diff --git a/daemons/execd/cts-exec-helper.c b/daemons/execd/cts-exec-helper.c index a74f5993a6..47ab5bbe38 100644 --- a/daemons/execd/cts-exec-helper.c +++ b/daemons/execd/cts-exec-helper.c @@ -1,626 +1,626 @@ /* * Copyright 2012-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include #include #include #define SUMMARY "cts-exec-helper - inject commands into the Pacemaker executor and watch for events" static int exec_call_id = 0; static gboolean start_test(gpointer user_data); static void try_connect(void); static char *key = NULL; static char *val = NULL; static struct { int verbose; int quiet; guint interval_ms; int timeout; int start_delay; int cancel_call_id; gboolean no_wait; gboolean is_running; gboolean no_connect; int exec_call_opts; const char *api_call; const char *rsc_id; const char *provider; const char *class; const char *type; const char *action; const char *listen; gboolean use_tls; lrmd_key_value_t *params; } options; static gboolean interval_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { return pcmk_parse_interval_spec(optarg, &options.interval_ms) == pcmk_rc_ok; } static gboolean notify_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { if (pcmk__str_any_of(option_name, "--notify-orig", "-n", NULL)) { options.exec_call_opts = lrmd_opt_notify_orig_only; } else if (pcmk__str_any_of(option_name, "--notify-changes", "-o", NULL)) { options.exec_call_opts = lrmd_opt_notify_changes_only; } return TRUE; } static gboolean param_key_val_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { if (pcmk__str_any_of(option_name, "--param-key", "-k", NULL)) { pcmk__str_update(&key, optarg); } else if (pcmk__str_any_of(option_name, "--param-val", "-v", NULL)) { pcmk__str_update(&val, optarg); } if (key != NULL && val != NULL) { options.params = lrmd_key_value_add(options.params, key, val); pcmk__str_update(&key, NULL); pcmk__str_update(&val, NULL); } return TRUE; } static GOptionEntry basic_entries[] = { { "api-call", 'c', 0, G_OPTION_ARG_STRING, &options.api_call, "Directly relates to executor API functions", NULL }, { "is-running", 'R', 0, G_OPTION_ARG_NONE, &options.is_running, "Determine if a resource is registered and running", NULL }, { "listen", 'l', 0, G_OPTION_ARG_STRING, &options.listen, "Listen for a specific event string", NULL }, { "no-wait", 'w', 0, G_OPTION_ARG_NONE, &options.no_wait, "Make api call and do not wait for result", NULL }, { "notify-changes", 'o', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, notify_cb, "Only notify client changes to recurring operations", NULL }, { "notify-orig", 'n', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, notify_cb, "Only notify this client of the results of an API action", NULL }, { "tls", 'S', 0, G_OPTION_ARG_NONE, &options.use_tls, "Use TLS backend for local connection", NULL }, { NULL } }; static GOptionEntry api_call_entries[] = { { "action", 'a', 0, G_OPTION_ARG_STRING, &options.action, NULL, NULL }, { "cancel-call-id", 'x', 0, G_OPTION_ARG_INT, &options.cancel_call_id, NULL, NULL }, { "class", 'C', 0, G_OPTION_ARG_STRING, &options.class, NULL, NULL }, { "interval", 'i', 0, G_OPTION_ARG_CALLBACK, interval_cb, NULL, NULL }, { "param-key", 'k', 0, G_OPTION_ARG_CALLBACK, param_key_val_cb, NULL, NULL }, { "param-val", 'v', 0, G_OPTION_ARG_CALLBACK, param_key_val_cb, NULL, NULL }, { "provider", 'P', 0, G_OPTION_ARG_STRING, &options.provider, NULL, NULL }, { "rsc-id", 'r', 0, G_OPTION_ARG_STRING, &options.rsc_id, NULL, NULL }, { "start-delay", 's', 0, G_OPTION_ARG_INT, &options.start_delay, NULL, NULL }, { "timeout", 't', 0, G_OPTION_ARG_INT, &options.timeout, NULL, NULL }, { "type", 'T', 0, G_OPTION_ARG_STRING, &options.type, NULL, NULL }, { NULL } }; static GMainLoop *mainloop = NULL; static lrmd_t *lrmd_conn = NULL; static char event_buf_v0[1024]; static crm_exit_t test_exit(crm_exit_t exit_code) { lrmd_api_delete(lrmd_conn); return crm_exit(exit_code); } #define print_result(fmt, args...) \ if (!options.quiet) { \ printf(fmt "\n" , ##args); \ } #define report_event(event) \ snprintf(event_buf_v0, sizeof(event_buf_v0), "NEW_EVENT event_type:%s rsc_id:%s action:%s rc:%s op_status:%s", \ lrmd_event_type2str(event->type), \ event->rsc_id, \ event->op_type ? event->op_type : "none", \ crm_exit_str((crm_exit_t) event->rc), \ pcmk_exec_status_str(event->op_status)); \ crm_info("%s", event_buf_v0); static void test_shutdown(int nsig) { lrmd_api_delete(lrmd_conn); lrmd_conn = NULL; } static void read_events(lrmd_event_data_t * event) { report_event(event); if (options.listen) { if (pcmk__str_eq(options.listen, event_buf_v0, pcmk__str_casei)) { print_result("LISTEN EVENT SUCCESSFUL"); test_exit(CRM_EX_OK); } } if (exec_call_id && (event->call_id == exec_call_id)) { if (event->op_status == 0 && event->rc == 0) { print_result("API-CALL SUCCESSFUL for 'exec'"); } else { print_result("API-CALL FAILURE for 'exec', rc:%d lrmd_op_status:%s", event->rc, pcmk_exec_status_str(event->op_status)); test_exit(CRM_EX_ERROR); } if (!options.listen) { test_exit(CRM_EX_OK); } } } static gboolean timeout_err(gpointer data) { print_result("LISTEN EVENT FAILURE - timeout occurred, never found"); test_exit(CRM_EX_TIMEOUT); return FALSE; } static void connection_events(lrmd_event_data_t * event) { int rc = event->connection_rc; if (event->type != lrmd_event_connect) { /* ignore */ return; } if (!rc) { crm_info("Executor client connection established"); start_test(NULL); return; } else { sleep(1); try_connect(); crm_notice("Executor client connection failed"); } } static void try_connect(void) { int tries = 10; static int num_tries = 0; int rc = 0; lrmd_conn->cmds->set_callback(lrmd_conn, connection_events); for (; num_tries < tries; num_tries++) { rc = lrmd_conn->cmds->connect_async(lrmd_conn, crm_system_name, 3000); if (!rc) { return; /* we'll hear back in async callback */ } sleep(1); } print_result("API CONNECTION FAILURE"); test_exit(CRM_EX_ERROR); } static gboolean start_test(gpointer user_data) { int rc = 0; if (!options.no_connect) { if (!lrmd_conn->cmds->is_connected(lrmd_conn)) { try_connect(); /* async connect -- this function will get called back into */ return 0; } } lrmd_conn->cmds->set_callback(lrmd_conn, read_events); if (options.timeout) { pcmk__create_timer(options.timeout, timeout_err, NULL); } if (!options.api_call) { return 0; } if (pcmk__str_eq(options.api_call, "exec", pcmk__str_casei)) { rc = lrmd_conn->cmds->exec(lrmd_conn, options.rsc_id, options.action, NULL, options.interval_ms, options.timeout, options.start_delay, options.exec_call_opts, options.params); if (rc > 0) { exec_call_id = rc; print_result("API-CALL 'exec' action pending, waiting on response"); } } else if (pcmk__str_eq(options.api_call, "register_rsc", pcmk__str_casei)) { rc = lrmd_conn->cmds->register_rsc(lrmd_conn, options.rsc_id, options.class, options.provider, options.type, 0); } else if (pcmk__str_eq(options.api_call, "get_rsc_info", pcmk__str_casei)) { lrmd_rsc_info_t *rsc_info; rsc_info = lrmd_conn->cmds->get_rsc_info(lrmd_conn, options.rsc_id, 0); if (rsc_info) { print_result("RSC_INFO: id:%s class:%s provider:%s type:%s", rsc_info->id, rsc_info->standard, (rsc_info->provider? rsc_info->provider : ""), rsc_info->type); lrmd_free_rsc_info(rsc_info); rc = pcmk_ok; } else { rc = -1; } } else if (pcmk__str_eq(options.api_call, "unregister_rsc", pcmk__str_casei)) { rc = lrmd_conn->cmds->unregister_rsc(lrmd_conn, options.rsc_id, 0); } else if (pcmk__str_eq(options.api_call, "cancel", pcmk__str_casei)) { rc = lrmd_conn->cmds->cancel(lrmd_conn, options.rsc_id, options.action, options.interval_ms); } else if (pcmk__str_eq(options.api_call, "metadata", pcmk__str_casei)) { char *output = NULL; rc = lrmd_conn->cmds->get_metadata(lrmd_conn, options.class, options.provider, options.type, &output, 0); if (rc == pcmk_ok) { print_result("%s", output); free(output); } } else if (pcmk__str_eq(options.api_call, "list_agents", pcmk__str_casei)) { lrmd_list_t *list = NULL; lrmd_list_t *iter = NULL; rc = lrmd_conn->cmds->list_agents(lrmd_conn, &list, options.class, options.provider); if (rc > 0) { print_result("%d agents found", rc); for (iter = list; iter != NULL; iter = iter->next) { print_result("%s", iter->val); } lrmd_list_freeall(list); rc = 0; } else { print_result("API_CALL FAILURE - no agents found"); rc = -1; } } else if (pcmk__str_eq(options.api_call, "list_ocf_providers", pcmk__str_casei)) { lrmd_list_t *list = NULL; lrmd_list_t *iter = NULL; rc = lrmd_conn->cmds->list_ocf_providers(lrmd_conn, options.type, &list); if (rc > 0) { print_result("%d providers found", rc); for (iter = list; iter != NULL; iter = iter->next) { print_result("%s", iter->val); } lrmd_list_freeall(list); rc = 0; } else { print_result("API_CALL FAILURE - no providers found"); rc = -1; } } else if (pcmk__str_eq(options.api_call, "list_standards", pcmk__str_casei)) { lrmd_list_t *list = NULL; lrmd_list_t *iter = NULL; rc = lrmd_conn->cmds->list_standards(lrmd_conn, &list); if (rc > 0) { print_result("%d standards found", rc); for (iter = list; iter != NULL; iter = iter->next) { print_result("%s", iter->val); } lrmd_list_freeall(list); rc = 0; } else { print_result("API_CALL FAILURE - no providers found"); rc = -1; } } else if (pcmk__str_eq(options.api_call, "get_recurring_ops", pcmk__str_casei)) { GList *op_list = NULL; GList *op_item = NULL; rc = lrmd_conn->cmds->get_recurring_ops(lrmd_conn, options.rsc_id, 0, 0, &op_list); for (op_item = op_list; op_item != NULL; op_item = op_item->next) { lrmd_op_info_t *op_info = op_item->data; print_result("RECURRING_OP: %s_%s_%s timeout=%sms", op_info->rsc_id, op_info->action, op_info->interval_ms_s, op_info->timeout_ms_s); lrmd_free_op_info(op_info); } g_list_free(op_list); } else if (options.api_call) { print_result("API-CALL FAILURE unknown action '%s'", options.action); test_exit(CRM_EX_ERROR); } if (rc < 0) { print_result("API-CALL FAILURE for '%s' api_rc:%d", options.api_call, rc); test_exit(CRM_EX_ERROR); } if (options.api_call && rc == pcmk_ok) { print_result("API-CALL SUCCESSFUL for '%s'", options.api_call); if (!options.listen) { test_exit(CRM_EX_OK); } } if (options.no_wait) { /* just make the call and exit regardless of anything else. */ test_exit(CRM_EX_OK); } return 0; } /*! * \internal * \brief Generate resource parameters from CIB if none explicitly given * * \return Standard Pacemaker return code */ static int generate_params(void) { int rc = pcmk_rc_ok; pcmk_scheduler_t *scheduler = NULL; xmlNode *cib_xml_copy = NULL; pcmk_resource_t *rsc = NULL; GHashTable *params = NULL; GHashTable *meta = NULL; GHashTableIter iter; char *key = NULL; char *value = NULL; if (options.params != NULL) { return pcmk_rc_ok; // User specified parameters explicitly } // Retrieve and update CIB rc = cib__signon_query(NULL, NULL, &cib_xml_copy); if (rc != pcmk_rc_ok) { return rc; } rc = pcmk__update_configured_schema(&cib_xml_copy, false); if (rc != pcmk_rc_ok) { return rc; } // Calculate cluster status - scheduler = pe_new_working_set(); + scheduler = pcmk_new_scheduler(); if (scheduler == NULL) { crm_crit("Could not allocate scheduler data"); return ENOMEM; } pcmk__set_scheduler_flags(scheduler, pcmk__sched_no_counts); scheduler->input = cib_xml_copy; scheduler->priv->now = crm_time_new(NULL); cluster_status(scheduler); // Find resource in CIB rsc = pe_find_resource_with_flags(scheduler->priv->resources, options.rsc_id, pcmk_rsc_match_history |pcmk_rsc_match_basename); if (rsc == NULL) { crm_err("Resource does not exist in config"); - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); return EINVAL; } // Add resource instance parameters to options.params params = pe_rsc_params(rsc, NULL, scheduler); if (params != NULL) { g_hash_table_iter_init(&iter, params); while (g_hash_table_iter_next(&iter, (gpointer *) &key, (gpointer *) &value)) { options.params = lrmd_key_value_add(options.params, key, value); } } // Add resource meta-attributes to options.params meta = pcmk__strkey_table(free, free); get_meta_attributes(meta, rsc, NULL, scheduler); g_hash_table_iter_init(&iter, meta); while (g_hash_table_iter_next(&iter, (gpointer *) &key, (gpointer *) &value)) { char *crm_name = crm_meta_name(key); options.params = lrmd_key_value_add(options.params, crm_name, value); free(crm_name); } g_hash_table_destroy(meta); - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); return rc; } static GOptionContext * build_arg_context(pcmk__common_args_t *args, GOptionGroup **group) { GOptionContext *context = NULL; context = pcmk__build_arg_context(args, NULL, group, NULL); pcmk__add_main_args(context, basic_entries); pcmk__add_arg_group(context, "api-call", "API Call Options:", "Parameters for api-call option", api_call_entries); return context; } int main(int argc, char **argv) { GError *error = NULL; crm_exit_t exit_code = CRM_EX_OK; crm_trigger_t *trig = NULL; pcmk__common_args_t *args = pcmk__new_common_args(SUMMARY); /* Typically we'd pass all the single character options that take an argument * as the second parameter here (and there's a bunch of those in this tool). * However, we control how this program is called so we can just not call it * in a way where the preprocessing ever matters. */ gchar **processed_args = pcmk__cmdline_preproc(argv, NULL); GOptionContext *context = build_arg_context(args, NULL); if (!g_option_context_parse_strv(context, &processed_args, &error)) { exit_code = CRM_EX_USAGE; goto done; } /* We have to use crm_log_init here to set up the logging because there's * different handling for daemons vs. command line programs, and * pcmk__cli_init_logging is set up to only handle the latter. */ crm_log_init(NULL, LOG_INFO, TRUE, (args->verbosity? TRUE : FALSE), argc, argv, FALSE); for (int i = 0; i < args->verbosity; i++) { crm_bump_log_level(argc, argv); } if (!options.listen && pcmk__strcase_any_of(options.api_call, "metadata", "list_agents", "list_standards", "list_ocf_providers", NULL)) { options.no_connect = TRUE; } if (options.is_running) { int rc = pcmk_rc_ok; if (options.rsc_id == NULL) { exit_code = CRM_EX_USAGE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "--is-running requires --rsc-id"); goto done; } options.interval_ms = 0; if (options.timeout == 0) { options.timeout = 30000; } rc = generate_params(); if (rc != pcmk_rc_ok) { exit_code = pcmk_rc2exitc(rc); g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Can not determine resource status: " "unable to get parameters from CIB"); goto done; } options.api_call = "exec"; options.action = PCMK_ACTION_MONITOR; options.exec_call_opts = lrmd_opt_notify_orig_only; } if (!options.api_call && !options.listen) { exit_code = CRM_EX_USAGE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Must specify at least one of --api-call, --listen, " "or --is-running"); goto done; } if (options.use_tls) { lrmd_conn = lrmd_remote_api_new(NULL, "localhost", 0); } else { lrmd_conn = lrmd_api_new(); } trig = mainloop_add_trigger(G_PRIORITY_HIGH, start_test, NULL); mainloop_set_trigger(trig); mainloop_add_signal(SIGTERM, test_shutdown); crm_info("Starting"); mainloop = g_main_loop_new(NULL, FALSE); g_main_loop_run(mainloop); done: g_strfreev(processed_args); pcmk__free_arg_context(context); free(key); free(val); pcmk__output_and_clear_error(&error, NULL); return test_exit(exit_code); } diff --git a/daemons/fenced/fenced_scheduler.c b/daemons/fenced/fenced_scheduler.c index 358b6264e4..8817cd4747 100644 --- a/daemons/fenced/fenced_scheduler.c +++ b/daemons/fenced/fenced_scheduler.c @@ -1,254 +1,254 @@ /* * Copyright 2009-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include static pcmk_scheduler_t *scheduler = NULL; /*! * \internal * \brief Initialize scheduler data for fencer purposes * * \return Standard Pacemaker return code */ int fenced_scheduler_init(void) { pcmk__output_t *logger = NULL; int rc = pcmk__log_output_new(&logger); if (rc != pcmk_rc_ok) { return rc; } - scheduler = pe_new_working_set(); + scheduler = pcmk_new_scheduler(); if (scheduler == NULL) { pcmk__output_free(logger); return ENOMEM; } pe__register_messages(logger); pcmk__register_lib_messages(logger); pcmk__output_set_log_level(logger, LOG_TRACE); scheduler->priv->out = logger; return pcmk_rc_ok; } /*! * \internal * \brief Set the local node name for scheduling purposes * * \param[in] node_name Name to set as local node name */ void fenced_set_local_node(const char *node_name) { pcmk__assert(scheduler != NULL); scheduler->priv->local_node_name = pcmk__str_copy(node_name); } /*! * \internal * \brief Get the local node name * * \return Local node name */ const char * fenced_get_local_node(void) { if (scheduler == NULL) { return NULL; } return scheduler->priv->local_node_name; } /*! * \internal * \brief Free all scheduler-related resources */ void fenced_scheduler_cleanup(void) { if (scheduler != NULL) { pcmk__output_t *logger = scheduler->priv->out; if (logger != NULL) { logger->finish(logger, CRM_EX_OK, true, NULL); pcmk__output_free(logger); scheduler->priv->out = NULL; } - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); scheduler = NULL; } } /*! * \internal * \brief Check whether the local node is in a resource's allowed node list * * \param[in] rsc Resource to check * * \return Pointer to node if found, otherwise NULL */ static pcmk_node_t * local_node_allowed_for(const pcmk_resource_t *rsc) { if ((rsc != NULL) && (scheduler->priv->local_node_name != NULL)) { GHashTableIter iter; pcmk_node_t *node = NULL; g_hash_table_iter_init(&iter, rsc->priv->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (void **) &node)) { if (pcmk__str_eq(node->priv->name, scheduler->priv->local_node_name, pcmk__str_casei)) { return node; } } } return NULL; } /*! * \internal * \brief If a given resource or any of its children are fencing devices, * register the devices * * \param[in,out] data Resource to check * \param[in,out] user_data Ignored */ static void register_if_fencing_device(gpointer data, gpointer user_data) { pcmk_resource_t *rsc = data; const char *rsc_id = pcmk__s(rsc->priv->history_id, rsc->id); xmlNode *xml = NULL; GHashTableIter hash_iter; pcmk_node_t *node = NULL; const char *name = NULL; const char *value = NULL; const char *agent = NULL; const char *rsc_provides = NULL; stonith_key_value_t *params = NULL; // If this is a collective resource, check children instead if (rsc->priv->children != NULL) { for (GList *iter = rsc->priv->children; iter != NULL; iter = iter->next) { register_if_fencing_device(iter->data, NULL); if (pcmk__is_clone(rsc)) { return; // Only one instance needs to be checked for clones } } return; } if (!pcmk_is_set(rsc->flags, pcmk__rsc_fence_device)) { return; // Not a fencing device } if (pe__resource_is_disabled(rsc)) { crm_info("Ignoring fencing device %s because it is disabled", rsc->id); return; } if ((stonith_watchdog_timeout_ms <= 0) && pcmk__str_eq(rsc->id, STONITH_WATCHDOG_ID, pcmk__str_none)) { crm_info("Ignoring fencing device %s " "because watchdog fencing is disabled", rsc->id); return; } // Check whether local node is allowed to run resource node = local_node_allowed_for(rsc); if (node == NULL) { crm_info("Ignoring fencing device %s " "because local node is not allowed to run it", rsc->id); return; } if (node->assign->score < 0) { crm_info("Ignoring fencing device %s " "because local node has preference %s for it", rsc->id, pcmk_readable_score(node->assign->score)); return; } // If device is in a group, check whether local node is allowed for group if (pcmk__is_group(rsc->priv->parent)) { pcmk_node_t *group_node = local_node_allowed_for(rsc->priv->parent); if ((group_node != NULL) && (group_node->assign->score < 0)) { crm_info("Ignoring fencing device %s " "because local node has preference %s for its group", rsc->id, pcmk_readable_score(group_node->assign->score)); return; } } crm_debug("Reloading configuration of fencing device %s", rsc->id); agent = crm_element_value(rsc->priv->xml, PCMK_XA_TYPE); get_meta_attributes(rsc->priv->meta, rsc, NULL, scheduler); rsc_provides = g_hash_table_lookup(rsc->priv->meta, PCMK_STONITH_PROVIDES); g_hash_table_iter_init(&hash_iter, pe_rsc_params(rsc, node, scheduler)); while (g_hash_table_iter_next(&hash_iter, (gpointer *) &name, (gpointer *) &value)) { if ((name == NULL) || (value == NULL)) { continue; } params = stonith_key_value_add(params, name, value); } xml = create_device_registration_xml(rsc_id, st_namespace_any, agent, params, rsc_provides); stonith_key_value_freeall(params, 1, 1); pcmk__assert(stonith_device_register(xml, TRUE) == pcmk_ok); pcmk__xml_free(xml); } /*! * \internal * \brief Run the scheduler for fencer purposes * * \param[in] cib Cluster's current CIB */ void fenced_scheduler_run(xmlNode *cib) { CRM_CHECK((cib != NULL) && (scheduler != NULL), return); if (scheduler->priv->now != NULL) { crm_time_free(scheduler->priv->now); scheduler->priv->now = NULL; } pcmk__schedule_actions(cib, pcmk__sched_location_only |pcmk__sched_no_counts, scheduler); g_list_foreach(scheduler->priv->resources, register_if_fencing_device, NULL); scheduler->input = NULL; // Wasn't a copy, so don't let API free it - pe_reset_working_set(scheduler); + pcmk_reset_scheduler(scheduler); } diff --git a/daemons/schedulerd/schedulerd_messages.c b/daemons/schedulerd/schedulerd_messages.c index 9e93fcd24c..a51a47c9e9 100644 --- a/daemons/schedulerd/schedulerd_messages.c +++ b/daemons/schedulerd/schedulerd_messages.c @@ -1,330 +1,330 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include "pacemaker-schedulerd.h" static GHashTable *schedulerd_handlers = NULL; static pcmk_scheduler_t * -init_working_set(void) +init_scheduler(void) { - pcmk_scheduler_t *scheduler = pe_new_working_set(); + pcmk_scheduler_t *scheduler = pcmk_new_scheduler(); pcmk__mem_assert(scheduler); scheduler->priv->out = logger_out; return scheduler; } static xmlNode * handle_pecalc_request(pcmk__request_t *request) { static struct series_s { const char *name; const char *param; /* Maximum number of inputs of this kind to save to disk. * If -1, save all; if 0, save none. */ int wrap; } series[] = { { "pe-error", PCMK_OPT_PE_ERROR_SERIES_MAX, -1 }, { "pe-warn", PCMK_OPT_PE_WARN_SERIES_MAX, 5000 }, { "pe-input", PCMK_OPT_PE_INPUT_SERIES_MAX, 4000 }, }; xmlNode *msg = request->xml; xmlNode *wrapper = pcmk__xe_first_child(msg, PCMK__XE_CRM_XML, NULL, NULL); xmlNode *xml_data = pcmk__xe_first_child(wrapper, NULL, NULL, NULL); static char *last_digest = NULL; static char *filename = NULL; unsigned int seq = 0U; int series_id = 0; int series_wrap = 0; char *digest = NULL; const char *value = NULL; time_t execution_date = time(NULL); xmlNode *converted = NULL; xmlNode *reply = NULL; bool is_repoke = false; bool process = true; - pcmk_scheduler_t *scheduler = init_working_set(); + pcmk_scheduler_t *scheduler = init_scheduler(); pcmk__ipc_send_ack(request->ipc_client, request->ipc_id, request->ipc_flags, PCMK__XE_ACK, NULL, CRM_EX_INDETERMINATE); digest = pcmk__digest_xml(xml_data, false); converted = pcmk__xml_copy(NULL, xml_data); if (pcmk__update_configured_schema(&converted, true) != pcmk_rc_ok) { scheduler->priv->graph = pcmk__xe_create(NULL, PCMK__XE_TRANSITION_GRAPH); crm_xml_add_int(scheduler->priv->graph, "transition_id", 0); crm_xml_add_int(scheduler->priv->graph, PCMK_OPT_CLUSTER_DELAY, 0); process = false; free(digest); } else if (pcmk__str_eq(digest, last_digest, pcmk__str_casei)) { is_repoke = true; free(digest); } else { free(last_digest); last_digest = digest; } if (process) { pcmk__schedule_actions(converted, pcmk__sched_no_counts |pcmk__sched_show_utilization, scheduler); } // Get appropriate index into series[] array if (pcmk_is_set(scheduler->flags, pcmk__sched_processing_error) || pcmk__config_has_error) { series_id = 0; } else if (pcmk_is_set(scheduler->flags, pcmk__sched_processing_warning) || pcmk__config_has_warning) { series_id = 1; } else { series_id = 2; } value = pcmk__cluster_option(scheduler->priv->options, series[series_id].param); if ((value == NULL) || (pcmk__scan_min_int(value, &series_wrap, -1) != pcmk_rc_ok)) { series_wrap = series[series_id].wrap; } if (pcmk__read_series_sequence(PCMK_SCHEDULER_INPUT_DIR, series[series_id].name, &seq) != pcmk_rc_ok) { // @TODO maybe handle errors better ... seq = 0U; } crm_trace("Series %s: wrap=%d, seq=%u, pref=%s", series[series_id].name, series_wrap, seq, value); scheduler->input = NULL; reply = pcmk__new_reply(msg, scheduler->priv->graph); if (reply == NULL) { pcmk__format_result(&request->result, CRM_EX_ERROR, PCMK_EXEC_ERROR, "Failed building ping reply for client %s", pcmk__client_name(request->ipc_client)); goto done; } if (series_wrap == 0) { // Don't save any inputs of this kind free(filename); filename = NULL; } else if (!is_repoke) { // Input changed, save to disk free(filename); filename = pcmk__series_filename(PCMK_SCHEDULER_INPUT_DIR, series[series_id].name, seq, true); } crm_xml_add(reply, PCMK__XA_CRM_TGRAPH_IN, filename); pcmk__log_transition_summary(scheduler, filename); if (series_wrap == 0) { crm_debug("Not saving input to disk (disabled by configuration)"); } else if (is_repoke) { crm_info("Input has not changed since last time, not saving to disk"); } else { unlink(filename); crm_xml_add_ll(xml_data, PCMK_XA_EXECUTION_DATE, (long long) execution_date); pcmk__xml_write_file(xml_data, filename, true); pcmk__write_series_sequence(PCMK_SCHEDULER_INPUT_DIR, series[series_id].name, ++seq, series_wrap); } pcmk__set_result(&request->result, CRM_EX_OK, PCMK_EXEC_DONE, NULL); done: pcmk__xml_free(converted); - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); return reply; } static xmlNode * handle_unknown_request(pcmk__request_t *request) { pcmk__ipc_send_ack(request->ipc_client, request->ipc_id, request->ipc_flags, PCMK__XE_ACK, NULL, CRM_EX_INVALID_PARAM); pcmk__format_result(&request->result, CRM_EX_PROTOCOL, PCMK_EXEC_INVALID, "Unknown IPC request type '%s' (bug?)", pcmk__client_name(request->ipc_client)); return NULL; } static xmlNode * handle_hello_request(pcmk__request_t *request) { pcmk__ipc_send_ack(request->ipc_client, request->ipc_id, request->ipc_flags, PCMK__XE_ACK, NULL, CRM_EX_INDETERMINATE); crm_trace("Received IPC hello from %s", pcmk__client_name(request->ipc_client)); pcmk__set_result(&request->result, CRM_EX_OK, PCMK_EXEC_DONE, NULL); return NULL; } static void schedulerd_register_handlers(void) { pcmk__server_command_t handlers[] = { { CRM_OP_HELLO, handle_hello_request }, { CRM_OP_PECALC, handle_pecalc_request }, { NULL, handle_unknown_request }, }; schedulerd_handlers = pcmk__register_handlers(handlers); } static int32_t pe_ipc_accept(qb_ipcs_connection_t * c, uid_t uid, gid_t gid) { crm_trace("Connection %p", c); if (pcmk__new_client(c, uid, gid) == NULL) { return -ENOMEM; } return 0; } static int32_t pe_ipc_dispatch(qb_ipcs_connection_t * qbc, void *data, size_t size) { uint32_t id = 0; uint32_t flags = 0; xmlNode *msg = NULL; pcmk__client_t *c = pcmk__find_client(qbc); const char *sys_to = NULL; CRM_CHECK(c != NULL, return 0); if (schedulerd_handlers == NULL) { schedulerd_register_handlers(); } msg = pcmk__client_data2xml(c, data, &id, &flags); if (msg == NULL) { pcmk__ipc_send_ack(c, id, flags, PCMK__XE_ACK, NULL, CRM_EX_PROTOCOL); return 0; } sys_to = crm_element_value(msg, PCMK__XA_CRM_SYS_TO); if (pcmk__str_eq(crm_element_value(msg, PCMK__XA_SUBT), PCMK__VALUE_RESPONSE, pcmk__str_none)) { pcmk__ipc_send_ack(c, id, flags, PCMK__XE_ACK, NULL, CRM_EX_INDETERMINATE); crm_info("Ignoring IPC reply from %s", pcmk__client_name(c)); } else if (!pcmk__str_eq(sys_to, CRM_SYSTEM_PENGINE, pcmk__str_none)) { pcmk__ipc_send_ack(c, id, flags, PCMK__XE_ACK, NULL, CRM_EX_INDETERMINATE); crm_info("Ignoring invalid IPC message: to '%s' not " CRM_SYSTEM_PENGINE, pcmk__s(sys_to, "")); } else { char *log_msg = NULL; const char *reason = NULL; xmlNode *reply = NULL; pcmk__request_t request = { .ipc_client = c, .ipc_id = id, .ipc_flags = flags, .peer = NULL, .xml = msg, .call_options = 0, .result = PCMK__UNKNOWN_RESULT, }; request.op = crm_element_value_copy(request.xml, PCMK__XA_CRM_TASK); CRM_CHECK(request.op != NULL, return 0); reply = pcmk__process_request(&request, schedulerd_handlers); if (reply != NULL) { pcmk__ipc_send_xml(c, id, reply, crm_ipc_server_event); pcmk__xml_free(reply); } reason = request.result.exit_reason; log_msg = crm_strdup_printf("Processed %s request from %s %s: %s%s%s%s", request.op, pcmk__request_origin_type(&request), pcmk__request_origin(&request), pcmk_exec_status_str(request.result.execution_status), (reason == NULL)? "" : " (", (reason == NULL)? "" : reason, (reason == NULL)? "" : ")"); if (!pcmk__result_ok(&request.result)) { crm_warn("%s", log_msg); } else { crm_debug("%s", log_msg); } free(log_msg); pcmk__reset_request(&request); } pcmk__xml_free(msg); return 0; } /* Error code means? */ static int32_t pe_ipc_closed(qb_ipcs_connection_t * c) { pcmk__client_t *client = pcmk__find_client(c); if (client == NULL) { return 0; } crm_trace("Connection %p", c); pcmk__free_client(client); return 0; } static void pe_ipc_destroy(qb_ipcs_connection_t * c) { crm_trace("Connection %p", c); pe_ipc_closed(c); } struct qb_ipcs_service_handlers ipc_callbacks = { .connection_accept = pe_ipc_accept, .connection_created = NULL, .msg_process = pe_ipc_dispatch, .connection_closed = pe_ipc_closed, .connection_destroyed = pe_ipc_destroy }; diff --git a/include/crm/common/action_relation_internal.h b/include/crm/common/action_relation_internal.h index 2f594b3edf..895bf29eca 100644 --- a/include/crm/common/action_relation_internal.h +++ b/include/crm/common/action_relation_internal.h @@ -1,188 +1,191 @@ /* * Copyright 2023-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #ifndef PCMK__CRM_COMMON_ACTION_RELATION_INTERNAL__H #define PCMK__CRM_COMMON_ACTION_RELATION_INTERNAL__H #include // bool #include // uint32_t +#include // gpointer #include // pcmk_resource_t, pcmk_action_t #ifdef __cplusplus extern "C" { #endif // Flags to indicate the relationship between two actions enum pcmk__action_relation_flags { //! No relation (compare with equality rather than bit set) pcmk__ar_none = 0U, //! Actions are ordered (optionally, if no other flags are set) pcmk__ar_ordered = (1U << 0), //! Relation applies only if 'first' cannot be part of a live migration pcmk__ar_if_first_unmigratable = (1U << 1), /*! * If 'then' is required, 'first' becomes required (and becomes unmigratable * if 'then' is); also, if 'first' is a stop of a blocked resource, 'then' * becomes unrunnable */ pcmk__ar_then_implies_first = (1U << 4), /*! * If 'first' is required, 'then' becomes required; if 'first' is a stop of * a blocked resource, 'then' becomes unrunnable */ pcmk__ar_first_implies_then = (1U << 5), /*! * If 'then' is required and for a promoted instance, 'first' becomes * required (and becomes unmigratable if 'then' is) */ pcmk__ar_promoted_then_implies_first = (1U << 6), /*! * 'first' is runnable only if 'then' is both runnable and migratable, * and 'first' becomes required if 'then' is */ pcmk__ar_unmigratable_then_blocks = (1U << 7), //! 'then' is runnable (and migratable) only if 'first' is runnable pcmk__ar_unrunnable_first_blocks = (1U << 8), //! If 'first' is unrunnable, 'then' becomes a real, unmigratable action pcmk__ar_first_else_then = (1U << 9), //! If 'first' is required, 'then' action for instance on same node is pcmk__ar_first_implies_same_node_then = (1U << 10), /*! * Disable relation if 'first' is unrunnable and for an active resource, * otherwise order actions and make 'then' unrunnable if 'first' is. * * This is used to order a bundle replica's start of its container before a * probe of its remote connection resource, in case the connection uses the * REMOTE_CONTAINER_HACK to replace the connection address with where the * container is running. */ pcmk__ar_nested_remote_probe = (1U << 11), /*! * If 'first' is for a blocked resource, make 'then' unrunnable. * * If 'then' is required, make 'first' required, make 'first' unmigratable * if 'then' is unmigratable, and make 'then' unrunnable if 'first' is * unrunnable. * * If 'then' is unrunnable and for the same resource as 'first', make * 'first' required if it is runnable, and make 'first' unmigratable if * 'then' is unmigratable. * * This is used for "stop then start primitive" (restarts) and * "stop group member then stop previous member". */ pcmk__ar_intermediate_stop = (1U << 12), /*! * The actions must be serialized if in the same transition but can be in * either order. (In practice, we always arrange them as 'first' then * 'then', so they end up being essentially the same as optional orderings.) * * @TODO Handle more intelligently -- for example, we could schedule the * action with the fewest inputs first, so we're more likely to execute at * least one if there is a failure during the transition. Or, we could * prefer certain action types over others, or base it on resource priority. */ pcmk__ar_serialize = (1U << 14), //! Relation applies only if actions are on same node pcmk__ar_if_on_same_node = (1U << 15), //! If 'then' is required, 'first' must be added to the transition graph pcmk__ar_then_implies_first_graphed = (1U << 16), //! If 'first' is required and runnable, 'then' must be in graph pcmk__ar_first_implies_then_graphed = (1U << 17), //! User-configured asymmetric ordering pcmk__ar_asymmetric = (1U << 20), //! Actions are ordered if on same node (or migration target for migrate_to) pcmk__ar_if_on_same_node_or_target = (1U << 21), //! 'then' action is runnable if certain number of 'first' instances are pcmk__ar_min_runnable = (1U << 22), //! Ordering applies only if 'first' is required and on same node as 'then' pcmk__ar_if_required_on_same_node = (1U << 23), //! Ordering applies even if 'first' runs on guest node created by 'then' pcmk__ar_guest_allowed = (1U << 24), //! If 'then' action becomes required, 'first' becomes optional pcmk__ar_then_cancels_first = (1U << 25), }; /* Action relation object * * The most common type of relation is an ordering, in which case action1 etc. * refers to the "first" action, and action2 etc. refers to the "then" action. */ typedef struct { int id; // Counter to identify relation uint32_t flags; // Group of enum pcmk__action_relation_flags pcmk_resource_t *rsc1; // Resource for first action, if any pcmk_action_t *action1; // First action in relation char *task1; // Action name or key for first action pcmk_resource_t *rsc2; // Resource for second action, if any pcmk_action_t *action2; // Second action in relation char *task2; // Action name or key for second action } pcmk__action_relation_t; // Action sequenced relative to another action typedef struct pcmk__related_action { pcmk_action_t *action; // Action to be sequenced uint32_t flags; // Group of enum pcmk__action_relation_flags bool graphed; // Whether action has been added to graph yet } pcmk__related_action_t; /*! * \internal * \brief Set action relation flags * * \param[in,out] ar_flags Flag group to modify * \param[in] flags_to_set enum pcmk__action_relation_flags to set */ #define pcmk__set_relation_flags(ar_flags, flags_to_set) do { \ ar_flags = pcmk__set_flags_as(__func__, __LINE__, LOG_TRACE, \ "Action relation", "constraint", \ ar_flags, (flags_to_set), \ #flags_to_set); \ } while (0) /*! * \internal * \brief Clear action relation flags * * \param[in,out] ar_flags Flag group to modify * \param[in] flags_to_clear enum pcmk__action_relation_flags to clear */ #define pcmk__clear_relation_flags(ar_flags, flags_to_clear) do { \ ar_flags = pcmk__clear_flags_as(__func__, __LINE__, LOG_TRACE, \ "Action relation", "constraint", \ ar_flags, (flags_to_clear), \ #flags_to_clear); \ } while (0) +void pcmk__free_action_relation(gpointer user_data); + #ifdef __cplusplus } #endif #endif // PCMK__CRM_COMMON_ACTION_RELATION_INTERNAL__H diff --git a/include/crm/common/actions_internal.h b/include/crm/common/actions_internal.h index 3d025cd5b4..072aa1adaf 100644 --- a/include/crm/common/actions_internal.h +++ b/include/crm/common/actions_internal.h @@ -1,269 +1,270 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #ifndef PCMK__CRM_COMMON_ACTIONS_INTERNAL__H #define PCMK__CRM_COMMON_ACTIONS_INTERNAL__H #include // bool #include // uint32_t, UINT32_C() #include // guint, GList, GHashTable #include // xmlNode #include // PCMK_ACTION_MONITOR #include // enum rsc_role_e #include // pcmk_resource_t, pcmk_node_t #include // pcmk__str_eq() #ifdef __cplusplus extern "C" { #endif //! printf-style format to create operation key from resource, action, interval #define PCMK__OP_FMT "%s_%s_%u" /*! * \internal * \brief Set action flags for an action * * \param[in,out] action Action to set flags for * \param[in] flags_to_set Group of enum pcmk__action_flags to set */ #define pcmk__set_action_flags(action, flags_to_set) do { \ (action)->flags = pcmk__set_flags_as(__func__, __LINE__, \ LOG_TRACE, \ "Action", (action)->uuid, \ (action)->flags, \ (flags_to_set), \ #flags_to_set); \ } while (0) /*! * \internal * \brief Clear action flags for an action * * \param[in,out] action Action to clear flags for * \param[in] flags_to_clear Group of enum pcmk__action_flags to clear */ #define pcmk__clear_action_flags(action, flags_to_clear) do { \ (action)->flags = pcmk__clear_flags_as(__func__, __LINE__, \ LOG_TRACE, \ "Action", (action)->uuid, \ (action)->flags, \ (flags_to_clear), \ #flags_to_clear); \ } while (0) /*! * \internal * \brief Set action flags for a flag group * * \param[in,out] action_flags Flag group to set flags for * \param[in] action_name Name of action being modified (for logging) * \param[in] to_set Group of enum pcmk__action_flags to set */ #define pcmk__set_raw_action_flags(action_flags, action_name, to_set) do { \ action_flags = pcmk__set_flags_as(__func__, __LINE__, \ LOG_TRACE, "Action", action_name, \ (action_flags), \ (to_set), #to_set); \ } while (0) /*! * \internal * \brief Clear action flags for a flag group * * \param[in,out] action_flags Flag group to clear flags for * \param[in] action_name Name of action being modified (for logging) * \param[in] to_clear Group of enum pcmk__action_flags to clear */ #define pcmk__clear_raw_action_flags(action_flags, action_name, to_clear) \ do { \ action_flags = pcmk__clear_flags_as(__func__, __LINE__, LOG_TRACE, \ "Action", action_name, \ (action_flags), \ (to_clear), #to_clear); \ } while (0) // Possible actions (including some pseudo-actions) enum pcmk__action_type { pcmk__action_unspecified = 0, // Unspecified or unknown action pcmk__action_monitor, // Monitor // Each "completed" action must be the regular action plus 1 pcmk__action_stop, // Stop pcmk__action_stopped, // Stop completed pcmk__action_start, // Start pcmk__action_started, // Start completed pcmk__action_notify, // Notify pcmk__action_notified, // Notify completed pcmk__action_promote, // Promote pcmk__action_promoted, // Promoted pcmk__action_demote, // Demote pcmk__action_demoted, // Demoted pcmk__action_shutdown, // Shut down node pcmk__action_fence, // Fence node }; // Action scheduling flags enum pcmk__action_flags { // No action flags set (compare with equality rather than bit set) pcmk__no_action_flags = 0, // Whether action does not require invoking an agent pcmk__action_pseudo = (UINT32_C(1) << 0), // Whether action is runnable pcmk__action_runnable = (UINT32_C(1) << 1), // Whether action should not be executed pcmk__action_optional = (UINT32_C(1) << 2), // Whether action should be added to transition graph even if optional pcmk__action_always_in_graph = (UINT32_C(1) << 3), // Whether operation-specific instance attributes have been unpacked yet pcmk__action_attrs_evaluated = (UINT32_C(1) << 4), // Whether action is allowed to be part of a live migration pcmk__action_migratable = (UINT32_C(1) << 7), // Whether action has been added to transition graph pcmk__action_added_to_graph = (UINT32_C(1) << 8), // Whether action is a stop to abort a dangling migration pcmk__action_migration_abort = (UINT32_C(1) << 11), // Whether action is recurring monitor that must be rescheduled if active pcmk__action_reschedule = (UINT32_C(1) << 13), // Whether action has already been processed by a recursive procedure pcmk__action_detect_loop = (UINT32_C(1) << 14), // Whether action's inputs have been de-duplicated yet pcmk__action_inputs_deduplicated = (UINT32_C(1) << 15), // Whether action can be executed on DC rather than own node pcmk__action_on_dc = (UINT32_C(1) << 16), }; /* Possible responses to a resource action failure * * The order is significant; the values are in order of increasing severity so * that they can be compared with less than and greater than. */ enum pcmk__on_fail { pcmk__on_fail_ignore, // Act as if failure didn't happen pcmk__on_fail_demote, // Demote if promotable, else stop pcmk__on_fail_restart, // Restart resource /* Fence the remote node created by the resource if fencing is enabled, * otherwise attempt to restart the resource (used internally for some * remote connection failures). */ pcmk__on_fail_reset_remote, pcmk__on_fail_restart_container, // Restart resource's container pcmk__on_fail_ban, // Ban resource from current node pcmk__on_fail_block, // Treat resource as unmanaged pcmk__on_fail_stop, // Stop resource and leave stopped pcmk__on_fail_standby_node, // Put resource's node in standby pcmk__on_fail_fence_node, // Fence resource's node }; // What resource needs before it can be recovered from a failed node enum pcmk__requires { pcmk__requires_nothing = 0, // Resource can be recovered immediately pcmk__requires_quorum = 1, // Resource can be recovered if quorate pcmk__requires_fencing = 2, // Resource can be recovered after fencing }; // Implementation of pcmk_action_t struct pcmk__action { int id; // Counter to identify action /* * When the controller aborts a transition graph, it sets an abort priority. * If this priority is higher, the action will still be executed anyway. * Pseudo-actions are always allowed, so this is irrelevant for them. */ int priority; pcmk_resource_t *rsc; // Resource to apply action to, if any pcmk_node_t *node; // Copy of node to execute action on, if any xmlNode *op_entry; // Action XML configuration, if any char *task; // Action name char *uuid; // Action key char *cancel_task; // If task is "cancel", the action being cancelled char *reason; // Readable description of why action is needed uint32_t flags; // Group of enum pcmk__action_flags enum pcmk__requires needs; // Prerequisite for recovery enum pcmk__on_fail on_fail; // Response to failure enum rsc_role_e fail_role; // Resource role if action fails GHashTable *meta; // Meta-attributes relevant to action GHashTable *extra; // Action-specific instance attributes pcmk_scheduler_t *scheduler; // Scheduler data this action is part of /* Current count of runnable instance actions for "first" action in an * ordering dependency with pcmk__ar_min_runnable set. */ int runnable_before; /* * Number of instance actions for "first" action in an ordering dependency * with pcmk__ar_min_runnable set that must be runnable before this action * can be runnable. */ int required_runnable_before; // Actions in a relation with this one (as pcmk__related_action_t *) GList *actions_before; GList *actions_after; }; +void pcmk__free_action(gpointer user_data); char *pcmk__op_key(const char *rsc_id, const char *op_type, guint interval_ms); char *pcmk__notify_key(const char *rsc_id, const char *notify_type, const char *op_type); char *pcmk__transition_key(int transition_id, int action_id, int target_rc, const char *node); void pcmk__filter_op_for_digest(xmlNode *param_set); bool pcmk__is_fencing_action(const char *action); enum pcmk__action_type pcmk__parse_action(const char *action_name); const char *pcmk__action_text(enum pcmk__action_type action); const char *pcmk__on_fail_text(enum pcmk__on_fail on_fail); /*! * \internal * \brief Get a human-friendly action name * * \param[in] action_name Actual action name * \param[in] interval_ms Action interval (in milliseconds) * * \return Action name suitable for display */ static inline const char * pcmk__readable_action(const char *action_name, guint interval_ms) { if ((interval_ms == 0) && pcmk__str_eq(action_name, PCMK_ACTION_MONITOR, pcmk__str_none)) { return "probe"; } return action_name; } #ifdef __cplusplus } #endif #endif // PCMK__CRM_COMMON_ACTIONS_INTERNAL__H diff --git a/include/crm/common/location_internal.h b/include/crm/common/location_internal.h index 39aba91c4b..1ca3024c3c 100644 --- a/include/crm/common/location_internal.h +++ b/include/crm/common/location_internal.h @@ -1,36 +1,38 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #ifndef PCMK__CRM_COMMON_LOCATION_INTERNAL__H #define PCMK__CRM_COMMON_LOCATION_INTERNAL__H -#include // GList +#include // gpointer, GList #include // enum pcmk__probe_mode #include // enum rsc_role_e #include // pcmk_resource_t #ifdef __cplusplus extern "C" { #endif //! Location constraint object typedef struct { char *id; // XML ID of location constraint pcmk_resource_t *rsc; // Resource with location preference enum rsc_role_e role_filter; // Limit to instances with this role enum pcmk__probe_mode probe_mode; // How to probe resource on node GList *nodes; // Copies of affected nodes, with score } pcmk__location_t; +void pcmk__free_location(gpointer user_data); + #ifdef __cplusplus } #endif #endif // PCMK__CRM_COMMON_LOCATION_INTERNAL__H diff --git a/include/crm/common/nodes_internal.h b/include/crm/common/nodes_internal.h index 4e0dd86691..b8525758a3 100644 --- a/include/crm/common/nodes_internal.h +++ b/include/crm/common/nodes_internal.h @@ -1,202 +1,204 @@ /* * Copyright 2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #ifndef PCMK__CRM_COMMON_NODES_INTERNAL__H #define PCMK__CRM_COMMON_NODES_INTERNAL__H #include // NULL #include // bool #include // uint32_t, UINT32_C() -#include +#include // gpointer, GList, GHashTable #include #ifdef __cplusplus extern "C" { #endif /* * Special node attributes */ #define PCMK__NODE_ATTR_SHUTDOWN "shutdown" /* @COMPAT Deprecated since 2.1.8. Use a location constraint with * PCMK_XA_RSC_PATTERN=".*" and PCMK_XA_RESOURCE_DISCOVERY="never" instead of * PCMK__NODE_ATTR_RESOURCE_DISCOVERY_ENABLED="false". */ #define PCMK__NODE_ATTR_RESOURCE_DISCOVERY_ENABLED "resource-discovery-enabled" enum pcmk__node_variant { // Possible node types pcmk__node_variant_cluster = 1, // Cluster layer node pcmk__node_variant_remote = 2, // Pacemaker Remote node }; enum pcmk__node_flags { pcmk__node_none = UINT32_C(0), // Whether node is in standby mode pcmk__node_standby = (UINT32_C(1) << 0), // Whether node is in standby mode due to PCMK_META_ON_FAIL pcmk__node_fail_standby = (UINT32_C(1) << 1), // Whether node has ever joined cluster (and thus has node state in CIB) pcmk__node_seen = (UINT32_C(1) << 2), // Whether expected join state is member pcmk__node_expected_up = (UINT32_C(1) << 3), // Whether probes are allowed on node pcmk__node_probes_allowed = (UINT32_C(1) << 4), /* Whether this either is a guest node whose guest resource must be * recovered or a remote node that must be fenced */ pcmk__node_remote_reset = (UINT32_C(1) << 5), /* Whether this is a Pacemaker Remote node that was fenced since it was last * connected by the cluster */ pcmk__node_remote_fenced = (UINT32_C(1) << 6), /* * Whether this is a Pacemaker Remote node previously marked in its * node state as being in maintenance mode */ pcmk__node_remote_maint = (UINT32_C(1) << 7), // Whether node history has been unpacked pcmk__node_unpacked = (UINT32_C(1) << 8), }; // When to probe a resource on a node (as specified in location constraints) enum pcmk__probe_mode { pcmk__probe_always = 0, // Always probe resource on node pcmk__probe_never = 1, // Never probe resource on node pcmk__probe_exclusive = 2, // Probe only on designated nodes }; /* Per-node data used in resource assignment * * @COMPAT When we can make the pcmk_node_t implementation internal, move these * there and drop this struct. */ struct pcmk__node_assignment { int score; // Node's score for relevant resource int count; // Counter reused by assignment and promotion code enum pcmk__probe_mode probe_mode; // When to probe resource on this node }; /* Implementation of pcmk__node_private_t (pcmk_node_t objects are shallow * copies, so all pcmk_node_t objects for the same node will share the same * private data) */ struct pcmk__node_private { /* Node's XML ID in the CIB (the cluster layer ID for cluster nodes, * the node name for Pacemaker Remote nodes) */ const char *id; /* * Sum of priorities of all resources active on node and on any guest nodes * connected to this node, with +1 for promoted instances (used to compare * nodes for PCMK_OPT_PRIORITY_FENCING_DELAY) */ int priority; const char *name; // Node name in cluster enum pcmk__node_variant variant; // Node variant uint32_t flags; // Group of enum pcmk__node_flags GHashTable *attrs; // Node attributes GHashTable *utilization; // Node utilization attributes int num_resources; // Number of active resources on node GList *assigned_resources; // List of resources assigned to node GHashTable *digest_cache; // Cache of calculated resource digests pcmk_resource_t *remote; // Pacemaker Remote connection (if any) pcmk_scheduler_t *scheduler; // Scheduler data that node is part of }; void pcmk__free_node_copy(void *data); pcmk_node_t *pcmk__find_node_in_list(const GList *nodes, const char *node_name); /*! * \internal * \brief Set node flags * * \param[in,out] node Node to set flags for * \param[in] flags_to_set Group of enum pcmk_node_flags to set */ #define pcmk__set_node_flags(node, flags_to_set) do { \ (node)->priv->flags = pcmk__set_flags_as(__func__, __LINE__, \ LOG_TRACE, "Node", pcmk__node_name(node), \ (node)->priv->flags, (flags_to_set), #flags_to_set); \ } while (0) /*! * \internal * \brief Clear node flags * * \param[in,out] node Node to clear flags for * \param[in] flags_to_clear Group of enum pcmk_node_flags to clear */ #define pcmk__clear_node_flags(node, flags_to_clear) do { \ (node)->priv->flags = pcmk__clear_flags_as(__func__, __LINE__, \ LOG_TRACE, "Node", pcmk__node_name(node), \ (node)->priv->flags, (flags_to_clear), #flags_to_clear); \ } while (0) +void pcmk__free_node(gpointer user_data); + /*! * \internal * \brief Return a string suitable for logging as a node name * * \param[in] node Node to return a node name string for * * \return Node name if available, otherwise node ID if available, * otherwise "unspecified node" if node is NULL or "unidentified node" * if node has neither a name nor ID. */ static inline const char * pcmk__node_name(const pcmk_node_t *node) { if (node == NULL) { return "unspecified node"; } else if (node->priv->name != NULL) { return node->priv->name; } else if (node->priv->id != NULL) { return node->priv->id; } else { return "unidentified node"; } } /*! * \internal * \brief Check whether two node objects refer to the same node * * \param[in] node1 First node object to compare * \param[in] node2 Second node object to compare * * \return true if \p node1 and \p node2 refer to the same node */ static inline bool pcmk__same_node(const pcmk_node_t *node1, const pcmk_node_t *node2) { return (node1 != NULL) && (node2 != NULL) && (node1->priv == node2->priv); } #ifdef __cplusplus } #endif #endif // PCMK__CRM_COMMON_NODES_INTERNAL__H diff --git a/include/crm/common/resources_internal.h b/include/crm/common/resources_internal.h index 87ac4c609c..2200ee153b 100644 --- a/include/crm/common/resources_internal.h +++ b/include/crm/common/resources_internal.h @@ -1,468 +1,469 @@ /* * Copyright 2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #ifndef PCMK__CRM_COMMON_RESOURCES_INTERNAL__H #define PCMK__CRM_COMMON_RESOURCES_INTERNAL__H #include // uint32_t -#include // gboolean, guint, GHashTable, GList +#include // gboolean, gpointer, guint, etc. #include // xmlNode #include // pcmk_resource_t #include // enum rsc_role_e #include // pcmk_node_t, etc. #ifdef __cplusplus extern "C" { #endif /*! * \internal * \brief Set resource flags * * \param[in,out] resource Resource to set flags for * \param[in] flags_to_set Group of enum pcmk_rsc_flags to set */ #define pcmk__set_rsc_flags(resource, flags_to_set) do { \ (resource)->flags = pcmk__set_flags_as(__func__, __LINE__, \ LOG_TRACE, "Resource", (resource)->id, (resource)->flags, \ (flags_to_set), #flags_to_set); \ } while (0) /*! * \internal * \brief Clear resource flags * * \param[in,out] resource Resource to clear flags for * \param[in] flags_to_clear Group of enum pcmk_rsc_flags to clear */ #define pcmk__clear_rsc_flags(resource, flags_to_clear) do { \ (resource)->flags = pcmk__clear_flags_as(__func__, __LINE__, \ LOG_TRACE, "Resource", (resource)->id, (resource)->flags, \ (flags_to_clear), #flags_to_clear); \ } while (0) //! Resource variants supported by Pacemaker enum pcmk__rsc_variant { // Order matters: some code compares greater or lesser than pcmk__rsc_variant_unknown = -1, //!< Unknown resource variant pcmk__rsc_variant_primitive = 0, //!< Primitive resource pcmk__rsc_variant_group = 1, //!< Group resource pcmk__rsc_variant_clone = 2, //!< Clone resource pcmk__rsc_variant_bundle = 3, //!< Bundle resource }; //! How to recover a resource that is incorrectly active on multiple nodes enum pcmk__multiply_active { pcmk__multiply_active_restart, //!< Stop on all, start on desired pcmk__multiply_active_stop, //!< Stop on all and leave stopped pcmk__multiply_active_block, //!< Do nothing to resource pcmk__multiply_active_unexpected, //!< Stop unexpected instances }; //! Resource scheduling flags enum pcmk__rsc_flags { // No resource flags set (compare with equality rather than bit set) pcmk__no_rsc_flags = 0ULL, // Whether resource has been removed from the configuration pcmk__rsc_removed = (1ULL << 0), /* NOTE: sbd (at least as of 1.5.2) uses pe_rsc_managed which equates to * this value, so the value should not be changed */ // Whether resource is managed pcmk__rsc_managed = (1ULL << 1), // Whether resource is blocked from further action pcmk__rsc_blocked = (1ULL << 2), // Whether resource has been removed but was launched pcmk__rsc_removed_launched = (1ULL << 3), // Whether resource has clone notifications enabled pcmk__rsc_notify = (1ULL << 4), // Whether resource is not an anonymous clone instance pcmk__rsc_unique = (1ULL << 5), // Whether resource's class is "stonith" pcmk__rsc_fence_device = (1ULL << 6), // Whether resource can be promoted and demoted pcmk__rsc_promotable = (1ULL << 7), // Whether resource has not yet been assigned to a node pcmk__rsc_unassigned = (1ULL << 8), // Whether resource is in the process of being assigned to a node pcmk__rsc_assigning = (1ULL << 9), // Whether resource is in the process of modifying allowed node scores pcmk__rsc_updating_nodes = (1ULL << 10), // Whether resource is in the process of scheduling actions to restart pcmk__rsc_restarting = (1ULL << 11), // Whether resource must be stopped (instead of demoted) if it is failed pcmk__rsc_stop_if_failed = (1ULL << 12), // Whether a reload action has been scheduled for resource pcmk__rsc_reload = (1ULL << 13), // Whether resource is a remote connection allowed to run on a remote node pcmk__rsc_remote_nesting_allowed = (1ULL << 14), // Whether resource has \c PCMK_META_CRITICAL meta-attribute enabled pcmk__rsc_critical = (1ULL << 15), // Whether resource is considered failed pcmk__rsc_failed = (1ULL << 16), // Flag for non-scheduler code to use to detect recursion loops pcmk__rsc_detect_loop = (1ULL << 17), // Whether resource is a Pacemaker Remote connection pcmk__rsc_is_remote_connection = (1ULL << 18), // Whether resource has pending start action in history pcmk__rsc_start_pending = (1ULL << 19), // Whether resource is probed only on nodes marked exclusive pcmk__rsc_exclusive_probes = (1ULL << 20), /* * Whether resource is multiply active with recovery set to * \c PCMK_VALUE_STOP_UNEXPECTED */ pcmk__rsc_stop_unexpected = (1ULL << 22), // Whether resource is allowed to live-migrate pcmk__rsc_migratable = (1ULL << 23), // Whether resource has an ignorable failure pcmk__rsc_ignore_failure = (1ULL << 24), // Whether resource is an implicit container resource for a bundle replica pcmk__rsc_replica_container = (1ULL << 25), // Whether resource, its node, or entire cluster is in maintenance mode pcmk__rsc_maintenance = (1ULL << 26), // Whether resource can be started or promoted only on quorate nodes pcmk__rsc_needs_quorum = (1ULL << 28), // Whether resource requires fencing before recovery if on unclean node pcmk__rsc_needs_fencing = (1ULL << 29), // Whether resource can be started or promoted only on unfenced nodes pcmk__rsc_needs_unfencing = (1ULL << 30), }; // Where to look for a resource enum pcmk__rsc_node { pcmk__rsc_node_none = 0U, // Nowhere pcmk__rsc_node_assigned = (1U << 0), // Where resource is assigned pcmk__rsc_node_current = (1U << 1), // Where resource is running pcmk__rsc_node_pending = (1U << 2), // Where resource is pending }; //! Resource assignment methods (implementation defined by libpacemaker) typedef struct pcmk__assignment_methods pcmk__assignment_methods_t; //! Resource object methods typedef struct { /*! * \internal * \brief Parse variant-specific resource XML from CIB into struct members * * \param[in,out] rsc Partially unpacked resource * \param[in,out] scheduler Scheduler data * * \return TRUE if resource was unpacked successfully, otherwise FALSE */ gboolean (*unpack)(pcmk_resource_t *rsc, pcmk_scheduler_t *scheduler); /*! * \internal * \brief Search for a resource ID in a resource and its children * * \param[in] rsc Search this resource and its children * \param[in] id Search for this resource ID * \param[in] on_node If not NULL, limit search to resources on this node * \param[in] flags Group of enum pe_find flags * * \return Resource that matches search criteria if any, otherwise NULL */ pcmk_resource_t *(*find_rsc)(pcmk_resource_t *rsc, const char *search, const pcmk_node_t *node, int flags); /*! * \internal * \brief Get value of a resource instance attribute * * \param[in,out] rsc Resource to check * \param[in] node Node to use to evaluate rules * \param[in] create Ignored * \param[in] name Name of instance attribute to check * \param[in,out] scheduler Scheduler data * * \return Value of requested attribute if available, otherwise NULL * \note The caller is responsible for freeing the result using free(). */ char *(*parameter)(pcmk_resource_t *rsc, pcmk_node_t *node, gboolean create, const char *name, pcmk_scheduler_t *scheduler); /*! * \internal * \brief Check whether a resource is active * * \param[in] rsc Resource to check * \param[in] all If \p rsc is collective, all instances must be active * * \return TRUE if \p rsc is active, otherwise FALSE */ gboolean (*active)(pcmk_resource_t *rsc, gboolean all); /*! * \internal * \brief Get resource's current or assigned role * * \param[in] rsc Resource to check * \param[in] current If TRUE, check current role, otherwise assigned role * * \return Current or assigned role of \p rsc */ enum rsc_role_e (*state)(const pcmk_resource_t *rsc, gboolean current); /*! * \internal * \brief List nodes where a resource (or any of its children) is * * \param[in] rsc Resource to check * \param[out] list List to add result to * \param[in] target Which resource conditions to target (group of * enum pcmk__rsc_node flags) * * \return If list contains only one node, that node, otherwise NULL */ pcmk_node_t *(*location)(const pcmk_resource_t *rsc, GList **list, uint32_t target); /*! * \internal * \brief Free all memory used by a resource * * \param[in,out] rsc Resource to free */ void (*free)(pcmk_resource_t *rsc); /*! * \internal * \brief Increment cluster's instance counts for a resource * * Given a resource, increment its cluster's ninstances, disabled_resources, * and blocked_resources counts for the resource and its descendants. * * \param[in,out] rsc Resource to count */ void (*count)(pcmk_resource_t *rsc); /*! * \internal * \brief Check whether a given resource is in a list of resources * * \param[in] rsc Resource ID to check for * \param[in] only_rsc List of resource IDs to check * \param[in] check_parent If TRUE, check top ancestor as well * * \return TRUE if \p rsc, its top parent if requested, or '*' is in * \p only_rsc, otherwise FALSE */ gboolean (*is_filtered)(const pcmk_resource_t *rsc, GList *only_rsc, gboolean check_parent); /*! * \internal * \brief Find a node (and optionally count all) where resource is active * * \param[in] rsc Resource to check * \param[out] count_all If not NULL, set this to count of active nodes * \param[out] count_clean If not NULL, set this to count of clean nodes * * \return A node where the resource is active, preferring the source node * if the resource is involved in a partial migration, or a clean, * online node if the resource's \c PCMK_META_REQUIRES is * \c PCMK_VALUE_QUORUM or \c PCMK_VALUE_NOTHING, otherwise \c NULL. */ pcmk_node_t *(*active_node)(const pcmk_resource_t *rsc, unsigned int *count_all, unsigned int *count_clean); /*! * \internal * \brief Get maximum resource instances per node * * \param[in] rsc Resource to check * * \return Maximum number of \p rsc instances that can be active on one node */ unsigned int (*max_per_node)(const pcmk_resource_t *rsc); } pcmk__rsc_methods_t; // Implementation of pcmk__resource_private_t struct pcmk__resource_private { enum pcmk__rsc_variant variant; // Resource variant void *variant_opaque; // Variant-specific data char *history_id; // Resource instance ID in history GHashTable *meta; // Resource meta-attributes GHashTable *utilization; // Resource utilization attributes int priority; // Priority relative other resources int promotion_priority; // Promotion priority on assigned node enum rsc_role_e orig_role; // Resource's role at start of transition enum rsc_role_e next_role; // Resource's role at end of transition int stickiness; // Extra preference for current node guint failure_expiration_ms; // Failures expire after this much time int ban_after_failures; // Ban from node after this many failures guint remote_reconnect_ms; // Retry interval for remote connections char *pending_action; // Pending action in history, if any const pcmk_node_t *pending_node;// Node on which pending_action is happening time_t lock_time; // When shutdown lock started const pcmk_node_t *lock_node; // Node that resource is shutdown-locked to GList *actions; // Actions scheduled for resource GList *children; // Resource's child resources, if any pcmk_resource_t *parent; // Resource's parent resource, if any pcmk_scheduler_t *scheduler; // Scheduler data containing resource // Resource configuration (possibly expanded from template) xmlNode *xml; // Original resource configuration, if using template xmlNode *orig_xml; // Configuration of resource operations (possibly expanded from template) xmlNode *ops_xml; /* * Resource parameters may have node-attribute-based rules, which means the * values can vary by node. This table has node names as keys and parameter * name/value tables as values. Use pe_rsc_params() to get the table for a * given node rather than use this directly. */ GHashTable *parameter_cache; /* A "launcher" is defined in one of these ways: * * - A Pacemaker Remote connection for a guest node or bundle node has its * launcher set to the resource that starts the guest or the bundle * replica's container. * * - If the user configures the PCMK__META_CONTAINER meta-attribute for this * resource, the launcher is set to that. * * If the launcher is a Pacemaker Remote connection resource, this * resource may run only on the node created by that connection. * * Otherwise, this resource will be colocated with and ordered after the * launcher, and failures of this resource will cause the launcher to be * recovered instead of this one. This is appropriate for monitoring-only * resources that represent a service launched by the other resource. */ pcmk_resource_t *launcher; // Resources launched by this one, if any (pcmk_resource_t *) GList *launched; // What to do if the resource is incorrectly active on multiple nodes enum pcmk__multiply_active multiply_active_policy; /* The assigned node (if not NULL) is the one where the resource *should* * be active by the end of the current scheduler transition. Only primitive * resources have an assigned node. This is a node copy (created by * pe__copy_node()) and so must be freed using pcmk__free_node_copy(). * * @TODO This should probably be part of the primitive variant data. */ pcmk_node_t *assigned_node; /* The active nodes are ones where the resource is (or might be, if * insufficient information is available to be sure) already active at the * start of the current scheduler transition. * * For primitive resources, there should be at most one, but could be more * if it is (incorrectly) multiply active. For collective resources, this * combines active nodes of all descendants. */ GList *active_nodes; /* The next two tables store node copies (created by pe__copy_node()), which * share some members with the original node objects and must be freed with * pcmk__free_node_copy(). */ // Nodes where resource has been probed (key is node ID, not name) GHashTable *probed_nodes; // Nodes where resource is allowed to run (key is node ID, not name) GHashTable *allowed_nodes; // The source node, if migrate_to completed but migrate_from has not pcmk_node_t *partial_migration_source; // The destination node, if migrate_to completed but migrate_from has not pcmk_node_t *partial_migration_target; // Source nodes where stop is needed after migrate_from and migrate_to GList *dangling_migration_sources; /* Pay special attention to whether you want to use with_this_colocations * and this_with_colocations directly, which include only colocations * explicitly involving this resource, or call libpacemaker's * pcmk__with_this_colocations() and pcmk__this_with_colocations() * functions, which may return relevant colocations involving the resource's * ancestors as well. */ // Colocations of other resources with this one GList *with_this_colocations; // Colocations of this resource with others GList *this_with_colocations; GList *location_constraints; // Location constraints for resource GList *ticket_constraints; // Ticket constraints for resource const pcmk__rsc_methods_t *fns; // Resource object methods const pcmk__assignment_methods_t *cmds; // Resource assignment methods }; +void pcmk__free_resource(gpointer user_data); const char *pcmk__multiply_active_text(const pcmk_resource_t *rsc); /*! * \internal * \brief Get node where resource is currently active (if any) * * \param[in] rsc Resource to check * * \return Node that \p rsc is active on, if any, otherwise NULL */ static inline pcmk_node_t * pcmk__current_node(const pcmk_resource_t *rsc) { if (rsc == NULL) { return NULL; } return rsc->priv->fns->active_node(rsc, NULL, NULL); } #ifdef __cplusplus } #endif #endif // PCMK__CRM_COMMON_RESOURCES_INTERNAL__H diff --git a/include/crm/common/scheduler.h b/include/crm/common/scheduler.h index 5d12c9db28..86d0cd7d9e 100644 --- a/include/crm/common/scheduler.h +++ b/include/crm/common/scheduler.h @@ -1,116 +1,120 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #ifndef PCMK__CRM_COMMON_SCHEDULER__H #define PCMK__CRM_COMMON_SCHEDULER__H #include // time_t #include // xmlNode #include // guint, GList, GHashTable #include // crm_time_t #include #include #include #include #include #include #ifdef __cplusplus extern "C" { #endif /*! * \file * \brief Scheduler API * \ingroup core */ // NOTE: sbd (as of at least 1.5.2) uses this enum //! Possible responses to loss of quorum enum pe_quorum_policy { pcmk_no_quorum_freeze, //!< Do not recover resources from outside partition pcmk_no_quorum_stop, //!< Stop all resources in partition pcmk_no_quorum_ignore, //!< Act as if partition still holds quorum pcmk_no_quorum_fence, //!< Fence all nodes in partition pcmk_no_quorum_demote, //!< Demote promotable resources and stop all others #if !defined(PCMK_ALLOW_DEPRECATED) || (PCMK_ALLOW_DEPRECATED == 1) // NOTE: sbd (as of at least 1.5.2) uses this value //! \deprecated Use pcmk_no_quorum_freeze instead no_quorum_freeze = pcmk_no_quorum_freeze, // NOTE: sbd (as of at least 1.5.2) uses this value //! \deprecated Use pcmk_no_quorum_stop instead no_quorum_stop = pcmk_no_quorum_stop, // NOTE: sbd (as of at least 1.5.2) uses this value //! \deprecated Use pcmk_no_quorum_ignore instead no_quorum_ignore = pcmk_no_quorum_ignore, //! \deprecated Use pcmk_no_quorum_fence instead no_quorum_suicide = pcmk_no_quorum_fence, // NOTE: sbd (as of at least 1.5.2) uses this value //! \deprecated Use pcmk_no_quorum_demote instead no_quorum_demote = pcmk_no_quorum_demote, #endif }; //! \internal Do not use typedef struct pcmk__scheduler_private pcmk__scheduler_private_t; /* Implementation of pcmk_scheduler_t * * @COMPAT Drop this struct once all members are moved to * pcmk__scheduler_private_t, and repoint pcmk_scheduler_t to that */ //! \deprecated Do not use (public access will be removed in a future release) struct pcmk__scheduler { // Be careful about when each piece of information is available and final // NOTE: sbd (as of at least 1.5.2) uses this //! \deprecated Set scheduler input with pcmk_set_scheduler_cib() instead xmlNode *input; // CIB XML // NOTE: sbd (as of at least 1.5.2) uses this //! \deprecated Call pcmk_get_dc() instead pcmk_node_t *dc_node; // Node object for DC // NOTE: sbd (as of at least 1.5.2) uses this //! \deprecated Call pcmk_has_quorum() to check quorum uint64_t flags; // Group of enum pcmk__scheduler_flags // NOTE: sbd (as of at least 1.5.2) uses this //! \deprecated Call pcmk_get_no_quorum_policy() to get no-quorum policy enum pe_quorum_policy no_quorum_policy; // Response to loss of quorum // NOTE: sbd (as of at least 1.5.2) uses this //! \deprecated Call pcmk_find_node() to find a node instead GList *nodes; // Nodes in cluster pcmk__scheduler_private_t *priv; //!< \internal For Pacemaker use only }; +pcmk_scheduler_t *pcmk_new_scheduler(void); +void pcmk_reset_scheduler(pcmk_scheduler_t *scheduler); +void pcmk_free_scheduler(pcmk_scheduler_t *scheduler); + pcmk_node_t *pcmk_get_dc(const pcmk_scheduler_t *scheduler); enum pe_quorum_policy pcmk_get_no_quorum_policy(const pcmk_scheduler_t *scheduler); int pcmk_set_scheduler_cib(pcmk_scheduler_t *scheduler, xmlNode *cib); bool pcmk_has_quorum(const pcmk_scheduler_t *scheduler); pcmk_node_t *pcmk_find_node(const pcmk_scheduler_t *scheduler, const char *node_name); #ifdef __cplusplus } #endif #endif // PCMK__CRM_COMMON_SCHEDULER__H diff --git a/include/crm/common/scheduler_internal.h b/include/crm/common/scheduler_internal.h index 39e806d92c..8ff4eb9886 100644 --- a/include/crm/common/scheduler_internal.h +++ b/include/crm/common/scheduler_internal.h @@ -1,286 +1,300 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #ifndef PCMK__CRM_COMMON_SCHEDULER_INTERNAL__H #define PCMK__CRM_COMMON_SCHEDULER_INTERNAL__H #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #ifdef __cplusplus extern "C" { #endif enum pcmk__check_parameters { /* Clear fail count if parameters changed for un-expired start or monitor * last_failure. */ pcmk__check_last_failure, /* Clear fail count if parameters changed for start, monitor, promote, or * migrate_from actions for active resources. */ pcmk__check_active, }; // Scheduling options and conditions enum pcmk__scheduler_flags { // No scheduler flags set (compare with equality rather than bit set) pcmk__sched_none = 0ULL, /* These flags are dynamically determined conditions */ // Whether partition has quorum (via \c PCMK_XA_HAVE_QUORUM attribute) //! \deprecated Call pcmk_has_quorum() to check quorum instead pcmk__sched_quorate = (1ULL << 0), // Whether cluster is symmetric (via symmetric-cluster property) pcmk__sched_symmetric_cluster = (1ULL << 1), // Whether scheduling encountered a non-configuration error pcmk__sched_processing_error = (1ULL << 2), // Whether cluster is in maintenance mode (via maintenance-mode property) pcmk__sched_in_maintenance = (1ULL << 3), // Whether fencing is enabled (via stonith-enabled property) pcmk__sched_fencing_enabled = (1ULL << 4), // Whether cluster has a fencing resource (via CIB resources) /*! \deprecated To indicate the cluster has a fencing resource, add either a * fencing resource configuration or the have-watchdog cluster option to the * input CIB */ pcmk__sched_have_fencing = (1ULL << 5), // Whether any resource provides or requires unfencing (via CIB resources) pcmk__sched_enable_unfencing = (1ULL << 6), // Whether concurrent fencing is allowed (via concurrent-fencing property) pcmk__sched_concurrent_fencing = (1ULL << 7), /* * Whether resources removed from the configuration should be stopped (via * stop-orphan-resources property) */ pcmk__sched_stop_removed_resources = (1ULL << 8), /* * Whether recurring actions removed from the configuration should be * cancelled (via stop-orphan-actions property) */ pcmk__sched_cancel_removed_actions = (1ULL << 9), // Whether to stop all resources (via stop-all-resources property) pcmk__sched_stop_all = (1ULL << 10), // Whether scheduler processing encountered a warning pcmk__sched_processing_warning = (1ULL << 11), /* * Whether start failure should be treated as if * \c PCMK_META_MIGRATION_THRESHOLD is 1 (via * \c PCMK_OPT_START_FAILURE_IS_FATAL property) */ pcmk__sched_start_failure_fatal = (1ULL << 12), // Whether unseen nodes should be fenced (via startup-fencing property) pcmk__sched_startup_fencing = (1ULL << 14), /* * Whether resources should be left stopped when their node shuts down * cleanly (via shutdown-lock property) */ pcmk__sched_shutdown_lock = (1ULL << 15), /* * Whether resources' current state should be probed (when unknown) before * scheduling any other actions (via the enable-startup-probes property) */ pcmk__sched_probe_resources = (1ULL << 16), // Whether the CIB status section has been parsed yet pcmk__sched_have_status = (1ULL << 17), // Whether the cluster includes any Pacemaker Remote nodes (via CIB) pcmk__sched_have_remote_nodes = (1ULL << 18), /* The remaining flags are scheduling options that must be set explicitly */ /* * Whether to skip unpacking the CIB status section and stop the scheduling * sequence after applying node-specific location criteria (skipping * assignment, ordering, actions, etc.). */ pcmk__sched_location_only = (1ULL << 20), // Whether sensitive resource attributes have been masked pcmk__sched_sanitized = (1ULL << 21), // Skip counting of total, disabled, and blocked resource instances pcmk__sched_no_counts = (1ULL << 23), // Whether node scores should be output instead of logged pcmk__sched_output_scores = (1ULL << 25), // Whether to show node and resource utilization (in log or output) pcmk__sched_show_utilization = (1ULL << 26), /* * Whether to stop the scheduling sequence after unpacking the CIB, * calculating cluster status, and applying node health (skipping * applying node-specific location criteria, assignment, etc.) */ pcmk__sched_validate_only = (1ULL << 27), }; // Implementation of pcmk__scheduler_private_t struct pcmk__scheduler_private { // Be careful about when each piece of information is available and final char *local_node_name; // Name of node running scheduler (if known) crm_time_t *now; // Time to use when evaluating rules pcmk__output_t *out; // Output object for displaying messages GHashTable *options; // Cluster options const char *fence_action; // Default fencing action guint fence_timeout_ms; // Default fencing action timeout (in ms) guint priority_fencing_ms; // Priority-based fencing delay (in ms) guint shutdown_lock_ms; // How long to lock resources (in ms) guint node_pending_ms; // Pending join times out after this (in ms) const char *placement_strategy; // Value of placement-strategy property xmlNode *rsc_defaults; // Configured resource defaults xmlNode *op_defaults; // Configured operation defaults GList *resources; // Resources in cluster GHashTable *templates; // Key = template ID, value = resource list GHashTable *tags; // Key = tag ID, value = element list GList *actions; // All scheduled actions GHashTable *singletons; // Scheduled non-resource actions int next_action_id; // Counter used as ID for actions xmlNode *failed; // History entries of failed actions GList *param_check; // History entries that need to be checked GList *stop_needed; // Containers that need stop actions GList *location_constraints; // Location constraints GList *colocation_constraints; // Colocation constraints GList *ordering_constraints; // Ordering constraints GHashTable *ticket_constraints; // Key = ticket ID, value = pcmk__ticket_t int next_ordering_id; // Counter used as ID for orderings int ninstances; // Total number of resource instances int blocked_resources; // Number of blocked resources in cluster int disabled_resources; // Number of disabled resources in cluster time_t recheck_by; // Hint to controller when to reschedule xmlNode *graph; // Transition graph int synapse_count; // Number of transition graph synapses }; // Group of enum pcmk__warnings flags for warnings we want to log once extern uint32_t pcmk__warnings; /*! * \internal * \brief Log a resource-tagged message at info severity * * \param[in] rsc Tag message with this resource's ID * \param[in] fmt... printf(3)-style format and arguments */ #define pcmk__rsc_info(rsc, fmt, args...) \ crm_log_tag(LOG_INFO, ((rsc) == NULL)? "" : (rsc)->id, (fmt), ##args) /*! * \internal * \brief Log a resource-tagged message at debug severity * * \param[in] rsc Tag message with this resource's ID * \param[in] fmt... printf(3)-style format and arguments */ #define pcmk__rsc_debug(rsc, fmt, args...) \ crm_log_tag(LOG_DEBUG, ((rsc) == NULL)? "" : (rsc)->id, (fmt), ##args) /*! * \internal * \brief Log a resource-tagged message at trace severity * * \param[in] rsc Tag message with this resource's ID * \param[in] fmt... printf(3)-style format and arguments */ #define pcmk__rsc_trace(rsc, fmt, args...) \ crm_log_tag(LOG_TRACE, ((rsc) == NULL)? "" : (rsc)->id, (fmt), ##args) /*! * \internal * \brief Log an error and remember that current scheduler input has errors * * \param[in,out] scheduler Scheduler data * \param[in] fmt... printf(3)-style format and arguments */ #define pcmk__sched_err(scheduler, fmt...) do { \ pcmk__set_scheduler_flags((scheduler), \ pcmk__sched_processing_error); \ crm_err(fmt); \ } while (0) /*! * \internal * \brief Log a warning and remember that current scheduler input has warnings * * \param[in,out] scheduler Scheduler data * \param[in] fmt... printf(3)-style format and arguments */ #define pcmk__sched_warn(scheduler, fmt...) do { \ pcmk__set_scheduler_flags((scheduler), \ pcmk__sched_processing_warning); \ crm_warn(fmt); \ } while (0) /*! * \internal * \brief Set scheduler flags * * \param[in,out] scheduler Scheduler data * \param[in] flags_to_set Group of enum pcmk__scheduler_flags to set */ #define pcmk__set_scheduler_flags(scheduler, flags_to_set) do { \ (scheduler)->flags = pcmk__set_flags_as(__func__, __LINE__, \ LOG_TRACE, "Scheduler", crm_system_name, \ (scheduler)->flags, (flags_to_set), #flags_to_set); \ } while (0) /*! * \internal * \brief Clear scheduler flags * * \param[in,out] scheduler Scheduler data * \param[in] flags_to_clear Group of enum pcmk__scheduler_flags to clear */ #define pcmk__clear_scheduler_flags(scheduler, flags_to_clear) do { \ (scheduler)->flags = pcmk__clear_flags_as(__func__, __LINE__, \ LOG_TRACE, "Scheduler", crm_system_name, \ (scheduler)->flags, (flags_to_clear), #flags_to_clear); \ } while (0) +void pcmk__set_scheduler_defaults(pcmk_scheduler_t *scheduler); +time_t pcmk__scheduler_epoch_time(pcmk_scheduler_t *scheduler); +void pcmk__update_recheck_time(time_t recheck, pcmk_scheduler_t *scheduler, + const char *reason); + +void pcmk__add_param_check(const xmlNode *rsc_op, pcmk_resource_t *rsc, + pcmk_node_t *node, enum pcmk__check_parameters, + pcmk_scheduler_t *scheduler); +void pcmk__foreach_param_check(pcmk_scheduler_t *scheduler, + void (*cb)(pcmk_resource_t*, pcmk_node_t*, + const xmlNode*, + enum pcmk__check_parameters)); +void pcmk__free_param_checks(pcmk_scheduler_t *scheduler); + #ifdef __cplusplus } #endif #endif // PCMK__CRM_COMMON_SCHEDULER_INTERNAL__H diff --git a/include/crm/pengine/internal.h b/include/crm/pengine/internal.h index 41ef4dc215..a9f4791cdb 100644 --- a/include/crm/pengine/internal.h +++ b/include/crm/pengine/internal.h @@ -1,413 +1,400 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #ifndef PCMK__CRM_PENGINE_INTERNAL__H #define PCMK__CRM_PENGINE_INTERNAL__H #include #include #include #include #include #include #include #include #include #include #ifdef __cplusplus extern "C" { #endif const char *pe__resource_description(const pcmk_resource_t *rsc, uint32_t show_opts); bool pe__clone_is_ordered(const pcmk_resource_t *clone); int pe__set_clone_flag(pcmk_resource_t *clone, enum pcmk__clone_flags flag); bool pe__clone_flag_is_set(const pcmk_resource_t *clone, uint32_t flags); bool pe__group_flag_is_set(const pcmk_resource_t *group, uint32_t flags); pcmk_resource_t *pe__last_group_member(const pcmk_resource_t *group); const pcmk_resource_t *pe__const_top_resource(const pcmk_resource_t *rsc, bool include_bundle); int pe__clone_max(const pcmk_resource_t *clone); int pe__clone_node_max(const pcmk_resource_t *clone); int pe__clone_promoted_max(const pcmk_resource_t *clone); int pe__clone_promoted_node_max(const pcmk_resource_t *clone); void pe__create_clone_notifications(pcmk_resource_t *clone); void pe__free_clone_notification_data(pcmk_resource_t *clone); void pe__create_clone_notif_pseudo_ops(pcmk_resource_t *clone, pcmk_action_t *start, pcmk_action_t *started, pcmk_action_t *stop, pcmk_action_t *stopped); pcmk_action_t *pe__new_rsc_pseudo_action(pcmk_resource_t *rsc, const char *task, bool optional, bool runnable); void pe__create_promotable_pseudo_ops(pcmk_resource_t *clone, bool any_promoting, bool any_demoting); bool pe_can_fence(const pcmk_scheduler_t *scheduler, const pcmk_node_t *node); char *native_parameter(pcmk_resource_t *rsc, pcmk_node_t *node, gboolean create, const char *name, pcmk_scheduler_t *scheduler); pcmk_node_t *native_location(const pcmk_resource_t *rsc, GList **list, uint32_t target); void native_add_running(pcmk_resource_t *rsc, pcmk_node_t *node, pcmk_scheduler_t *scheduler, gboolean failed); gboolean native_unpack(pcmk_resource_t *rsc, pcmk_scheduler_t *scheduler); gboolean group_unpack(pcmk_resource_t *rsc, pcmk_scheduler_t *scheduler); gboolean clone_unpack(pcmk_resource_t *rsc, pcmk_scheduler_t *scheduler); gboolean pe__unpack_bundle(pcmk_resource_t *rsc, pcmk_scheduler_t *scheduler); pcmk_resource_t *native_find_rsc(pcmk_resource_t *rsc, const char *id, const pcmk_node_t *node, int flags); gboolean native_active(pcmk_resource_t *rsc, gboolean all); gboolean group_active(pcmk_resource_t *rsc, gboolean all); gboolean clone_active(pcmk_resource_t *rsc, gboolean all); gboolean pe__bundle_active(pcmk_resource_t *rsc, gboolean all); gchar *pcmk__native_output_string(const pcmk_resource_t *rsc, const char *name, const pcmk_node_t *node, uint32_t show_opts, const char *target_role, bool show_nodes); int pe__name_and_nvpairs_xml(pcmk__output_t *out, bool is_list, const char *tag_name, ...) G_GNUC_NULL_TERMINATED; char *pe__node_display_name(pcmk_node_t *node, bool print_detail); // Clone notifications (pe_notif.c) void pe__order_notifs_after_fencing(const pcmk_action_t *action, pcmk_resource_t *rsc, pcmk_action_t *stonith_op); // Resource output methods int pe__clone_xml(pcmk__output_t *out, va_list args); int pe__clone_default(pcmk__output_t *out, va_list args); int pe__group_xml(pcmk__output_t *out, va_list args); int pe__group_default(pcmk__output_t *out, va_list args); int pe__bundle_xml(pcmk__output_t *out, va_list args); int pe__bundle_html(pcmk__output_t *out, va_list args); int pe__bundle_text(pcmk__output_t *out, va_list args); int pe__node_html(pcmk__output_t *out, va_list args); int pe__node_text(pcmk__output_t *out, va_list args); int pe__node_xml(pcmk__output_t *out, va_list args); int pe__resource_xml(pcmk__output_t *out, va_list args); int pe__resource_html(pcmk__output_t *out, va_list args); int pe__resource_text(pcmk__output_t *out, va_list args); void native_free(pcmk_resource_t *rsc); void group_free(pcmk_resource_t *rsc); void clone_free(pcmk_resource_t *rsc); void pe__free_bundle(pcmk_resource_t *rsc); enum rsc_role_e native_resource_state(const pcmk_resource_t *rsc, gboolean current); enum rsc_role_e group_resource_state(const pcmk_resource_t *rsc, gboolean current); enum rsc_role_e clone_resource_state(const pcmk_resource_t *rsc, gboolean current); enum rsc_role_e pe__bundle_resource_state(const pcmk_resource_t *rsc, gboolean current); void pe__count_common(pcmk_resource_t *rsc); void pe__count_bundle(pcmk_resource_t *rsc); void common_free(pcmk_resource_t *rsc); pcmk_node_t *pe__copy_node(const pcmk_node_t *this_node); -time_t get_effective_time(pcmk_scheduler_t *scheduler); /* Failure handling utilities (from failcounts.c) */ int pe_get_failcount(const pcmk_node_t *node, pcmk_resource_t *rsc, time_t *last_failure, uint32_t flags, const xmlNode *xml_op); pcmk_action_t *pe__clear_failcount(pcmk_resource_t *rsc, const pcmk_node_t *node, const char *reason, pcmk_scheduler_t *scheduler); /* Functions for finding/counting a resource's active nodes */ bool pe__count_active_node(const pcmk_resource_t *rsc, pcmk_node_t *node, pcmk_node_t **active, unsigned int *count_all, unsigned int *count_clean); pcmk_node_t *pe__find_active_requires(const pcmk_resource_t *rsc, unsigned int *count); /* Binary like operators for lists of nodes */ GHashTable *pe__node_list2table(const GList *list); pcmk_action_t *get_pseudo_op(const char *name, pcmk_scheduler_t *scheduler); gboolean order_actions(pcmk_action_t *first, pcmk_action_t *then, uint32_t flags); void pe__show_node_scores_as(const char *file, const char *function, int line, bool to_log, const pcmk_resource_t *rsc, const char *comment, GHashTable *nodes, pcmk_scheduler_t *scheduler); #define pe__show_node_scores(level, rsc, text, nodes, scheduler) \ pe__show_node_scores_as(__FILE__, __func__, __LINE__, \ (level), (rsc), (text), (nodes), (scheduler)) GHashTable *pcmk__unpack_action_meta(pcmk_resource_t *rsc, const pcmk_node_t *node, const char *action_name, guint interval_ms, const xmlNode *action_config); GHashTable *pcmk__unpack_action_rsc_params(const xmlNode *action_xml, GHashTable *node_attrs, pcmk_scheduler_t *data_set); xmlNode *pcmk__find_action_config(const pcmk_resource_t *rsc, const char *action_name, guint interval_ms, bool include_disabled); enum pcmk__requires pcmk__action_requires(const pcmk_resource_t *rsc, const char *action_name); enum pcmk__on_fail pcmk__parse_on_fail(const pcmk_resource_t *rsc, const char *action_name, guint interval_ms, const char *value); enum rsc_role_e pcmk__role_after_failure(const pcmk_resource_t *rsc, const char *action_name, enum pcmk__on_fail on_fail, GHashTable *meta); pcmk_action_t *custom_action(pcmk_resource_t *rsc, char *key, const char *task, const pcmk_node_t *on_node, gboolean optional, pcmk_scheduler_t *scheduler); #define delete_key(rsc) pcmk__op_key((rsc)->id, PCMK_ACTION_DELETE, 0) #define stop_key(rsc) pcmk__op_key((rsc)->id, PCMK_ACTION_STOP, 0) #define reload_key(rsc) pcmk__op_key((rsc)->id, PCMK_ACTION_RELOAD_AGENT, 0) #define start_key(rsc) pcmk__op_key((rsc)->id, PCMK_ACTION_START, 0) #define promote_key(rsc) pcmk__op_key((rsc)->id, PCMK_ACTION_PROMOTE, 0) #define demote_key(rsc) pcmk__op_key((rsc)->id, PCMK_ACTION_DEMOTE, 0) #define delete_action(rsc, node, optional) \ custom_action((rsc), delete_key(rsc), PCMK_ACTION_DELETE, \ (node), (optional), (rsc)->priv->scheduler) #define stop_action(rsc, node, optional) \ custom_action((rsc), stop_key(rsc), PCMK_ACTION_STOP, \ (node), (optional), (rsc)->priv->scheduler) #define start_action(rsc, node, optional) \ custom_action((rsc), start_key(rsc), PCMK_ACTION_START, \ (node), (optional), (rsc)->priv->scheduler) #define promote_action(rsc, node, optional) \ custom_action((rsc), promote_key(rsc), PCMK_ACTION_PROMOTE, \ (node), (optional), (rsc)->priv->scheduler) #define demote_action(rsc, node, optional) \ custom_action((rsc), demote_key(rsc), PCMK_ACTION_DEMOTE, \ (node), (optional), (rsc)->priv->scheduler) pcmk_action_t *find_first_action(const GList *input, const char *uuid, const char *task, const pcmk_node_t *on_node); enum pcmk__action_type get_complex_task(const pcmk_resource_t *rsc, const char *name); GList *find_actions(GList *input, const char *key, const pcmk_node_t *on_node); GList *find_actions_exact(GList *input, const char *key, const pcmk_node_t *on_node); GList *pe__resource_actions(const pcmk_resource_t *rsc, const pcmk_node_t *node, const char *task, bool require_node); -extern void pe_free_action(pcmk_action_t *action); - void resource_location(pcmk_resource_t *rsc, const pcmk_node_t *node, int score, const char *tag, pcmk_scheduler_t *scheduler); int pe__is_newer_op(const xmlNode *xml_a, const xmlNode *xml_b); extern gint sort_op_by_callid(gconstpointer a, gconstpointer b); gboolean get_target_role(const pcmk_resource_t *rsc, enum rsc_role_e *role); void pe__set_next_role(pcmk_resource_t *rsc, enum rsc_role_e role, const char *why); extern void destroy_ticket(gpointer data); pcmk__ticket_t *ticket_new(const char *ticket_id, pcmk_scheduler_t *scheduler); // Resources for manipulating resource names const char *pe_base_name_end(const char *id); char *clone_strip(const char *last_rsc_id); char *clone_zero(const char *last_rsc_id); static inline bool pe_base_name_eq(const pcmk_resource_t *rsc, const char *id) { if (id && rsc && rsc->id) { // Number of characters in rsc->id before any clone suffix size_t base_len = pe_base_name_end(rsc->id) - rsc->id + 1; return (strlen(id) == base_len) && !strncmp(id, rsc->id, base_len); } return false; } int pe__target_rc_from_xml(const xmlNode *xml_op); gint pe__cmp_node_name(gconstpointer a, gconstpointer b); bool is_set_recursive(const pcmk_resource_t *rsc, long long flag, bool any); pcmk__op_digest_t *pe__calculate_digests(pcmk_resource_t *rsc, const char *task, guint *interval_ms, const pcmk_node_t *node, const xmlNode *xml_op, GHashTable *overrides, bool calc_secure, pcmk_scheduler_t *scheduler); void pe__free_digests(gpointer ptr); pcmk__op_digest_t *rsc_action_digest_cmp(pcmk_resource_t *rsc, const xmlNode *xml_op, pcmk_node_t *node, pcmk_scheduler_t *scheduler); pcmk_action_t *pe_fence_op(pcmk_node_t *node, const char *op, bool optional, const char *reason, bool priority_delay, pcmk_scheduler_t *scheduler); void trigger_unfencing(pcmk_resource_t *rsc, pcmk_node_t *node, const char *reason, pcmk_action_t *dependency, pcmk_scheduler_t *scheduler); char *pe__action2reason(const pcmk_action_t *action, enum pcmk__action_flags flag); void pe_action_set_reason(pcmk_action_t *action, const char *reason, bool overwrite); void pe__add_action_expected_result(pcmk_action_t *action, int expected_result); void pe__set_resource_flags_recursive(pcmk_resource_t *rsc, uint64_t flags); void pe__clear_resource_flags_recursive(pcmk_resource_t *rsc, uint64_t flags); void pe__clear_resource_flags_on_all(pcmk_scheduler_t *scheduler, uint64_t flag); int pe__rscs_brief_output(pcmk__output_t *out, GList *rsc_list, unsigned int options); void pe_fence_node(pcmk_scheduler_t *scheduler, pcmk_node_t *node, const char *reason, bool priority_delay); pcmk_node_t *pe_create_node(const char *id, const char *uname, const char *type, int score, pcmk_scheduler_t *scheduler); int pe__common_output_text(pcmk__output_t *out, const pcmk_resource_t *rsc, const char *name, const pcmk_node_t *node, unsigned int options); int pe__common_output_html(pcmk__output_t *out, const pcmk_resource_t *rsc, const char *name, const pcmk_node_t *node, unsigned int options); GList *pe__bundle_containers(const pcmk_resource_t *bundle); int pe__bundle_max(const pcmk_resource_t *rsc); bool pe__node_is_bundle_instance(const pcmk_resource_t *bundle, const pcmk_node_t *node); pcmk_resource_t *pe__bundled_resource(const pcmk_resource_t *rsc); const pcmk_resource_t *pe__get_rsc_in_container(const pcmk_resource_t *instance); pcmk_resource_t *pe__first_container(const pcmk_resource_t *bundle); void pe__foreach_bundle_replica(pcmk_resource_t *bundle, bool (*fn)(pcmk__bundle_replica_t *, void *), void *user_data); void pe__foreach_const_bundle_replica(const pcmk_resource_t *bundle, bool (*fn)(const pcmk__bundle_replica_t *, void *), void *user_data); pcmk_resource_t *pe__find_bundle_replica(const pcmk_resource_t *bundle, const pcmk_node_t *node); bool pe__bundle_needs_remote_name(pcmk_resource_t *rsc); const char *pe__add_bundle_remote_name(pcmk_resource_t *rsc, xmlNode *xml, const char *field); bool pe__is_universal_clone(const pcmk_resource_t *rsc, const pcmk_scheduler_t *scheduler); -void pe__add_param_check(const xmlNode *rsc_op, pcmk_resource_t *rsc, - pcmk_node_t *node, enum pcmk__check_parameters, - pcmk_scheduler_t *scheduler); -void pe__foreach_param_check(pcmk_scheduler_t *scheduler, - void (*cb)(pcmk_resource_t*, pcmk_node_t*, - const xmlNode*, - enum pcmk__check_parameters)); -void pe__free_param_checks(pcmk_scheduler_t *scheduler); bool pe__shutdown_requested(const pcmk_node_t *node); -void pe__update_recheck_time(time_t recheck, pcmk_scheduler_t *scheduler, - const char *reason); /*! * \internal * \brief Register xml formatting message functions. * * \param[in,out] out Output object to register messages with */ void pe__register_messages(pcmk__output_t *out); void pe__unpack_dataset_nvpairs(const xmlNode *xml_obj, const char *set_name, const pcmk_rule_input_t *rule_input, GHashTable *hash, const char *always_first, pcmk_scheduler_t *scheduler); bool pe__resource_is_disabled(const pcmk_resource_t *rsc); void pe__clear_resource_history(pcmk_resource_t *rsc, const pcmk_node_t *node); GList *pe__rscs_with_tag(pcmk_scheduler_t *scheduler, const char *tag_name); GList *pe__unames_with_tag(pcmk_scheduler_t *scheduler, const char *tag_name); bool pe__rsc_has_tag(pcmk_scheduler_t *scheduler, const char *rsc, const char *tag); bool pe__uname_has_tag(pcmk_scheduler_t *scheduler, const char *node, const char *tag); bool pe__rsc_running_on_only(const pcmk_resource_t *rsc, const pcmk_node_t *node); bool pe__rsc_running_on_any(pcmk_resource_t *rsc, GList *node_list); GList *pe__filter_rsc_list(GList *rscs, GList *filter); GList * pe__build_node_name_list(pcmk_scheduler_t *scheduler, const char *s); GList * pe__build_rsc_list(pcmk_scheduler_t *scheduler, const char *s); bool pcmk__rsc_filtered_by_node(pcmk_resource_t *rsc, GList *only_node); gboolean pe__bundle_is_filtered(const pcmk_resource_t *rsc, GList *only_rsc, gboolean check_parent); gboolean pe__clone_is_filtered(const pcmk_resource_t *rsc, GList *only_rsc, gboolean check_parent); gboolean pe__group_is_filtered(const pcmk_resource_t *rsc, GList *only_rsc, gboolean check_parent); gboolean pe__native_is_filtered(const pcmk_resource_t *rsc, GList *only_rsc, gboolean check_parent); xmlNode *pe__failed_probe_for_rsc(const pcmk_resource_t *rsc, const char *name); const char *pe__clone_child_id(const pcmk_resource_t *rsc); int pe__sum_node_health_scores(const pcmk_node_t *node, int base_health); int pe__node_health(pcmk_node_t *node); static inline enum pcmk__health_strategy pe__health_strategy(pcmk_scheduler_t *scheduler) { const char *strategy = pcmk__cluster_option(scheduler->priv->options, PCMK_OPT_NODE_HEALTH_STRATEGY); return pcmk__parse_health_strategy(strategy); } #ifdef __cplusplus } #endif #endif // PCMK__CRM_PENGINE_INTERNAL__H diff --git a/include/crm/pengine/status.h b/include/crm/pengine/status.h index 80fae41ae4..4d894dc96e 100644 --- a/include/crm/pengine/status.h +++ b/include/crm/pengine/status.h @@ -1,68 +1,56 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #ifndef PCMK__CRM_PENGINE_STATUS__H # define PCMK__CRM_PENGINE_STATUS__H # include // gboolean # include // bool # include // pcmk_is_set() # include # include // pcmk_node_t, pcmk_resource_t, etc. # include #ifdef __cplusplus extern "C" { #endif /*! * \file * \brief Cluster status and scheduling * \ingroup pengine */ const char *rsc_printable_id(const pcmk_resource_t *rsc); // NOTE: sbd (as of at least 1.5.2) uses this gboolean cluster_status(pcmk_scheduler_t *scheduler); -// NOTE: sbd (as of at least 1.5.2) uses this -pcmk_scheduler_t *pe_new_working_set(void); - -// NOTE: sbd (as of at least 1.5.2) uses this -void pe_free_working_set(pcmk_scheduler_t *scheduler); - -void set_working_set_defaults(pcmk_scheduler_t *scheduler); -void cleanup_calculations(pcmk_scheduler_t *scheduler); - -// NOTE: sbd (as of at least 1.5.2) uses this -void pe_reset_working_set(pcmk_scheduler_t *scheduler); - pcmk_resource_t *pe_find_resource(GList *rsc_list, const char *id); pcmk_resource_t *pe_find_resource_with_flags(GList *rsc_list, const char *id, enum pe_find flags); pcmk_node_t *pe_find_node_id(const GList *node_list, const char *id); pcmk_node_t *pe_find_node_any(const GList *node_list, const char *id, const char *node_name); GList *find_operations(const char *rsc, const char *node, gboolean active_filter, pcmk_scheduler_t *scheduler); void calculate_active_ops(const GList *sorted_op_list, int *start_index, int *stop_index); int pe_bundle_replicas(const pcmk_resource_t *rsc); #ifdef __cplusplus } #endif #if !defined(PCMK_ALLOW_DEPRECATED) || (PCMK_ALLOW_DEPRECATED == 1) #include #include #endif #endif diff --git a/include/crm/pengine/status_compat.h b/include/crm/pengine/status_compat.h index 5488bb8e7d..af45129d2d 100644 --- a/include/crm/pengine/status_compat.h +++ b/include/crm/pengine/status_compat.h @@ -1,38 +1,56 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #ifndef PCMK__CRM_PENGINE_STATUS_COMPAT__H #define PCMK__CRM_PENGINE_STATUS_COMPAT__H #include // bool #include // pcmk_is_set() #include // pcmk_resource_t, pcmk__rsc_unique, etc. #ifdef __cplusplus extern "C" { #endif /** * \file * \brief Deprecated Pacemaker scheduler utilities * \ingroup pengine * \deprecated Do not include this header directly. The utilities in this * header, and the header itself, will be removed in a future * release. */ +// NOTE: sbd (as of at least 1.5.2) uses this +//! \deprecated Use pcmk_new_scheduler() instead +pcmk_scheduler_t *pe_new_working_set(void); + +// NOTE: sbd (as of at least 1.5.2) uses this +//! \deprecated Use pcmk_reset_scheduler() instead +void pe_reset_working_set(pcmk_scheduler_t *scheduler); + +//! \deprecated Use pcmk_reset_scheduler() instead +void cleanup_calculations(pcmk_scheduler_t *scheduler); + +//! \deprecated Use pcmk_reset_scheduler() instead +void set_working_set_defaults(pcmk_scheduler_t *scheduler); + +// NOTE: sbd (as of at least 1.5.2) uses this +//! \deprecated Use pcmk_free_scheduler() instead +void pe_free_working_set(pcmk_scheduler_t *scheduler); + // NOTE: sbd (as of at least 1.5.2) uses this //! \deprecated Use pcmk_find_node() with scheduler object instead pcmk_node_t *pe_find_node(const GList *node_list, const char *node_name); #ifdef __cplusplus } #endif #endif // PCMK__CRM_PENGINE_STATUS_COMPAT__H diff --git a/lib/common/Makefile.am b/lib/common/Makefile.am index 45fd0a0265..32fdf0602a 100644 --- a/lib/common/Makefile.am +++ b/lib/common/Makefile.am @@ -1,142 +1,144 @@ # # Copyright 2004-2024 the Pacemaker project contributors # # The version control history for this file may have further details. # # This source code is licensed under the GNU General Public License version 2 # or later (GPLv2+) WITHOUT ANY WARRANTY. # include $(top_srcdir)/mk/common.mk ## libraries lib_LTLIBRARIES = libcrmcommon.la check_LTLIBRARIES = libcrmcommon_test.la # Disable -Wcast-qual if used, because we do some hacky casting, # and because libxml2 has some signatures that should be const but aren't # for backward compatibility reasons. # s390 needs -fPIC # s390-suse-linux/bin/ld: .libs/ipc.o: relocation R_390_PC32DBL against `__stack_chk_fail@@GLIBC_2.4' can not be used when making a shared object; recompile with -fPIC CFLAGS = $(CFLAGS_COPY:-Wcast-qual=) -fPIC # Without "." here, check-recursive will run through the subdirectories first # and then run "make check" here. This will fail, because there's things in # the subdirectories that need check_LTLIBRARIES built first. Adding "." here # changes the order so the subdirectories are processed afterwards. SUBDIRS = . tests noinst_HEADERS = crmcommon_private.h \ mock_private.h libcrmcommon_la_LDFLAGS = -version-info 68:0:0 libcrmcommon_la_CFLAGS = $(CFLAGS_HARDENED_LIB) libcrmcommon_la_LDFLAGS += $(LDFLAGS_HARDENED_LIB) # If configured with --with-profiling or --with-coverage, BUILD_PROFILING will # be set and -fno-builtin will be added to the CFLAGS. However, libcrmcommon # uses the fabs() function which is normally supplied by gcc as one of its # builtins. Therefore we need to explicitly link against libm here or the # tests won't link. if BUILD_PROFILING libcrmcommon_la_LIBADD = -lm endif ## Library sources (*must* use += format for bumplibs) libcrmcommon_la_SOURCES = libcrmcommon_la_SOURCES += acl.c +libcrmcommon_la_SOURCES += action_relation.c libcrmcommon_la_SOURCES += actions.c libcrmcommon_la_SOURCES += agents.c libcrmcommon_la_SOURCES += alerts.c libcrmcommon_la_SOURCES += attrs.c libcrmcommon_la_SOURCES += cib.c if BUILD_CIBSECRETS libcrmcommon_la_SOURCES += cib_secrets.c endif libcrmcommon_la_SOURCES += cmdline.c libcrmcommon_la_SOURCES += digest.c libcrmcommon_la_SOURCES += health.c libcrmcommon_la_SOURCES += io.c libcrmcommon_la_SOURCES += ipc_attrd.c libcrmcommon_la_SOURCES += ipc_client.c libcrmcommon_la_SOURCES += ipc_common.c libcrmcommon_la_SOURCES += ipc_controld.c libcrmcommon_la_SOURCES += ipc_pacemakerd.c libcrmcommon_la_SOURCES += ipc_schedulerd.c libcrmcommon_la_SOURCES += ipc_server.c libcrmcommon_la_SOURCES += iso8601.c libcrmcommon_la_SOURCES += lists.c +libcrmcommon_la_SOURCES += location.c libcrmcommon_la_SOURCES += logging.c libcrmcommon_la_SOURCES += mainloop.c libcrmcommon_la_SOURCES += messages.c libcrmcommon_la_SOURCES += nodes.c libcrmcommon_la_SOURCES += nvpair.c libcrmcommon_la_SOURCES += options.c libcrmcommon_la_SOURCES += options_display.c libcrmcommon_la_SOURCES += output.c libcrmcommon_la_SOURCES += output_html.c libcrmcommon_la_SOURCES += output_log.c libcrmcommon_la_SOURCES += output_none.c libcrmcommon_la_SOURCES += output_text.c libcrmcommon_la_SOURCES += output_xml.c libcrmcommon_la_SOURCES += patchset.c libcrmcommon_la_SOURCES += patchset_display.c libcrmcommon_la_SOURCES += pid.c libcrmcommon_la_SOURCES += probes.c libcrmcommon_la_SOURCES += procfs.c libcrmcommon_la_SOURCES += remote.c libcrmcommon_la_SOURCES += resources.c libcrmcommon_la_SOURCES += results.c libcrmcommon_la_SOURCES += roles.c libcrmcommon_la_SOURCES += rules.c libcrmcommon_la_SOURCES += scheduler.c libcrmcommon_la_SOURCES += schemas.c libcrmcommon_la_SOURCES += scores.c libcrmcommon_la_SOURCES += servers.c libcrmcommon_la_SOURCES += strings.c libcrmcommon_la_SOURCES += tls.c libcrmcommon_la_SOURCES += utils.c libcrmcommon_la_SOURCES += watchdog.c libcrmcommon_la_SOURCES += xml.c libcrmcommon_la_SOURCES += xml_attr.c libcrmcommon_la_SOURCES += xml_comment.c libcrmcommon_la_SOURCES += xml_display.c libcrmcommon_la_SOURCES += xml_element.c libcrmcommon_la_SOURCES += xml_idref.c libcrmcommon_la_SOURCES += xml_io.c libcrmcommon_la_SOURCES += xpath.c # # libcrmcommon_test is used only with unit tests, so we can mock system calls. # See mock.c for details. # include $(top_srcdir)/mk/tap.mk libcrmcommon_test_la_SOURCES = $(libcrmcommon_la_SOURCES) libcrmcommon_test_la_SOURCES += mock.c libcrmcommon_test_la_SOURCES += unittest.c libcrmcommon_test_la_LDFLAGS = $(libcrmcommon_la_LDFLAGS) \ -rpath $(libdir) \ $(LDFLAGS_WRAP) # If GCC emits a builtin function in place of something we've mocked up, that will # get used instead of the mocked version which leads to unexpected test results. So # disable all builtins. Older versions of GCC (at least, on RHEL7) will still emit # replacement code for strdup (and possibly other functions) unless -fno-inline is # also added. libcrmcommon_test_la_CFLAGS = $(libcrmcommon_la_CFLAGS) \ -DPCMK__UNIT_TESTING \ -fno-builtin \ -fno-inline # If -fno-builtin is used, -lm also needs to be added. See the comment at # BUILD_PROFILING above. libcrmcommon_test_la_LIBADD = $(libcrmcommon_la_LIBADD) if BUILD_COVERAGE libcrmcommon_test_la_LIBADD += -lgcov endif libcrmcommon_test_la_LIBADD += -lcmocka libcrmcommon_test_la_LIBADD += -lm nodist_libcrmcommon_test_la_SOURCES = $(nodist_libcrmcommon_la_SOURCES) diff --git a/lib/common/action_relation.c b/lib/common/action_relation.c new file mode 100644 index 0000000000..22343cc551 --- /dev/null +++ b/lib/common/action_relation.c @@ -0,0 +1,31 @@ +/* + * Copyright 2024 the Pacemaker project contributors + * + * The version control history for this file may have further details. + * + * This source code is licensed under the GNU Lesser General Public License + * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. + */ + +#include + +#include // free() +#include // gpointer + +#include + +/*! + * \internal + * \brief Free an action relation + * + * \param[in,out] user_data Action relation to free + */ +void +pcmk__free_action_relation(gpointer user_data) +{ + pcmk__action_relation_t *relation = user_data; + + free(relation->task1); + free(relation->task2); + free(relation); +} diff --git a/lib/common/actions.c b/lib/common/actions.c index cbfe6ebe8d..61076f50b1 100644 --- a/lib/common/actions.c +++ b/lib/common/actions.c @@ -1,584 +1,614 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include #include #include /*! * \internal * \brief Get string equivalent of an action type * * \param[in] action Action type * * \return Static string describing \p action */ const char * pcmk__action_text(enum pcmk__action_type action) { switch (action) { case pcmk__action_stop: return PCMK_ACTION_STOP; case pcmk__action_stopped: return PCMK_ACTION_STOPPED; case pcmk__action_start: return PCMK_ACTION_START; case pcmk__action_started: return PCMK_ACTION_RUNNING; case pcmk__action_shutdown: return PCMK_ACTION_DO_SHUTDOWN; case pcmk__action_fence: return PCMK_ACTION_STONITH; case pcmk__action_monitor: return PCMK_ACTION_MONITOR; case pcmk__action_notify: return PCMK_ACTION_NOTIFY; case pcmk__action_notified: return PCMK_ACTION_NOTIFIED; case pcmk__action_promote: return PCMK_ACTION_PROMOTE; case pcmk__action_promoted: return PCMK_ACTION_PROMOTED; case pcmk__action_demote: return PCMK_ACTION_DEMOTE; case pcmk__action_demoted: return PCMK_ACTION_DEMOTED; default: // pcmk__action_unspecified or invalid return "no_action"; } } /*! * \internal * \brief Parse an action type from an action name * * \param[in] action_name Action name * * \return Action type corresponding to \p action_name */ enum pcmk__action_type pcmk__parse_action(const char *action_name) { if (pcmk__str_eq(action_name, PCMK_ACTION_STOP, pcmk__str_none)) { return pcmk__action_stop; } else if (pcmk__str_eq(action_name, PCMK_ACTION_STOPPED, pcmk__str_none)) { return pcmk__action_stopped; } else if (pcmk__str_eq(action_name, PCMK_ACTION_START, pcmk__str_none)) { return pcmk__action_start; } else if (pcmk__str_eq(action_name, PCMK_ACTION_RUNNING, pcmk__str_none)) { return pcmk__action_started; } else if (pcmk__str_eq(action_name, PCMK_ACTION_DO_SHUTDOWN, pcmk__str_none)) { return pcmk__action_shutdown; } else if (pcmk__str_eq(action_name, PCMK_ACTION_STONITH, pcmk__str_none)) { return pcmk__action_fence; } else if (pcmk__str_eq(action_name, PCMK_ACTION_MONITOR, pcmk__str_none)) { return pcmk__action_monitor; } else if (pcmk__str_eq(action_name, PCMK_ACTION_NOTIFY, pcmk__str_none)) { return pcmk__action_notify; } else if (pcmk__str_eq(action_name, PCMK_ACTION_NOTIFIED, pcmk__str_none)) { return pcmk__action_notified; } else if (pcmk__str_eq(action_name, PCMK_ACTION_PROMOTE, pcmk__str_none)) { return pcmk__action_promote; } else if (pcmk__str_eq(action_name, PCMK_ACTION_DEMOTE, pcmk__str_none)) { return pcmk__action_demote; } else if (pcmk__str_eq(action_name, PCMK_ACTION_PROMOTED, pcmk__str_none)) { return pcmk__action_promoted; } else if (pcmk__str_eq(action_name, PCMK_ACTION_DEMOTED, pcmk__str_none)) { return pcmk__action_demoted; } return pcmk__action_unspecified; } /*! * \internal * \brief Get string equivalent of a failure handling type * * \param[in] on_fail Failure handling type * * \return Static string describing \p on_fail */ const char * pcmk__on_fail_text(enum pcmk__on_fail on_fail) { switch (on_fail) { case pcmk__on_fail_ignore: return "ignore"; case pcmk__on_fail_demote: return "demote"; case pcmk__on_fail_block: return "block"; case pcmk__on_fail_restart: return "recover"; case pcmk__on_fail_ban: return "migrate"; case pcmk__on_fail_stop: return "stop"; case pcmk__on_fail_fence_node: return "fence"; case pcmk__on_fail_standby_node: return "standby"; case pcmk__on_fail_restart_container: return "restart-container"; case pcmk__on_fail_reset_remote: return "reset-remote"; } return ""; } +/*! + * \internal + * \brief Free an action object + * + * \param[in,out] user_data Action object to free + */ +void +pcmk__free_action(gpointer user_data) +{ + pcmk_action_t *action = user_data; + + if (action == NULL) { + return; + } + g_list_free_full(action->actions_before, free); + g_list_free_full(action->actions_after, free); + if (action->extra != NULL) { + g_hash_table_destroy(action->extra); + } + if (action->meta != NULL) { + g_hash_table_destroy(action->meta); + } + pcmk__free_node_copy(action->node); + free(action->cancel_task); + free(action->reason); + free(action->task); + free(action->uuid); + free(action); +} + /*! * \brief Generate an operation key (RESOURCE_ACTION_INTERVAL) * * \param[in] rsc_id ID of resource being operated on * \param[in] op_type Operation name * \param[in] interval_ms Operation interval * * \return Newly allocated memory containing operation key as string * * \note This function asserts on errors, so it will never return NULL. * The caller is responsible for freeing the result with free(). */ char * pcmk__op_key(const char *rsc_id, const char *op_type, guint interval_ms) { pcmk__assert((rsc_id != NULL) && (op_type != NULL)); return crm_strdup_printf(PCMK__OP_FMT, rsc_id, op_type, interval_ms); } static inline gboolean convert_interval(const char *s, guint *interval_ms) { unsigned long l; errno = 0; l = strtoul(s, NULL, 10); if (errno != 0) { return FALSE; } *interval_ms = (guint) l; return TRUE; } /*! * \internal * \brief Check for underbar-separated substring match * * \param[in] key Overall string being checked * \param[in] position Match before underbar at this \p key index * \param[in] matches Substrings to match (may contain underbars) * * \return \p key index of underbar before any matching substring, * or 0 if none */ static size_t match_before(const char *key, size_t position, const char **matches) { for (int i = 0; matches[i] != NULL; ++i) { const size_t match_len = strlen(matches[i]); // Must have at least X_MATCH before position if (position > (match_len + 1)) { const size_t possible = position - match_len - 1; if ((key[possible] == '_') && (strncmp(key + possible + 1, matches[i], match_len) == 0)) { return possible; } } } return 0; } gboolean parse_op_key(const char *key, char **rsc_id, char **op_type, guint *interval_ms) { guint local_interval_ms = 0; const size_t key_len = (key == NULL)? 0 : strlen(key); // Operation keys must be formatted as RSC_ACTION_INTERVAL size_t action_underbar = 0; // Index in key of underbar before ACTION size_t interval_underbar = 0; // Index in key of underbar before INTERVAL size_t possible = 0; /* Underbar was a poor choice of separator since both RSC and ACTION can * contain underbars. Here, list action names and name prefixes that can. */ const char *actions_with_underbars[] = { PCMK_ACTION_MIGRATE_FROM, PCMK_ACTION_MIGRATE_TO, NULL }; const char *action_prefixes_with_underbars[] = { "pre_" PCMK_ACTION_NOTIFY, "post_" PCMK_ACTION_NOTIFY, "confirmed-pre_" PCMK_ACTION_NOTIFY, "confirmed-post_" PCMK_ACTION_NOTIFY, NULL, }; // Initialize output variables in case of early return if (rsc_id) { *rsc_id = NULL; } if (op_type) { *op_type = NULL; } if (interval_ms) { *interval_ms = 0; } // RSC_ACTION_INTERVAL implies a minimum of 5 characters if (key_len < 5) { return FALSE; } // Find, parse, and validate interval interval_underbar = key_len - 2; while ((interval_underbar > 2) && (key[interval_underbar] != '_')) { --interval_underbar; } if ((interval_underbar == 2) || !convert_interval(key + interval_underbar + 1, &local_interval_ms)) { return FALSE; } // Find the base (OCF) action name, disregarding prefixes action_underbar = match_before(key, interval_underbar, actions_with_underbars); if (action_underbar == 0) { action_underbar = interval_underbar - 2; while ((action_underbar > 0) && (key[action_underbar] != '_')) { --action_underbar; } if (action_underbar == 0) { return FALSE; } } possible = match_before(key, action_underbar, action_prefixes_with_underbars); if (possible != 0) { action_underbar = possible; } // Set output variables if (rsc_id != NULL) { *rsc_id = strndup(key, action_underbar); pcmk__mem_assert(*rsc_id); } if (op_type != NULL) { *op_type = strndup(key + action_underbar + 1, interval_underbar - action_underbar - 1); pcmk__mem_assert(*op_type); } if (interval_ms != NULL) { *interval_ms = local_interval_ms; } return TRUE; } char * pcmk__notify_key(const char *rsc_id, const char *notify_type, const char *op_type) { CRM_CHECK(rsc_id != NULL, return NULL); CRM_CHECK(op_type != NULL, return NULL); CRM_CHECK(notify_type != NULL, return NULL); return crm_strdup_printf("%s_%s_notify_%s_0", rsc_id, notify_type, op_type); } /*! * \brief Parse a transition magic string into its constituent parts * * \param[in] magic Magic string to parse (must be non-NULL) * \param[out] uuid If non-NULL, where to store copy of parsed UUID * \param[out] transition_id If non-NULL, where to store parsed transition ID * \param[out] action_id If non-NULL, where to store parsed action ID * \param[out] op_status If non-NULL, where to store parsed result status * \param[out] op_rc If non-NULL, where to store parsed actual rc * \param[out] target_rc If non-NULL, where to stored parsed target rc * * \return TRUE if key was valid, FALSE otherwise * \note If uuid is supplied and this returns TRUE, the caller is responsible * for freeing the memory for *uuid using free(). */ gboolean decode_transition_magic(const char *magic, char **uuid, int *transition_id, int *action_id, int *op_status, int *op_rc, int *target_rc) { int res = 0; char *key = NULL; gboolean result = TRUE; int local_op_status = -1; int local_op_rc = -1; CRM_CHECK(magic != NULL, return FALSE); #ifdef HAVE_SSCANF_M res = sscanf(magic, "%d:%d;%ms", &local_op_status, &local_op_rc, &key); #else // magic must have >=4 other characters key = pcmk__assert_alloc(1, strlen(magic) - 3); res = sscanf(magic, "%d:%d;%s", &local_op_status, &local_op_rc, key); #endif if (res == EOF) { crm_err("Could not decode transition information '%s': %s", magic, pcmk_rc_str(errno)); result = FALSE; } else if (res < 3) { crm_warn("Transition information '%s' incomplete (%d of 3 expected items)", magic, res); result = FALSE; } else { if (op_status) { *op_status = local_op_status; } if (op_rc) { *op_rc = local_op_rc; } result = decode_transition_key(key, uuid, transition_id, action_id, target_rc); } free(key); return result; } char * pcmk__transition_key(int transition_id, int action_id, int target_rc, const char *node) { CRM_CHECK(node != NULL, return NULL); return crm_strdup_printf("%d:%d:%d:%-*s", action_id, transition_id, target_rc, 36, node); } /*! * \brief Parse a transition key into its constituent parts * * \param[in] key Transition key to parse (must be non-NULL) * \param[out] uuid If non-NULL, where to store copy of parsed UUID * \param[out] transition_id If non-NULL, where to store parsed transition ID * \param[out] action_id If non-NULL, where to store parsed action ID * \param[out] target_rc If non-NULL, where to stored parsed target rc * * \return TRUE if key was valid, FALSE otherwise * \note If uuid is supplied and this returns TRUE, the caller is responsible * for freeing the memory for *uuid using free(). */ gboolean decode_transition_key(const char *key, char **uuid, int *transition_id, int *action_id, int *target_rc) { int local_transition_id = -1; int local_action_id = -1; int local_target_rc = -1; char local_uuid[37] = { '\0' }; // Initialize any supplied output arguments if (uuid) { *uuid = NULL; } if (transition_id) { *transition_id = -1; } if (action_id) { *action_id = -1; } if (target_rc) { *target_rc = -1; } CRM_CHECK(key != NULL, return FALSE); if (sscanf(key, "%d:%d:%d:%36s", &local_action_id, &local_transition_id, &local_target_rc, local_uuid) != 4) { crm_err("Invalid transition key '%s'", key); return FALSE; } if (strlen(local_uuid) != 36) { crm_warn("Invalid UUID '%s' in transition key '%s'", local_uuid, key); } if (uuid) { *uuid = pcmk__str_copy(local_uuid); } if (transition_id) { *transition_id = local_transition_id; } if (action_id) { *action_id = local_action_id; } if (target_rc) { *target_rc = local_target_rc; } return TRUE; } int rsc_op_expected_rc(const lrmd_event_data_t *op) { int rc = 0; if (op && op->user_data) { decode_transition_key(op->user_data, NULL, NULL, NULL, &rc); } return rc; } gboolean did_rsc_op_fail(lrmd_event_data_t * op, int target_rc) { switch (op->op_status) { case PCMK_EXEC_CANCELLED: case PCMK_EXEC_PENDING: return FALSE; case PCMK_EXEC_NOT_SUPPORTED: case PCMK_EXEC_TIMEOUT: case PCMK_EXEC_ERROR: case PCMK_EXEC_NOT_CONNECTED: case PCMK_EXEC_NO_FENCE_DEVICE: case PCMK_EXEC_NO_SECRETS: case PCMK_EXEC_INVALID: return TRUE; default: if (target_rc != op->rc) { return TRUE; } } return FALSE; } /*! * \brief Create a CIB XML element for an operation * * \param[in,out] parent If not NULL, make new XML node a child of this * \param[in] prefix Generate an ID using this prefix * \param[in] task Operation task to set * \param[in] interval_spec Operation interval to set * \param[in] timeout If not NULL, operation timeout to set * * \return New XML object on success, NULL otherwise */ xmlNode * crm_create_op_xml(xmlNode *parent, const char *prefix, const char *task, const char *interval_spec, const char *timeout) { xmlNode *xml_op; CRM_CHECK(prefix && task && interval_spec, return NULL); xml_op = pcmk__xe_create(parent, PCMK_XE_OP); pcmk__xe_set_id(xml_op, "%s-%s-%s", prefix, task, interval_spec); crm_xml_add(xml_op, PCMK_META_INTERVAL, interval_spec); crm_xml_add(xml_op, PCMK_XA_NAME, task); if (timeout) { crm_xml_add(xml_op, PCMK_META_TIMEOUT, timeout); } return xml_op; } /*! * \brief Check whether an operation requires resource agent meta-data * * \param[in] rsc_class Resource agent class (or NULL to skip class check) * \param[in] op Operation action (or NULL to skip op check) * * \return true if operation needs meta-data, false otherwise * \note At least one of rsc_class and op must be specified. */ bool crm_op_needs_metadata(const char *rsc_class, const char *op) { /* Agent metadata is used to determine whether an agent reload is possible, * so if this op is not relevant to that feature, we don't need metadata. */ CRM_CHECK((rsc_class != NULL) || (op != NULL), return false); if ((rsc_class != NULL) && !pcmk_is_set(pcmk_get_ra_caps(rsc_class), pcmk_ra_cap_params)) { // Metadata is needed only for resource classes that use parameters return false; } if (op == NULL) { return true; } // Metadata is needed only for these actions return pcmk__str_any_of(op, PCMK_ACTION_START, PCMK_ACTION_MONITOR, PCMK_ACTION_PROMOTE, PCMK_ACTION_DEMOTE, PCMK_ACTION_RELOAD, PCMK_ACTION_RELOAD_AGENT, PCMK_ACTION_MIGRATE_TO, PCMK_ACTION_MIGRATE_FROM, PCMK_ACTION_NOTIFY, NULL); } /*! * \internal * \brief Check whether an action name is for a fencing action * * \param[in] action Action name to check * * \return \c true if \p action is \c PCMK_ACTION_OFF or \c PCMK_ACTION_REBOOT, * or \c false otherwise */ bool pcmk__is_fencing_action(const char *action) { return pcmk__str_any_of(action, PCMK_ACTION_OFF, PCMK_ACTION_REBOOT, NULL); } diff --git a/lib/common/location.c b/lib/common/location.c new file mode 100644 index 0000000000..e0f7707024 --- /dev/null +++ b/lib/common/location.c @@ -0,0 +1,31 @@ +/* + * Copyright 2024 the Pacemaker project contributors + * + * The version control history for this file may have further details. + * + * This source code is licensed under the GNU Lesser General Public License + * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. + */ + +#include + +#include // free() +#include // gpointer, g_list_free_full() + +#include + +/*! + * \internal + * \brief Free a location constraint + * + * \param[in,out] user_data Location constraint to free + */ +void +pcmk__free_location(gpointer user_data) +{ + pcmk__location_t *location = user_data; + + g_list_free_full(location->nodes, pcmk__free_node_copy); + free(location->id); + free(location); +} diff --git a/lib/common/nodes.c b/lib/common/nodes.c index fed3159464..4a059e6520 100644 --- a/lib/common/nodes.c +++ b/lib/common/nodes.c @@ -1,212 +1,254 @@ /* * Copyright 2022-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include // xmlNode #include +/*! + * \internal + * \brief Free a node object + * + * \param[in,out] user_data Node object to free + */ +void +pcmk__free_node(gpointer user_data) +{ + pcmk_node_t *node = user_data; + + if (node == NULL) { + return; + } + if (node->details == NULL) { + free(node); + return; + } + + /* This may be called after freeing resources, which means that we can't + * use node->private->name for Pacemaker Remote nodes. + */ + crm_trace("Freeing node %s", (pcmk__is_pacemaker_remote_node(node)? + "(guest or remote)" : pcmk__node_name(node))); + + if (node->priv->attrs != NULL) { + g_hash_table_destroy(node->priv->attrs); + } + if (node->priv->utilization != NULL) { + g_hash_table_destroy(node->priv->utilization); + } + if (node->priv->digest_cache != NULL) { + g_hash_table_destroy(node->priv->digest_cache); + } + g_list_free(node->details->running_rsc); + g_list_free(node->priv->assigned_resources); + free(node->priv); + free(node->details); + free(node->assign); + free(node); +} + /*! * \internal * \brief Free a copy of a node object * * \param[in] data Node copy (created by pe__copy_node()) to free */ void pcmk__free_node_copy(void *data) { if (data != NULL) { pcmk_node_t *node = data; if (node->assign != NULL) { // This is the only member allocated separately for a node copy free(node->assign); } free(node); } } /*! * \internal * \brief Check whether a node is online * * \param[in] node Node to check * * \return true if \p node is online, otherwise false */ bool pcmk_node_is_online(const pcmk_node_t *node) { return (node != NULL) && node->details->online; } /*! * \internal * \brief Check whether a node is pending * * Check whether a node is pending. A node is pending if it is a member of the * cluster but not the controller group, which means it is in the process of * either joining or leaving the cluster. * * \param[in] node Node to check * * \return true if \p node is pending, otherwise false */ bool pcmk_node_is_pending(const pcmk_node_t *node) { return (node != NULL) && node->details->pending; } /*! * \internal * \brief Check whether a node is clean * * Check whether a node is clean. A node is clean if it is a cluster node or * remote node that has been seen by the cluster at least once, or the * startup-fencing cluster option is false; and the node, and its host if a * guest or bundle node, are not scheduled to be fenced. * * \param[in] node Node to check * * \return true if \p node is clean, otherwise false */ bool pcmk_node_is_clean(const pcmk_node_t *node) { return (node != NULL) && !(node->details->unclean); } /*! * \internal * \brief Check whether a node is shutting down * * \param[in] node Node to check * * \return true if \p node is shutting down, otherwise false */ bool pcmk_node_is_shutting_down(const pcmk_node_t *node) { return (node != NULL) && node->details->shutdown; } /*! * \internal * \brief Check whether a node is in maintenance mode * * \param[in] node Node to check * * \return true if \p node is in maintenance mode, otherwise false */ bool pcmk_node_is_in_maintenance(const pcmk_node_t *node) { return (node != NULL) && node->details->maintenance; } /*! * \internal * \brief Call a function for each resource active on a node * * Call a caller-supplied function with a caller-supplied argument for each * resource that is active on a given node. If the function returns false, this * function will return immediately without processing any remaining resources. * * \param[in] node Node to check * * \return Result of last call of \p fn (or false if none) */ bool pcmk_foreach_active_resource(pcmk_node_t *node, bool (*fn)(pcmk_resource_t *, void *), void *user_data) { bool result = false; if ((node != NULL) && (fn != NULL)) { for (GList *item = node->details->running_rsc; item != NULL; item = item->next) { result = fn((pcmk_resource_t *) item->data, user_data); if (!result) { break; } } } return result; } void pcmk__xe_add_node(xmlNode *xml, const char *node, int nodeid) { pcmk__assert(xml != NULL); if (node != NULL) { crm_xml_add(xml, PCMK__XA_ATTR_HOST, node); } if (nodeid > 0) { crm_xml_add_int(xml, PCMK__XA_ATTR_HOST_ID, nodeid); } } /*! * \internal * \brief Find a node by name in a list of nodes * * \param[in] nodes List of nodes (as pcmk_node_t*) * \param[in] node_name Name of node to find * * \return Node from \p nodes that matches \p node_name if any, otherwise NULL */ pcmk_node_t * pcmk__find_node_in_list(const GList *nodes, const char *node_name) { if (node_name != NULL) { for (const GList *iter = nodes; iter != NULL; iter = iter->next) { pcmk_node_t *node = (pcmk_node_t *) iter->data; if (pcmk__str_eq(node->priv->name, node_name, pcmk__str_casei)) { return node; } } } return NULL; } #define XP_SHUTDOWN "//" PCMK__XE_NODE_STATE "[@" PCMK_XA_UNAME "='%s']/" \ PCMK__XE_TRANSIENT_ATTRIBUTES "/" PCMK_XE_INSTANCE_ATTRIBUTES "/" \ PCMK_XE_NVPAIR "[@" PCMK_XA_NAME "='" PCMK__NODE_ATTR_SHUTDOWN "']" /*! * \brief Get value of a node's shutdown attribute from CIB, if present * * \param[in] cib CIB to check * \param[in] node Name of node to check * * \return Value of shutdown attribute for \p node in \p cib if any, * otherwise NULL * \note The return value is a pointer into \p cib and so is valid only for the * lifetime of that object. */ const char * pcmk_cib_node_shutdown(xmlNode *cib, const char *node) { if ((cib != NULL) && (node != NULL)) { char *xpath = crm_strdup_printf(XP_SHUTDOWN, node); xmlNode *match = get_xpath_object(xpath, cib, LOG_TRACE); free(xpath); if (match != NULL) { return crm_element_value(match, PCMK_XA_VALUE); } } return NULL; } diff --git a/lib/common/resources.c b/lib/common/resources.c index bfc3ddfc5b..18890a99af 100644 --- a/lib/common/resources.c +++ b/lib/common/resources.c @@ -1,67 +1,83 @@ /* * Copyright 2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include // NULL #include // bool, false #include #include +/*! + * \internal + * \brief Free a resource object + * + * \param[in,out] user_data Resource object to free + */ +void +pcmk__free_resource(gpointer user_data) +{ + pcmk_resource_t *rsc = user_data; + + if (rsc != NULL) { + rsc->priv->fns->free(rsc); + } +} + /*! * \internal * \brief Get a resource's ID * * \param[in] rsc Resource to check * * \return ID of \p rsc (or NULL if \p rsc is NULL) */ const char * pcmk_resource_id(const pcmk_resource_t *rsc) { return (rsc == NULL)? NULL : rsc->id; } /*! * \internal * \brief Check whether a resource is managed by the cluster * * \param[in] rsc Resource to check * * \return true if \p rsc is managed, otherwise false */ bool pcmk_resource_is_managed(const pcmk_resource_t *rsc) { return (rsc == NULL)? false : pcmk_is_set(rsc->flags, pcmk__rsc_managed); } /*! * \brief Get readable description of a multiply-active recovery type * * \param[in] rsc Resource with recovery type to check * * \return Static string describing recovery type of \p rsc */ const char * pcmk__multiply_active_text(const pcmk_resource_t *rsc) { switch (rsc->priv->multiply_active_policy) { case pcmk__multiply_active_stop: return "shutting it down"; case pcmk__multiply_active_restart: return "attempting recovery"; case pcmk__multiply_active_block: return "waiting for an administrator"; case pcmk__multiply_active_unexpected: return "stopping unexpected instances"; } return "Unknown"; } diff --git a/lib/common/scheduler.c b/lib/common/scheduler.c index 0e73da7ba3..78413e7da7 100644 --- a/lib/common/scheduler.c +++ b/lib/common/scheduler.c @@ -1,106 +1,417 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include // uint32_t #include // EINVAL -#include // gboolean, FALSE +#include // gboolean, FALSE, etc. #include // xmlNode #include uint32_t pcmk__warnings = 0; +/*! + * \brief Create a new object to hold scheduler data + * + * \return New, initialized scheduler data, or NULL on memory error + * \note Only pcmk_scheduler_t objects created with this function (as opposed + * to statically declared or directly allocated) should be used with the + * functions in this library, to allow for future extensions to the + * data type. The caller is responsible for freeing the memory with + * pcmk_free_scheduler() when the instance is no longer needed. + */ +pcmk_scheduler_t * +pcmk_new_scheduler(void) +{ + pcmk_scheduler_t *scheduler = calloc(1, sizeof(pcmk_scheduler_t)); + + if (scheduler == NULL) { + return NULL; + } + scheduler->priv = calloc(1, sizeof(pcmk__scheduler_private_t)); + if (scheduler->priv == NULL) { + free(scheduler); + return NULL; + } + pcmk__set_scheduler_defaults(scheduler); + return scheduler; +} + +/*! + * \internal + * \brief Set non-zero default values in scheduler data + * + * \param[in,out] scheduler Scheduler data to modify + * + * \note Values that default to NULL or 0 will remain unchanged + */ +void +pcmk__set_scheduler_defaults(pcmk_scheduler_t *scheduler) +{ + pcmk__assert(scheduler != NULL); + scheduler->flags = 0U; +#if PCMK__CONCURRENT_FENCING_DEFAULT_TRUE + pcmk__set_scheduler_flags(scheduler, + pcmk__sched_symmetric_cluster + |pcmk__sched_concurrent_fencing + |pcmk__sched_stop_removed_resources + |pcmk__sched_cancel_removed_actions); +#else + pcmk__set_scheduler_flags(scheduler, + pcmk__sched_symmetric_cluster + |pcmk__sched_stop_removed_resources + |pcmk__sched_cancel_removed_actions); +#endif + scheduler->no_quorum_policy = pcmk_no_quorum_stop; + scheduler->priv->next_action_id = 1; + scheduler->priv->next_ordering_id = 1; +} + +/*! + * \brief Reset scheduler data to defaults + * + * Free scheduler data except the local node name and output object, and reset + * all other values to defaults, so the data is suitable for rerunning status + * + * \param[in,out] scheduler Scheduler data to reset + */ +void +pcmk_reset_scheduler(pcmk_scheduler_t *scheduler) +{ + CRM_CHECK(scheduler != NULL, return); + + /* Be careful about the order of freeing members. Many contain references to + * other members that will become dangling if those members are freed first. + * For example, the node name and ID of Pacemaker Remote nodes are pointers + * into resource objects. Ensure that earlier-freed members are not needed + * by any of the free functions for later-freed members. + */ + + pcmk__xml_free(scheduler->input); + scheduler->input = NULL; + + scheduler->dc_node = NULL; + + g_list_free_full(scheduler->nodes, pcmk__free_node); + scheduler->nodes = NULL; + + // Do not reset local_node_name or out + + crm_time_free(scheduler->priv->now); + scheduler->priv->now = NULL; + + if (scheduler->priv->options != NULL) { + g_hash_table_destroy(scheduler->priv->options); + scheduler->priv->options = NULL; + } + + scheduler->priv->fence_action = NULL; + scheduler->priv->fence_timeout_ms = 0U; + scheduler->priv->priority_fencing_ms = 0U; + scheduler->priv->shutdown_lock_ms = 0U; + scheduler->priv->node_pending_ms = 0U; + scheduler->priv->placement_strategy = NULL; + scheduler->priv->rsc_defaults = NULL; + scheduler->priv->op_defaults = NULL; + + g_list_free_full(scheduler->priv->resources, pcmk__free_resource); + scheduler->priv->resources = NULL; + + if (scheduler->priv->templates != NULL) { + g_hash_table_destroy(scheduler->priv->templates); + scheduler->priv->templates = NULL; + } + if (scheduler->priv->tags != NULL) { + g_hash_table_destroy(scheduler->priv->tags); + scheduler->priv->tags = NULL; + } + + g_list_free_full(scheduler->priv->actions, pcmk__free_action); + scheduler->priv->actions = NULL; + + if (scheduler->priv->singletons != NULL) { + g_hash_table_destroy(scheduler->priv->singletons); + scheduler->priv->singletons = NULL; + } + + pcmk__xml_free(scheduler->priv->failed); + scheduler->priv->failed = NULL; + + pcmk__free_param_checks(scheduler); + + g_list_free(scheduler->priv->stop_needed); + scheduler->priv->stop_needed = NULL; + + g_list_free_full(scheduler->priv->location_constraints, + pcmk__free_location); + scheduler->priv->location_constraints = NULL; + + g_list_free_full(scheduler->priv->colocation_constraints, free); + scheduler->priv->colocation_constraints = NULL; + + g_list_free_full(scheduler->priv->ordering_constraints, + pcmk__free_action_relation); + scheduler->priv->ordering_constraints = NULL; + + if (scheduler->priv->ticket_constraints != NULL) { + g_hash_table_destroy(scheduler->priv->ticket_constraints); + scheduler->priv->ticket_constraints = NULL; + } + + scheduler->priv->ninstances = 0; + scheduler->priv->blocked_resources = 0; + scheduler->priv->disabled_resources = 0; + scheduler->priv->recheck_by = 0; + + pcmk__xml_free(scheduler->priv->graph); + scheduler->priv->graph = NULL; + + scheduler->priv->synapse_count = 0; + + pcmk__set_scheduler_defaults(scheduler); + + /* @TODO We should probably reset pcmk__config_has_warning and + * pcmk__config_has_error as well + */ +} + +/*! + * \brief Free scheduler data + * + * \param[in,out] scheduler Scheduler data to free + */ +void +pcmk_free_scheduler(pcmk_scheduler_t *scheduler) +{ + if (scheduler != NULL) { + pcmk_reset_scheduler(scheduler); + free(scheduler->priv->local_node_name); + free(scheduler->priv); + free(scheduler); + } +} + /*! * \internal * \brief Get the Designated Controller node from scheduler data * * \param[in] scheduler Scheduler data * * \return Designated Controller node from scheduler data, or NULL if none */ pcmk_node_t * pcmk_get_dc(const pcmk_scheduler_t *scheduler) { return (scheduler == NULL)? NULL : scheduler->dc_node; } /*! * \internal * \brief Get the no quorum policy from scheduler data * * \param[in] scheduler Scheduler data * * \return No quorum policy from scheduler data */ enum pe_quorum_policy pcmk_get_no_quorum_policy(const pcmk_scheduler_t *scheduler) { if (scheduler == NULL) { return pcmk_no_quorum_stop; // The default } return scheduler->no_quorum_policy; } /*! * \internal * \brief Set CIB XML as scheduler input in scheduler data * * \param[out] scheduler Scheduler data * \param[in] cib CIB XML to set as scheduler input * * \return Standard Pacemaker return code (EINVAL if \p scheduler is NULL, * otherwise pcmk_rc_ok) * \note This will not free any previously set scheduler CIB. */ int pcmk_set_scheduler_cib(pcmk_scheduler_t *scheduler, xmlNode *cib) { if (scheduler == NULL) { return EINVAL; } scheduler->input = cib; return pcmk_rc_ok; } /*! * \internal * \brief Check whether cluster has quorum * * \param[in] scheduler Scheduler data * * \return true if cluster has quorum, otherwise false */ bool pcmk_has_quorum(const pcmk_scheduler_t *scheduler) { if (scheduler == NULL) { return false; } return pcmk_is_set(scheduler->flags, pcmk__sched_quorate); } /*! * \brief Find a node by name in scheduler data * * \param[in] scheduler Scheduler data * \param[in] node_name Name of node to find * * \return Node from scheduler data that matches \p node_name if any, * otherwise NULL */ pcmk_node_t * pcmk_find_node(const pcmk_scheduler_t *scheduler, const char *node_name) { if ((scheduler == NULL) || (node_name == NULL)) { return NULL; } return pcmk__find_node_in_list(scheduler->nodes, node_name); } + +/*! + * \internal + * \brief Get scheduler data's "now" in epoch time + * + * \param[in,out] scheduler Scheduler data + * + * \return Scheduler data's "now" as seconds since epoch (defaulting to current + * time) + */ +time_t +pcmk__scheduler_epoch_time(pcmk_scheduler_t *scheduler) +{ + if (scheduler == NULL) { + return time(NULL); + } + if (scheduler->priv->now == NULL) { + crm_trace("Scheduler 'now' set to current time"); + scheduler->priv->now = crm_time_new(NULL); + } + return crm_time_get_seconds_since_epoch(scheduler->priv->now); +} + +/*! + * \internal + * \brief Update "recheck by" time in scheduler data + * + * \param[in] recheck Epoch time when recheck should happen + * \param[in,out] scheduler Scheduler data + * \param[in] reason What time is being updated for (for logs) + */ +void +pcmk__update_recheck_time(time_t recheck, pcmk_scheduler_t *scheduler, + const char *reason) +{ + pcmk__assert(scheduler != NULL); + + if ((recheck > pcmk__scheduler_epoch_time(scheduler)) + && ((scheduler->priv->recheck_by == 0) + || (scheduler->priv->recheck_by > recheck))) { + scheduler->priv->recheck_by = recheck; + crm_debug("Updated next scheduler recheck to %s for %s", + pcmk__trim(ctime(&recheck)), + pcmk__s(reason, "some reason")); + } +} + +/* Fail count clearing for parameter changes normally happens when unpacking + * history, before resources are unpacked. However, for bundles using the + * REMOTE_CONTAINER_HACK, we can't check the conditions until after unpacking + * the bundle, so those parameter checks are deferred using the APIs below. + */ + +// History entry to be checked later for fail count clearing +struct param_check { + const xmlNode *rsc_history; // History entry XML + pcmk_resource_t *rsc; // Resource corresponding to history entry + pcmk_node_t *node; // Node corresponding to history entry + enum pcmk__check_parameters check_type; // What needs checking +}; + +/*! + * \internal + * \brief Add a deferred parameter check + * + * \param[in] rsc_history Resource history XML to check later + * \param[in] rsc Resource that history is for + * \param[in] node Node that history is for + * \param[in] flag What needs to be checked later + * \param[in,out] scheduler Scheduler data + */ +void +pcmk__add_param_check(const xmlNode *rsc_history, pcmk_resource_t *rsc, + pcmk_node_t *node, enum pcmk__check_parameters flag, + pcmk_scheduler_t *scheduler) +{ + struct param_check *param_check = NULL; + + CRM_CHECK((rsc_history != NULL) && (rsc != NULL) && (node != NULL) + && (scheduler != NULL), return); + + crm_trace("Deferring checks of %s until after assignment", + pcmk__xe_id(rsc_history)); + param_check = pcmk__assert_alloc(1, sizeof(struct param_check)); + param_check->rsc_history = rsc_history; + param_check->rsc = rsc; + param_check->node = node; + param_check->check_type = flag; + scheduler->priv->param_check = g_list_prepend(scheduler->priv->param_check, + param_check); +} + +/*! + * \internal + * \brief Call a function for each deferred parameter check + * + * \param[in,out] scheduler Scheduler data + * \param[in] cb Function to be called + */ +void +pcmk__foreach_param_check(pcmk_scheduler_t *scheduler, + void (*cb)(pcmk_resource_t*, pcmk_node_t*, + const xmlNode*, + enum pcmk__check_parameters)) +{ + CRM_CHECK((scheduler != NULL) && (cb != NULL), return); + + for (GList *item = scheduler->priv->param_check; + item != NULL; item = item->next) { + struct param_check *param_check = item->data; + + cb(param_check->rsc, param_check->node, param_check->rsc_history, + param_check->check_type); + } +} + +/*! + * \internal + * \brief Free all deferred parameter checks + * + * \param[in,out] scheduler Scheduler data + */ +void +pcmk__free_param_checks(pcmk_scheduler_t *scheduler) +{ + if ((scheduler != NULL) && (scheduler->priv->param_check != NULL)) { + g_list_free_full(scheduler->priv->param_check, free); + scheduler->priv->param_check = NULL; + } +} diff --git a/lib/common/tests/scheduler/Makefile.am b/lib/common/tests/scheduler/Makefile.am index 51bde6235e..abb30df85d 100644 --- a/lib/common/tests/scheduler/Makefile.am +++ b/lib/common/tests/scheduler/Makefile.am @@ -1,20 +1,23 @@ # # Copyright 2024 the Pacemaker project contributors # # The version control history for this file may have further details. # # This source code is licensed under the GNU General Public License version 2 # or later (GPLv2+) WITHOUT ANY WARRANTY. # include $(top_srcdir)/mk/common.mk include $(top_srcdir)/mk/tap.mk include $(top_srcdir)/mk/unittest.mk # Add "_test" to the end of all test program names to simplify .gitignore. -check_PROGRAMS = pcmk_get_dc_test \ +check_PROGRAMS = pcmk__set_scheduler_defaults_test \ + pcmk__update_recheck_time_test \ + pcmk_get_dc_test \ pcmk_get_no_quorum_policy_test \ pcmk_has_quorum_test \ + pcmk_new_scheduler_test \ pcmk_set_scheduler_cib_test TESTS = $(check_PROGRAMS) diff --git a/lib/pengine/tests/status/set_working_set_defaults_test.c b/lib/common/tests/scheduler/pcmk__set_scheduler_defaults_test.c similarity index 76% rename from lib/pengine/tests/status/set_working_set_defaults_test.c rename to lib/common/tests/scheduler/pcmk__set_scheduler_defaults_test.c index 9743e1c759..788d04bcf1 100644 --- a/lib/pengine/tests/status/set_working_set_defaults_test.c +++ b/lib/common/tests/scheduler/pcmk__set_scheduler_defaults_test.c @@ -1,50 +1,55 @@ /* * Copyright 2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include "mock_private.h" static void -check_defaults(void **state) { - uint32_t flags; - pcmk_scheduler_t *scheduler = pcmk__assert_alloc(1, - sizeof(pcmk_scheduler_t)); +null_scheduler(void **state) +{ + pcmk__assert_asserts(pcmk__set_scheduler_defaults(NULL)); +} + +static void +check_defaults(void **state) +{ + uint32_t flags = 0U; + pcmk_scheduler_t *scheduler = NULL; + scheduler = pcmk__assert_alloc(1, sizeof(pcmk_scheduler_t)); scheduler->priv = pcmk__assert_alloc(1, sizeof(pcmk__scheduler_private_t)); - set_working_set_defaults(scheduler); + pcmk__set_scheduler_defaults(scheduler); flags = pcmk__sched_symmetric_cluster #if PCMK__CONCURRENT_FENCING_DEFAULT_TRUE |pcmk__sched_concurrent_fencing #endif |pcmk__sched_stop_removed_resources |pcmk__sched_cancel_removed_actions; assert_null(scheduler->priv->out); assert_int_equal(scheduler->priv->next_ordering_id, 1); assert_int_equal(scheduler->priv->next_action_id, 1); assert_int_equal(scheduler->no_quorum_policy, pcmk_no_quorum_stop); assert_int_equal(scheduler->flags, flags); - /* Avoid calling pe_free_working_set here so we don't artificially - * inflate the coverage numbers. - */ free(scheduler->priv); free(scheduler); } PCMK__UNIT_TEST(NULL, NULL, + cmocka_unit_test(null_scheduler), cmocka_unit_test(check_defaults)) diff --git a/lib/common/tests/scheduler/pcmk__update_recheck_time_test.c b/lib/common/tests/scheduler/pcmk__update_recheck_time_test.c new file mode 100644 index 0000000000..7344891645 --- /dev/null +++ b/lib/common/tests/scheduler/pcmk__update_recheck_time_test.c @@ -0,0 +1,79 @@ +/* + * Copyright 2024 the Pacemaker project contributors + * + * The version control history for this file may have further details. + * + * This source code is licensed under the GNU General Public License version 2 + * or later (GPLv2+) WITHOUT ANY WARRANTY. + */ + +#include + +#include // NULL +#include // time_t + +#include +#include +#include + +static void +assert_recheck(time_t now_time, time_t orig_time, time_t update_time, + time_t expected_time, const char *reason) +{ + pcmk_scheduler_t *scheduler = pcmk_new_scheduler(); + + scheduler->priv->now = pcmk__copy_timet(now_time); + scheduler->priv->recheck_by = orig_time; + pcmk__update_recheck_time(update_time, scheduler, reason); + assert_int_equal(scheduler->priv->recheck_by, expected_time); + pcmk_free_scheduler(scheduler); +} + +// A NULL scheduler argument is invalid and should assert +static void +null_scheduler(void **state) +{ + pcmk__assert_asserts(pcmk__update_recheck_time(0, NULL, "reasons")); +} + +// Do not update recheck time if new value is before or equal to "now" +static void +too_early(void **state) +{ + // Recheck time is initially unset + assert_recheck(1423548000, 0, 1423547900, 0, NULL); + assert_recheck(1423548000, 0, 1423548000, 0, NULL); + + // Recheck time is initially set + assert_recheck(1423548000, 1423548100, 1423547900, 1423548100, NULL); + assert_recheck(1423548000, 1423548100, 1423548000, 1423548100, NULL); +} + +// Update recheck time if the existing value is 0 +static void +first_time(void **state) +{ + // This also tests that a NULL reason does not crash + assert_recheck(1423548000, 0, 1423548100, 1423548100, NULL); +} + +// Update recheck time if new value is earlier than the existing one +static void +earlier_time(void **state) +{ + assert_recheck(1423548000, 1423548500, 1423548200, 1423548200, "reasons"); +} + +// Do not update recheck time if new value is later than the existing one +static void +later_time(void **state) +{ + assert_recheck(1423548000, 1423548500, 1423548600, 1423548500, "reasons"); +} + +PCMK__UNIT_TEST(NULL, NULL, + cmocka_unit_test(null_scheduler), + cmocka_unit_test(too_early), + cmocka_unit_test(first_time), + cmocka_unit_test(earlier_time), + cmocka_unit_test(later_time)) diff --git a/lib/pengine/tests/status/pe_new_working_set_test.c b/lib/common/tests/scheduler/pcmk_new_scheduler_test.c similarity index 60% rename from lib/pengine/tests/status/pe_new_working_set_test.c rename to lib/common/tests/scheduler/pcmk_new_scheduler_test.c index b385f9cd11..e7545df762 100644 --- a/lib/pengine/tests/status/pe_new_working_set_test.c +++ b/lib/common/tests/scheduler/pcmk_new_scheduler_test.c @@ -1,46 +1,47 @@ /* - * Copyright 2022-2023 the Pacemaker project contributors + * Copyright 2022-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include "mock_private.h" static void -calloc_fails(void **state) { +calloc_fails(void **state) +{ pcmk__mock_calloc = true; // calloc() will return NULL expect_value(__wrap_calloc, nmemb, 1); expect_value(__wrap_calloc, size, sizeof(pcmk_scheduler_t)); - assert_null(pe_new_working_set()); + assert_null(pcmk_new_scheduler()); pcmk__mock_calloc = false; // Use real calloc() } static void -calloc_succeeds(void **state) { - pcmk_scheduler_t *scheduler = pe_new_working_set(); +calloc_succeeds(void **state) +{ + pcmk_scheduler_t *scheduler = pcmk_new_scheduler(); - /* Nothing else to test about this function, as all it does is call - * set_working_set_defaults which is also a public function and should - * get its own unit test. + /* We only need to test that the allocated memory is non-NULL, as all the + * function does otherwise is call pcmk__set_scheduler_defaults(), which + * should have its own unit test. */ assert_non_null(scheduler); + assert_non_null(scheduler->priv); - /* Avoid calling pe_free_working_set here so we don't artificially - * inflate the coverage numbers. - */ + free(scheduler->priv); free(scheduler); } PCMK__UNIT_TEST(NULL, NULL, cmocka_unit_test(calloc_fails), cmocka_unit_test(calloc_succeeds)) diff --git a/lib/pacemaker/pcmk_rule.c b/lib/pacemaker/pcmk_rule.c index 6eb344e6b1..0d0fbd5abe 100644 --- a/lib/pacemaker/pcmk_rule.c +++ b/lib/pacemaker/pcmk_rule.c @@ -1,210 +1,210 @@ /* * Copyright 2022-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include "libpacemaker_private.h" #define XPATH_NODE_RULE "//" PCMK_XE_RULE "[@" PCMK_XA_ID "='%s']" /*! * \internal * \brief Check whether a given rule is in effect * * \param[in] scheduler Scheduler data * \param[in] rule_id The ID of the rule to check * \param[out] error Where to store a rule evaluation error message * * \return Standard Pacemaker return code */ static int eval_rule(pcmk_scheduler_t *scheduler, const char *rule_id, const char **error) { xmlNodePtr cib_constraints = NULL; xmlNodePtr match = NULL; xmlXPathObjectPtr xpath_obj = NULL; char *xpath = NULL; int rc = pcmk_rc_ok; int num_results = 0; *error = NULL; /* Rules are under the constraints node in the XML, so first find that. */ cib_constraints = pcmk_find_cib_element(scheduler->input, PCMK_XE_CONSTRAINTS); /* Get all rules matching the given ID that are also simple enough for us * to check. For the moment, these rules must only have a single * date_expression child and: * - Do not have a date_spec operation, or * - Have a date_spec operation that contains years= * * We do this in steps to provide better error messages. First, check that * there's any rule with the given ID. */ xpath = crm_strdup_printf(XPATH_NODE_RULE, rule_id); xpath_obj = xpath_search(cib_constraints, xpath); num_results = numXpathResults(xpath_obj); free(xpath); freeXpathObject(xpath_obj); if (num_results == 0) { *error = "Rule not found"; return ENXIO; } if (num_results > 1) { // Should not be possible; schema prevents this *error = "Found more than one rule with matching ID"; return pcmk_rc_duplicate_id; } /* Next, make sure it has exactly one date_expression. */ xpath = crm_strdup_printf(XPATH_NODE_RULE "//date_expression", rule_id); xpath_obj = xpath_search(cib_constraints, xpath); num_results = numXpathResults(xpath_obj); free(xpath); freeXpathObject(xpath_obj); if (num_results != 1) { if (num_results == 0) { *error = "Rule does not have a date expression"; } else { *error = "Rule has more than one date expression"; } return EOPNOTSUPP; } /* Then, check that it's something we actually support. */ xpath = crm_strdup_printf(XPATH_NODE_RULE "//" PCMK_XE_DATE_EXPRESSION "[@" PCMK_XA_OPERATION "!='" PCMK_VALUE_DATE_SPEC "']", rule_id); xpath_obj = xpath_search(cib_constraints, xpath); num_results = numXpathResults(xpath_obj); free(xpath); if (num_results == 0) { freeXpathObject(xpath_obj); xpath = crm_strdup_printf(XPATH_NODE_RULE "//" PCMK_XE_DATE_EXPRESSION "[@" PCMK_XA_OPERATION "='" PCMK_VALUE_DATE_SPEC "' " "and " PCMK_XE_DATE_SPEC "/@" PCMK_XA_YEARS "]", rule_id); xpath_obj = xpath_search(cib_constraints, xpath); num_results = numXpathResults(xpath_obj); free(xpath); if (num_results == 0) { freeXpathObject(xpath_obj); *error = "Rule must either not use " PCMK_XE_DATE_SPEC ", or use " PCMK_XE_DATE_SPEC " with " PCMK_XA_YEARS "="; return EOPNOTSUPP; } } match = getXpathResult(xpath_obj, 0); /* We should have ensured this with the xpath query above, but double- * checking can't hurt. */ pcmk__assert((match != NULL) && (pcmk__condition_type(match) == pcmk__condition_datetime)); rc = pcmk__evaluate_date_expression(match, scheduler->priv->now, NULL); if ((rc != pcmk_rc_ok) && (rc != pcmk_rc_within_range)) { // Malformed or missing *error = "Error parsing rule"; } freeXpathObject(xpath_obj); return rc; } /*! * \internal * \brief Check whether each rule in a list is in effect * * \param[in,out] out Output object * \param[in] input The CIB XML to check (if \c NULL, use current CIB) * \param[in] date Check whether the rule is in effect at this date and * time (if \c NULL, use current date and time) * \param[in] rule_ids The IDs of the rules to check, as a NULL- * terminated list. * * \return Standard Pacemaker return code */ int pcmk__check_rules(pcmk__output_t *out, xmlNodePtr input, const crm_time_t *date, const char **rule_ids) { pcmk_scheduler_t *scheduler = NULL; int rc = pcmk_rc_ok; pcmk__assert(out != NULL); if (rule_ids == NULL) { // Trivial case; every rule specified is in effect return pcmk_rc_ok; } rc = pcmk__init_scheduler(out, input, date, &scheduler); if (rc != pcmk_rc_ok) { return rc; } for (const char **rule_id = rule_ids; *rule_id != NULL; rule_id++) { const char *error = NULL; int last_rc = eval_rule(scheduler, *rule_id, &error); out->message(out, "rule-check", *rule_id, last_rc, error); if (last_rc != pcmk_rc_ok) { rc = last_rc; } } - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); return rc; } // Documented in pacemaker.h int pcmk_check_rules(xmlNodePtr *xml, xmlNodePtr input, const crm_time_t *date, const char **rule_ids) { pcmk__output_t *out = NULL; int rc = pcmk_rc_ok; rc = pcmk__xml_output_new(&out, xml); if (rc != pcmk_rc_ok) { return rc; } pcmk__register_lib_messages(out); rc = pcmk__check_rules(out, input, date, rule_ids); pcmk__xml_output_finish(out, pcmk_rc2exitc(rc), xml); return rc; } diff --git a/lib/pacemaker/pcmk_sched_actions.c b/lib/pacemaker/pcmk_sched_actions.c index e6ed67c0d9..70e5f9083b 100644 --- a/lib/pacemaker/pcmk_sched_actions.c +++ b/lib/pacemaker/pcmk_sched_actions.c @@ -1,1942 +1,1942 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include "libpacemaker_private.h" /*! * \internal * \brief Get the action flags relevant to ordering constraints * * \param[in,out] action Action to check * \param[in] node Node that *other* action in the ordering is on * (used only for clone resource actions) * * \return Action flags that should be used for orderings */ static uint32_t action_flags_for_ordering(pcmk_action_t *action, const pcmk_node_t *node) { bool runnable = false; uint32_t flags; // For non-resource actions, return the action flags if (action->rsc == NULL) { return action->flags; } /* For non-clone resources, or a clone action not assigned to a node, * return the flags as determined by the resource method without a node * specified. */ flags = action->rsc->priv->cmds->action_flags(action, NULL); if ((node == NULL) || !pcmk__is_clone(action->rsc)) { return flags; } /* Otherwise (i.e., for clone resource actions on a specific node), first * remember whether the non-node-specific action is runnable. */ runnable = pcmk_is_set(flags, pcmk__action_runnable); // Then recheck the resource method with the node flags = action->rsc->priv->cmds->action_flags(action, node); /* For clones in ordering constraints, the node-specific "runnable" doesn't * matter, just the non-node-specific setting (i.e., is the action runnable * anywhere). * * This applies only to runnable, and only for ordering constraints. This * function shouldn't be used for other types of constraints without * changes. Not very satisfying, but it's logical and appears to work well. */ if (runnable && !pcmk_is_set(flags, pcmk__action_runnable)) { pcmk__set_raw_action_flags(flags, action->rsc->id, pcmk__action_runnable); } return flags; } /*! * \internal * \brief Get action UUID that should be used with a resource ordering * * When an action is ordered relative to an action for a collective resource * (clone, group, or bundle), it actually needs to be ordered after all * instances of the collective have completed the relevant action (for example, * given "start CLONE then start RSC", RSC must wait until all instances of * CLONE have started). Given the UUID and resource of the first action in an * ordering, this returns the UUID of the action that should actually be used * for ordering (for example, "CLONE_started_0" instead of "CLONE_start_0"). * * \param[in] first_uuid UUID of first action in ordering * \param[in] first_rsc Resource of first action in ordering * * \return Newly allocated copy of UUID to use with ordering * \note It is the caller's responsibility to free the return value. */ static char * action_uuid_for_ordering(const char *first_uuid, const pcmk_resource_t *first_rsc) { guint interval_ms = 0; char *uuid = NULL; char *rid = NULL; char *first_task_str = NULL; enum pcmk__action_type first_task = pcmk__action_unspecified; enum pcmk__action_type remapped_task = pcmk__action_unspecified; // Only non-notify actions for collective resources need remapping if ((strstr(first_uuid, PCMK_ACTION_NOTIFY) != NULL) || (first_rsc->priv->variant < pcmk__rsc_variant_group)) { goto done; } // Only non-recurring actions need remapping pcmk__assert(parse_op_key(first_uuid, &rid, &first_task_str, &interval_ms)); if (interval_ms > 0) { goto done; } first_task = pcmk__parse_action(first_task_str); switch (first_task) { case pcmk__action_stop: case pcmk__action_start: case pcmk__action_notify: case pcmk__action_promote: case pcmk__action_demote: remapped_task = first_task + 1; break; case pcmk__action_stopped: case pcmk__action_started: case pcmk__action_notified: case pcmk__action_promoted: case pcmk__action_demoted: remapped_task = first_task; break; case pcmk__action_monitor: case pcmk__action_shutdown: case pcmk__action_fence: break; default: crm_err("Unknown action '%s' in ordering", first_task_str); break; } if (remapped_task != pcmk__action_unspecified) { /* If a clone or bundle has notifications enabled, the ordering will be * relative to when notifications have been sent for the remapped task. */ if (pcmk_is_set(first_rsc->flags, pcmk__rsc_notify) && (pcmk__is_clone(first_rsc) || pcmk__is_bundled(first_rsc))) { uuid = pcmk__notify_key(rid, "confirmed-post", pcmk__action_text(remapped_task)); } else { uuid = pcmk__op_key(rid, pcmk__action_text(remapped_task), 0); } pcmk__rsc_trace(first_rsc, "Remapped action UUID %s to %s for ordering purposes", first_uuid, uuid); } done: free(first_task_str); free(rid); return (uuid != NULL)? uuid : pcmk__str_copy(first_uuid); } /*! * \internal * \brief Get actual action that should be used with an ordering * * When an action is ordered relative to an action for a collective resource * (clone, group, or bundle), it actually needs to be ordered after all * instances of the collective have completed the relevant action (for example, * given "start CLONE then start RSC", RSC must wait until all instances of * CLONE have started). Given the first action in an ordering, this returns the * the action that should actually be used for ordering (for example, the * started action instead of the start action). * * \param[in] action First action in an ordering * * \return Actual action that should be used for the ordering */ static pcmk_action_t * action_for_ordering(pcmk_action_t *action) { pcmk_action_t *result = action; pcmk_resource_t *rsc = action->rsc; if (rsc == NULL) { return result; } if ((rsc->priv->variant >= pcmk__rsc_variant_group) && (action->uuid != NULL)) { char *uuid = action_uuid_for_ordering(action->uuid, rsc); result = find_first_action(rsc->priv->actions, uuid, NULL, NULL); if (result == NULL) { crm_warn("Not remapping %s to %s because %s does not have " "remapped action", action->uuid, uuid, rsc->id); result = action; } free(uuid); } return result; } /*! * \internal * \brief Wrapper for update_ordered_actions() method for readability * * \param[in,out] rsc Resource to call method for * \param[in,out] first 'First' action in an ordering * \param[in,out] then 'Then' action in an ordering * \param[in] node If not NULL, limit scope of ordering to this * node (only used when interleaving instances) * \param[in] flags Action flags for \p first for ordering purposes * \param[in] filter Action flags to limit scope of certain updates * (may include pcmk__action_optional to affect only * mandatory actions, and pe_action_runnable to * affect only runnable actions) * \param[in] type Group of enum pcmk__action_relation_flags to apply * \param[in,out] scheduler Scheduler data * * \return Group of enum pcmk__updated flags indicating what was updated */ static inline uint32_t update(pcmk_resource_t *rsc, pcmk_action_t *first, pcmk_action_t *then, const pcmk_node_t *node, uint32_t flags, uint32_t filter, uint32_t type, pcmk_scheduler_t *scheduler) { return rsc->priv->cmds->update_ordered_actions(first, then, node, flags, filter, type, scheduler); } /*! * \internal * \brief Update flags for ordering's actions appropriately for ordering's flags * * \param[in,out] first First action in an ordering * \param[in,out] then Then action in an ordering * \param[in] first_flags Action flags for \p first for ordering purposes * \param[in] then_flags Action flags for \p then for ordering purposes * \param[in,out] order Action wrapper for \p first in ordering * \param[in,out] scheduler Scheduler data * * \return Group of enum pcmk__updated flags */ static uint32_t update_action_for_ordering_flags(pcmk_action_t *first, pcmk_action_t *then, uint32_t first_flags, uint32_t then_flags, pcmk__related_action_t *order, pcmk_scheduler_t *scheduler) { uint32_t changed = pcmk__updated_none; /* The node will only be used for clones. If interleaved, node will be NULL, * otherwise the ordering scope will be limited to the node. Normally, the * whole 'then' clone should restart if 'first' is restarted, so then->node * is needed. */ pcmk_node_t *node = then->node; if (pcmk_is_set(order->flags, pcmk__ar_first_implies_same_node_then)) { /* For unfencing, only instances of 'then' on the same node as 'first' * (the unfencing operation) should restart, so reset node to * first->node, at which point this case is handled like a normal * pcmk__ar_first_implies_then. */ pcmk__clear_relation_flags(order->flags, pcmk__ar_first_implies_same_node_then); pcmk__set_relation_flags(order->flags, pcmk__ar_first_implies_then); node = first->node; pcmk__rsc_trace(then->rsc, "%s then %s: mapped " "pcmk__ar_first_implies_same_node_then to " "pcmk__ar_first_implies_then on %s", first->uuid, then->uuid, pcmk__node_name(node)); } if (pcmk_is_set(order->flags, pcmk__ar_first_implies_then)) { if (then->rsc != NULL) { changed |= update(then->rsc, first, then, node, first_flags & pcmk__action_optional, pcmk__action_optional, pcmk__ar_first_implies_then, scheduler); } else if (!pcmk_is_set(first_flags, pcmk__action_optional) && pcmk_is_set(then->flags, pcmk__action_optional)) { pcmk__clear_action_flags(then, pcmk__action_optional); pcmk__set_updated_flags(changed, first, pcmk__updated_then); } pcmk__rsc_trace(then->rsc, "%s then %s: %s after pcmk__ar_first_implies_then", first->uuid, then->uuid, (changed? "changed" : "unchanged")); } if (pcmk_is_set(order->flags, pcmk__ar_intermediate_stop) && (then->rsc != NULL)) { enum pcmk__action_flags restart = pcmk__action_optional |pcmk__action_runnable; changed |= update(then->rsc, first, then, node, first_flags, restart, pcmk__ar_intermediate_stop, scheduler); pcmk__rsc_trace(then->rsc, "%s then %s: %s after pcmk__ar_intermediate_stop", first->uuid, then->uuid, (changed? "changed" : "unchanged")); } if (pcmk_is_set(order->flags, pcmk__ar_then_implies_first)) { if (first->rsc != NULL) { changed |= update(first->rsc, first, then, node, first_flags, pcmk__action_optional, pcmk__ar_then_implies_first, scheduler); } else if (!pcmk_is_set(first_flags, pcmk__action_optional) && pcmk_is_set(first->flags, pcmk__action_runnable)) { pcmk__clear_action_flags(first, pcmk__action_runnable); pcmk__set_updated_flags(changed, first, pcmk__updated_first); } pcmk__rsc_trace(then->rsc, "%s then %s: %s after pcmk__ar_then_implies_first", first->uuid, then->uuid, (changed? "changed" : "unchanged")); } if (pcmk_is_set(order->flags, pcmk__ar_promoted_then_implies_first)) { if (then->rsc != NULL) { changed |= update(then->rsc, first, then, node, first_flags & pcmk__action_optional, pcmk__action_optional, pcmk__ar_promoted_then_implies_first, scheduler); } pcmk__rsc_trace(then->rsc, "%s then %s: %s after " "pcmk__ar_promoted_then_implies_first", first->uuid, then->uuid, (changed? "changed" : "unchanged")); } if (pcmk_is_set(order->flags, pcmk__ar_min_runnable)) { if (then->rsc != NULL) { changed |= update(then->rsc, first, then, node, first_flags, pcmk__action_runnable, pcmk__ar_min_runnable, scheduler); } else if (pcmk_is_set(first_flags, pcmk__action_runnable)) { // We have another runnable instance of "first" then->runnable_before++; /* Mark "then" as runnable if it requires a certain number of * "before" instances to be runnable, and they now are. */ if ((then->runnable_before >= then->required_runnable_before) && !pcmk_is_set(then->flags, pcmk__action_runnable)) { pcmk__set_action_flags(then, pcmk__action_runnable); pcmk__set_updated_flags(changed, first, pcmk__updated_then); } } pcmk__rsc_trace(then->rsc, "%s then %s: %s after pcmk__ar_min_runnable", first->uuid, then->uuid, (changed? "changed" : "unchanged")); } if (pcmk_is_set(order->flags, pcmk__ar_nested_remote_probe) && (then->rsc != NULL)) { if (!pcmk_is_set(first_flags, pcmk__action_runnable) && (first->rsc != NULL) && (first->rsc->priv->active_nodes != NULL)) { pcmk__rsc_trace(then->rsc, "%s then %s: ignoring because first is stopping", first->uuid, then->uuid); order->flags = pcmk__ar_none; } else { changed |= update(then->rsc, first, then, node, first_flags, pcmk__action_runnable, pcmk__ar_unrunnable_first_blocks, scheduler); } pcmk__rsc_trace(then->rsc, "%s then %s: %s after pcmk__ar_nested_remote_probe", first->uuid, then->uuid, (changed? "changed" : "unchanged")); } if (pcmk_is_set(order->flags, pcmk__ar_unrunnable_first_blocks)) { if (then->rsc != NULL) { changed |= update(then->rsc, first, then, node, first_flags, pcmk__action_runnable, pcmk__ar_unrunnable_first_blocks, scheduler); } else if (!pcmk_is_set(first_flags, pcmk__action_runnable) && pcmk_is_set(then->flags, pcmk__action_runnable)) { pcmk__clear_action_flags(then, pcmk__action_runnable); pcmk__set_updated_flags(changed, first, pcmk__updated_then); } pcmk__rsc_trace(then->rsc, "%s then %s: %s after pcmk__ar_unrunnable_first_blocks", first->uuid, then->uuid, (changed? "changed" : "unchanged")); } if (pcmk_is_set(order->flags, pcmk__ar_unmigratable_then_blocks)) { if (then->rsc != NULL) { changed |= update(then->rsc, first, then, node, first_flags, pcmk__action_optional, pcmk__ar_unmigratable_then_blocks, scheduler); } pcmk__rsc_trace(then->rsc, "%s then %s: %s after " "pcmk__ar_unmigratable_then_blocks", first->uuid, then->uuid, (changed? "changed" : "unchanged")); } if (pcmk_is_set(order->flags, pcmk__ar_first_else_then)) { if (then->rsc != NULL) { changed |= update(then->rsc, first, then, node, first_flags, pcmk__action_optional, pcmk__ar_first_else_then, scheduler); } pcmk__rsc_trace(then->rsc, "%s then %s: %s after pcmk__ar_first_else_then", first->uuid, then->uuid, (changed? "changed" : "unchanged")); } if (pcmk_is_set(order->flags, pcmk__ar_ordered)) { if (then->rsc != NULL) { changed |= update(then->rsc, first, then, node, first_flags, pcmk__action_runnable, pcmk__ar_ordered, scheduler); } pcmk__rsc_trace(then->rsc, "%s then %s: %s after pcmk__ar_ordered", first->uuid, then->uuid, (changed? "changed" : "unchanged")); } if (pcmk_is_set(order->flags, pcmk__ar_asymmetric)) { if (then->rsc != NULL) { changed |= update(then->rsc, first, then, node, first_flags, pcmk__action_runnable, pcmk__ar_asymmetric, scheduler); } pcmk__rsc_trace(then->rsc, "%s then %s: %s after pcmk__ar_asymmetric", first->uuid, then->uuid, (changed? "changed" : "unchanged")); } if (pcmk_is_set(first->flags, pcmk__action_runnable) && pcmk_is_set(order->flags, pcmk__ar_first_implies_then_graphed) && !pcmk_is_set(first_flags, pcmk__action_optional)) { pcmk__rsc_trace(then->rsc, "%s will be in graph because %s is required", then->uuid, first->uuid); pcmk__set_action_flags(then, pcmk__action_always_in_graph); // Don't bother marking 'then' as changed just for this } if (pcmk_is_set(order->flags, pcmk__ar_then_implies_first_graphed) && !pcmk_is_set(then_flags, pcmk__action_optional)) { pcmk__rsc_trace(then->rsc, "%s will be in graph because %s is required", first->uuid, then->uuid); pcmk__set_action_flags(first, pcmk__action_always_in_graph); // Don't bother marking 'first' as changed just for this } if (pcmk_any_flags_set(order->flags, pcmk__ar_first_implies_then |pcmk__ar_then_implies_first |pcmk__ar_intermediate_stop) && (first->rsc != NULL) && !pcmk_is_set(first->rsc->flags, pcmk__rsc_managed) && pcmk_is_set(first->rsc->flags, pcmk__rsc_blocked) && !pcmk_is_set(first->flags, pcmk__action_runnable) && pcmk__str_eq(first->task, PCMK_ACTION_STOP, pcmk__str_none)) { if (pcmk_is_set(then->flags, pcmk__action_runnable)) { pcmk__clear_action_flags(then, pcmk__action_runnable); pcmk__set_updated_flags(changed, first, pcmk__updated_then); } pcmk__rsc_trace(then->rsc, "%s then %s: %s after checking whether first " "is blocked, unmanaged, unrunnable stop", first->uuid, then->uuid, (changed? "changed" : "unchanged")); } return changed; } // Convenience macros for logging action properties #define action_type_str(flags) \ (pcmk_is_set((flags), pcmk__action_pseudo)? "pseudo-action" : "action") #define action_optional_str(flags) \ (pcmk_is_set((flags), pcmk__action_optional)? "optional" : "required") #define action_runnable_str(flags) \ (pcmk_is_set((flags), pcmk__action_runnable)? "runnable" : "unrunnable") #define action_node_str(a) \ (((a)->node == NULL)? "no node" : (a)->node->priv->name) /*! * \internal * \brief Update an action's flags for all orderings where it is "then" * * \param[in,out] then Action to update * \param[in,out] scheduler Scheduler data */ void pcmk__update_action_for_orderings(pcmk_action_t *then, pcmk_scheduler_t *scheduler) { GList *lpc = NULL; uint32_t changed = pcmk__updated_none; int last_flags = then->flags; pcmk__rsc_trace(then->rsc, "Updating %s %s (%s %s) on %s", action_type_str(then->flags), then->uuid, action_optional_str(then->flags), action_runnable_str(then->flags), action_node_str(then)); if (then->required_runnable_before > 0) { /* Initialize current known "runnable before" actions. As * update_action_for_ordering_flags() is called for each of then's * before actions, this number will increment as runnable 'first' * actions are encountered. */ then->runnable_before = 0; /* The pcmk__ar_min_runnable clause of * update_action_for_ordering_flags() (called below) * will reset runnable if appropriate. */ pcmk__clear_action_flags(then, pcmk__action_runnable); } for (lpc = then->actions_before; lpc != NULL; lpc = lpc->next) { pcmk__related_action_t *other = lpc->data; pcmk_action_t *first = other->action; pcmk_node_t *then_node = then->node; pcmk_node_t *first_node = first->node; const uint32_t target = pcmk__rsc_node_assigned; if ((first->rsc != NULL) && pcmk__is_group(first->rsc) && pcmk__str_eq(first->task, PCMK_ACTION_START, pcmk__str_none)) { first_node = first->rsc->priv->fns->location(first->rsc, NULL, target); if (first_node != NULL) { pcmk__rsc_trace(first->rsc, "Found %s for 'first' %s", pcmk__node_name(first_node), first->uuid); } } if (pcmk__is_group(then->rsc) && pcmk__str_eq(then->task, PCMK_ACTION_START, pcmk__str_none)) { then_node = then->rsc->priv->fns->location(then->rsc, NULL, target); if (then_node != NULL) { pcmk__rsc_trace(then->rsc, "Found %s for 'then' %s", pcmk__node_name(then_node), then->uuid); } } // Disable constraint if it only applies when on same node, but isn't if (pcmk_is_set(other->flags, pcmk__ar_if_on_same_node) && (first_node != NULL) && (then_node != NULL) && !pcmk__same_node(first_node, then_node)) { pcmk__rsc_trace(then->rsc, "Disabled ordering %s on %s then %s on %s: " "not same node", other->action->uuid, pcmk__node_name(first_node), then->uuid, pcmk__node_name(then_node)); other->flags = pcmk__ar_none; continue; } pcmk__clear_updated_flags(changed, then, pcmk__updated_first); if ((first->rsc != NULL) && pcmk_is_set(other->flags, pcmk__ar_then_cancels_first) && !pcmk_is_set(then->flags, pcmk__action_optional)) { /* 'then' is required, so we must abandon 'first' * (e.g. a required stop cancels any agent reload). */ pcmk__set_action_flags(other->action, pcmk__action_optional); if (!strcmp(first->task, PCMK_ACTION_RELOAD_AGENT)) { pcmk__clear_rsc_flags(first->rsc, pcmk__rsc_reload); } } if ((first->rsc != NULL) && (then->rsc != NULL) && (first->rsc != then->rsc) && !is_parent(then->rsc, first->rsc)) { first = action_for_ordering(first); } if (first != other->action) { pcmk__rsc_trace(then->rsc, "Ordering %s after %s instead of %s", then->uuid, first->uuid, other->action->uuid); } pcmk__rsc_trace(then->rsc, "%s (%#.6x) then %s (%#.6x): type=%#.6x node=%s", first->uuid, first->flags, then->uuid, then->flags, other->flags, action_node_str(first)); if (first == other->action) { /* 'first' was not remapped (e.g. from 'start' to 'running'), which * could mean it is a non-resource action, a primitive resource * action, or already expanded. */ uint32_t first_flags, then_flags; first_flags = action_flags_for_ordering(first, then_node); then_flags = action_flags_for_ordering(then, first_node); changed |= update_action_for_ordering_flags(first, then, first_flags, then_flags, other, scheduler); /* 'first' was for a complex resource (clone, group, etc), * create a new dependency if necessary */ } else if (order_actions(first, then, other->flags)) { /* This was the first time 'first' and 'then' were associated, * start again to get the new actions_before list */ pcmk__set_updated_flags(changed, then, pcmk__updated_then); pcmk__rsc_trace(then->rsc, "Disabled ordering %s then %s in favor of %s " "then %s", other->action->uuid, then->uuid, first->uuid, then->uuid); other->flags = pcmk__ar_none; } if (pcmk_is_set(changed, pcmk__updated_first)) { crm_trace("Re-processing %s and its 'after' actions " "because it changed", first->uuid); for (GList *lpc2 = first->actions_after; lpc2 != NULL; lpc2 = lpc2->next) { pcmk__related_action_t *other = lpc2->data; pcmk__update_action_for_orderings(other->action, scheduler); } pcmk__update_action_for_orderings(first, scheduler); } } if (then->required_runnable_before > 0) { if (last_flags == then->flags) { pcmk__clear_updated_flags(changed, then, pcmk__updated_then); } else { pcmk__set_updated_flags(changed, then, pcmk__updated_then); } } if (pcmk_is_set(changed, pcmk__updated_then)) { crm_trace("Re-processing %s and its 'after' actions because it changed", then->uuid); if (pcmk_is_set(last_flags, pcmk__action_runnable) && !pcmk_is_set(then->flags, pcmk__action_runnable)) { pcmk__block_colocation_dependents(then); } pcmk__update_action_for_orderings(then, scheduler); for (lpc = then->actions_after; lpc != NULL; lpc = lpc->next) { pcmk__related_action_t *other = lpc->data; pcmk__update_action_for_orderings(other->action, scheduler); } } } static inline bool is_primitive_action(const pcmk_action_t *action) { return (action != NULL) && pcmk__is_primitive(action->rsc); } /*! * \internal * \brief Clear a single action flag and set reason text * * \param[in,out] action Action whose flag should be cleared * \param[in] flag Action flag that should be cleared * \param[in] reason Action that is the reason why flag is being cleared */ #define clear_action_flag_because(action, flag, reason) do { \ if (pcmk_is_set((action)->flags, (flag))) { \ pcmk__clear_action_flags(action, flag); \ if ((action)->rsc != (reason)->rsc) { \ char *reason_text = pe__action2reason((reason), (flag)); \ pe_action_set_reason((action), reason_text, false); \ free(reason_text); \ } \ } \ } while (0) /*! * \internal * \brief Update actions in an asymmetric ordering * * If the "first" action in an asymmetric ordering is unrunnable, make the * "second" action unrunnable as well, if appropriate. * * \param[in] first 'First' action in an asymmetric ordering * \param[in,out] then 'Then' action in an asymmetric ordering */ static void handle_asymmetric_ordering(const pcmk_action_t *first, pcmk_action_t *then) { /* Only resource actions after an unrunnable 'first' action need updates for * asymmetric ordering. */ if ((then->rsc == NULL) || pcmk_is_set(first->flags, pcmk__action_runnable)) { return; } // Certain optional 'then' actions are unaffected by unrunnable 'first' if (pcmk_is_set(then->flags, pcmk__action_optional)) { enum rsc_role_e then_rsc_role; then_rsc_role = then->rsc->priv->fns->state(then->rsc, TRUE); if ((then_rsc_role == pcmk_role_stopped) && pcmk__str_eq(then->task, PCMK_ACTION_STOP, pcmk__str_none)) { /* If 'then' should stop after 'first' but is already stopped, the * ordering is irrelevant. */ return; } else if ((then_rsc_role >= pcmk_role_started) && pcmk__str_eq(then->task, PCMK_ACTION_START, pcmk__str_none) && pe__rsc_running_on_only(then->rsc, then->node)) { /* Similarly if 'then' should start after 'first' but is already * started on a single node. */ return; } } // 'First' can't run, so 'then' can't either clear_action_flag_because(then, pcmk__action_optional, first); clear_action_flag_because(then, pcmk__action_runnable, first); } /*! * \internal * \brief Set action bits appropriately when pcmk__ar_intermediate_stop is used * * \param[in,out] first 'First' action in ordering * \param[in,out] then 'Then' action in ordering * \param[in] filter What action flags to care about * * \note pcmk__ar_intermediate_stop is set for "stop resource before starting * it" and "stop later group member before stopping earlier group member" */ static void handle_restart_ordering(pcmk_action_t *first, pcmk_action_t *then, uint32_t filter) { const char *reason = NULL; pcmk__assert(is_primitive_action(first) && is_primitive_action(then)); // We need to update the action in two cases: // ... if 'then' is required if (pcmk_is_set(filter, pcmk__action_optional) && !pcmk_is_set(then->flags, pcmk__action_optional)) { reason = "restart"; } /* ... if 'then' is unrunnable action on same resource (if a resource * should restart but can't start, we still want to stop) */ if (pcmk_is_set(filter, pcmk__action_runnable) && !pcmk_is_set(then->flags, pcmk__action_runnable) && pcmk_is_set(then->rsc->flags, pcmk__rsc_managed) && (first->rsc == then->rsc)) { reason = "stop"; } if (reason == NULL) { return; } pcmk__rsc_trace(first->rsc, "Handling %s -> %s for %s", first->uuid, then->uuid, reason); // Make 'first' required if it is runnable if (pcmk_is_set(first->flags, pcmk__action_runnable)) { clear_action_flag_because(first, pcmk__action_optional, then); } // Make 'first' required if 'then' is required if (!pcmk_is_set(then->flags, pcmk__action_optional)) { clear_action_flag_because(first, pcmk__action_optional, then); } // Make 'first' unmigratable if 'then' is unmigratable if (!pcmk_is_set(then->flags, pcmk__action_migratable)) { clear_action_flag_because(first, pcmk__action_migratable, then); } // Make 'then' unrunnable if 'first' is required but unrunnable if (!pcmk_is_set(first->flags, pcmk__action_optional) && !pcmk_is_set(first->flags, pcmk__action_runnable)) { clear_action_flag_because(then, pcmk__action_runnable, first); } } /*! * \internal * \brief Update two actions according to an ordering between them * * Given information about an ordering of two actions, update the actions' flags * (and runnable_before members if appropriate) as appropriate for the ordering. * Effects may cascade to other orderings involving the actions as well. * * \param[in,out] first 'First' action in an ordering * \param[in,out] then 'Then' action in an ordering * \param[in] node If not NULL, limit scope of ordering to this node * (ignored) * \param[in] flags Action flags for \p first for ordering purposes * \param[in] filter Action flags to limit scope of certain updates (may * include pcmk__action_optional to affect only * mandatory actions, and pcmk__action_runnable to * affect only runnable actions) * \param[in] type Group of enum pcmk__action_relation_flags to apply * \param[in,out] scheduler Scheduler data * * \return Group of enum pcmk__updated flags indicating what was updated */ uint32_t pcmk__update_ordered_actions(pcmk_action_t *first, pcmk_action_t *then, const pcmk_node_t *node, uint32_t flags, uint32_t filter, uint32_t type, pcmk_scheduler_t *scheduler) { uint32_t changed = pcmk__updated_none; uint32_t then_flags = 0U; uint32_t first_flags = 0U; pcmk__assert((first != NULL) && (then != NULL) && (scheduler != NULL)); then_flags = then->flags; first_flags = first->flags; if (pcmk_is_set(type, pcmk__ar_asymmetric)) { handle_asymmetric_ordering(first, then); } if (pcmk_is_set(type, pcmk__ar_then_implies_first) && !pcmk_is_set(then_flags, pcmk__action_optional)) { // Then is required, and implies first should be, too if (pcmk_is_set(filter, pcmk__action_optional) && !pcmk_is_set(flags, pcmk__action_optional) && pcmk_is_set(first_flags, pcmk__action_optional)) { clear_action_flag_because(first, pcmk__action_optional, then); } if (pcmk_is_set(flags, pcmk__action_migratable) && !pcmk_is_set(then->flags, pcmk__action_migratable)) { clear_action_flag_because(first, pcmk__action_migratable, then); } } if (pcmk_is_set(type, pcmk__ar_promoted_then_implies_first) && (then->rsc != NULL) && (then->rsc->priv->orig_role == pcmk_role_promoted) && pcmk_is_set(filter, pcmk__action_optional) && !pcmk_is_set(then->flags, pcmk__action_optional)) { clear_action_flag_because(first, pcmk__action_optional, then); if (pcmk_is_set(first->flags, pcmk__action_migratable) && !pcmk_is_set(then->flags, pcmk__action_migratable)) { clear_action_flag_because(first, pcmk__action_migratable, then); } } if (pcmk_is_set(type, pcmk__ar_unmigratable_then_blocks) && pcmk_is_set(filter, pcmk__action_optional)) { if (!pcmk_all_flags_set(then->flags, pcmk__action_migratable |pcmk__action_runnable)) { clear_action_flag_because(first, pcmk__action_runnable, then); } if (!pcmk_is_set(then->flags, pcmk__action_optional)) { clear_action_flag_because(first, pcmk__action_optional, then); } } if (pcmk_is_set(type, pcmk__ar_first_else_then) && pcmk_is_set(filter, pcmk__action_optional) && !pcmk_is_set(first->flags, pcmk__action_runnable)) { clear_action_flag_because(then, pcmk__action_migratable, first); pcmk__clear_action_flags(then, pcmk__action_pseudo); } if (pcmk_is_set(type, pcmk__ar_unrunnable_first_blocks) && pcmk_is_set(filter, pcmk__action_runnable) && pcmk_is_set(then->flags, pcmk__action_runnable) && !pcmk_is_set(flags, pcmk__action_runnable)) { clear_action_flag_because(then, pcmk__action_runnable, first); clear_action_flag_because(then, pcmk__action_migratable, first); } if (pcmk_is_set(type, pcmk__ar_first_implies_then) && pcmk_is_set(filter, pcmk__action_optional) && pcmk_is_set(then->flags, pcmk__action_optional) && !pcmk_is_set(flags, pcmk__action_optional) && !pcmk_is_set(first->flags, pcmk__action_migratable)) { clear_action_flag_because(then, pcmk__action_optional, first); } if (pcmk_is_set(type, pcmk__ar_intermediate_stop)) { handle_restart_ordering(first, then, filter); } if (then_flags != then->flags) { pcmk__set_updated_flags(changed, first, pcmk__updated_then); pcmk__rsc_trace(then->rsc, "%s on %s: flags are now %#.6x (was %#.6x) " "because of 'first' %s (%#.6x)", then->uuid, pcmk__node_name(then->node), then->flags, then_flags, first->uuid, first->flags); if ((then->rsc != NULL) && (then->rsc->priv->parent != NULL)) { // Required to handle "X_stop then X_start" for cloned groups pcmk__update_action_for_orderings(then, scheduler); } } if (first_flags != first->flags) { pcmk__set_updated_flags(changed, first, pcmk__updated_first); pcmk__rsc_trace(first->rsc, "%s on %s: flags are now %#.6x (was %#.6x) " "because of 'then' %s (%#.6x)", first->uuid, pcmk__node_name(first->node), first->flags, first_flags, then->uuid, then->flags); } return changed; } /*! * \internal * \brief Trace-log an action (optionally with its dependent actions) * * \param[in] pre_text If not NULL, prefix the log with this plus ": " * \param[in] action Action to log * \param[in] details If true, recursively log dependent actions */ void pcmk__log_action(const char *pre_text, const pcmk_action_t *action, bool details) { const char *node_uname = NULL; const char *node_uuid = NULL; const char *desc = NULL; CRM_CHECK(action != NULL, return); if (!pcmk_is_set(action->flags, pcmk__action_pseudo)) { if (action->node != NULL) { node_uname = action->node->priv->name; node_uuid = action->node->priv->id; } else { node_uname = ""; } } switch (pcmk__parse_action(action->task)) { case pcmk__action_fence: case pcmk__action_shutdown: if (pcmk_is_set(action->flags, pcmk__action_pseudo)) { desc = "Pseudo "; } else if (pcmk_is_set(action->flags, pcmk__action_optional)) { desc = "Optional "; } else if (!pcmk_is_set(action->flags, pcmk__action_runnable)) { desc = "!!Non-Startable!! "; } else { desc = "(Provisional) "; } crm_trace("%s%s%sAction %d: %s%s%s%s%s%s", ((pre_text == NULL)? "" : pre_text), ((pre_text == NULL)? "" : ": "), desc, action->id, action->uuid, (node_uname? "\ton " : ""), (node_uname? node_uname : ""), (node_uuid? "\t\t(" : ""), (node_uuid? node_uuid : ""), (node_uuid? ")" : "")); break; default: if (pcmk_is_set(action->flags, pcmk__action_optional)) { desc = "Optional "; } else if (pcmk_is_set(action->flags, pcmk__action_pseudo)) { desc = "Pseudo "; } else if (!pcmk_is_set(action->flags, pcmk__action_runnable)) { desc = "!!Non-Startable!! "; } else { desc = "(Provisional) "; } crm_trace("%s%s%sAction %d: %s %s%s%s%s%s%s", ((pre_text == NULL)? "" : pre_text), ((pre_text == NULL)? "" : ": "), desc, action->id, action->uuid, (action->rsc? action->rsc->id : ""), (node_uname? "\ton " : ""), (node_uname? node_uname : ""), (node_uuid? "\t\t(" : ""), (node_uuid? node_uuid : ""), (node_uuid? ")" : "")); break; } if (details) { const GList *iter = NULL; const pcmk__related_action_t *other = NULL; crm_trace("\t\t====== Preceding Actions"); for (iter = action->actions_before; iter != NULL; iter = iter->next) { other = (const pcmk__related_action_t *) iter->data; pcmk__log_action("\t\t", other->action, false); } crm_trace("\t\t====== Subsequent Actions"); for (iter = action->actions_after; iter != NULL; iter = iter->next) { other = (const pcmk__related_action_t *) iter->data; pcmk__log_action("\t\t", other->action, false); } crm_trace("\t\t====== End"); } else { crm_trace("\t\t(before=%d, after=%d)", g_list_length(action->actions_before), g_list_length(action->actions_after)); } } /*! * \internal * \brief Create a new shutdown action for a node * * \param[in,out] node Node being shut down * * \return Newly created shutdown action for \p node */ pcmk_action_t * pcmk__new_shutdown_action(pcmk_node_t *node) { char *shutdown_id = NULL; pcmk_action_t *shutdown_op = NULL; pcmk__assert(node != NULL); shutdown_id = crm_strdup_printf("%s-%s", PCMK_ACTION_DO_SHUTDOWN, node->priv->name); shutdown_op = custom_action(NULL, shutdown_id, PCMK_ACTION_DO_SHUTDOWN, node, FALSE, node->priv->scheduler); pcmk__order_stops_before_shutdown(node, shutdown_op); pcmk__insert_meta(shutdown_op, PCMK__META_OP_NO_WAIT, PCMK_VALUE_TRUE); return shutdown_op; } /*! * \internal * \brief Calculate and add an operation digest to XML * * Calculate an operation digest, which enables us to later determine when a * restart is needed due to the resource's parameters being changed, and add it * to given XML. * * \param[in] op Operation result from executor * \param[in,out] update XML to add digest to */ static void add_op_digest_to_xml(const lrmd_event_data_t *op, xmlNode *update) { char *digest = NULL; xmlNode *args_xml = NULL; if (op->params == NULL) { return; } args_xml = pcmk__xe_create(NULL, PCMK_XE_PARAMETERS); g_hash_table_foreach(op->params, hash2field, args_xml); pcmk__filter_op_for_digest(args_xml); digest = pcmk__digest_operation(args_xml); crm_xml_add(update, PCMK__XA_OP_DIGEST, digest); pcmk__xml_free(args_xml); free(digest); } #define FAKE_TE_ID "xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx" /*! * \internal * \brief Create XML for resource operation history update * * \param[in,out] parent Parent XML node to add to * \param[in,out] op Operation event data * \param[in] caller_version DC feature set * \param[in] target_rc Expected result of operation * \param[in] node Name of node on which operation was performed * \param[in] origin Arbitrary description of update source * * \return Newly created XML node for history update */ xmlNode * pcmk__create_history_xml(xmlNode *parent, lrmd_event_data_t *op, const char *caller_version, int target_rc, const char *node, const char *origin) { char *key = NULL; char *magic = NULL; char *op_id = NULL; char *op_id_additional = NULL; char *local_user_data = NULL; const char *exit_reason = NULL; xmlNode *xml_op = NULL; const char *task = NULL; CRM_CHECK(op != NULL, return NULL); crm_trace("Creating history XML for %s-interval %s action for %s on %s " "(DC version: %s, origin: %s)", pcmk__readable_interval(op->interval_ms), op->op_type, op->rsc_id, ((node == NULL)? "no node" : node), caller_version, origin); task = op->op_type; /* Record a successful agent reload as a start, and a failed one as a * monitor, to make life easier for the scheduler when determining the * current state. * * @COMPAT We should check "reload" here only if the operation was for a * pre-OCF-1.1 resource agent, but we don't know that here, and we should * only ever get results for actions scheduled by us, so we can reasonably * assume any "reload" is actually a pre-1.1 agent reload. */ if (pcmk__str_any_of(task, PCMK_ACTION_RELOAD, PCMK_ACTION_RELOAD_AGENT, NULL)) { if (op->op_status == PCMK_EXEC_DONE) { task = PCMK_ACTION_START; } else { task = PCMK_ACTION_MONITOR; } } key = pcmk__op_key(op->rsc_id, task, op->interval_ms); if (pcmk__str_eq(task, PCMK_ACTION_NOTIFY, pcmk__str_none)) { const char *n_type = crm_meta_value(op->params, "notify_type"); const char *n_task = crm_meta_value(op->params, "notify_operation"); CRM_LOG_ASSERT(n_type != NULL); CRM_LOG_ASSERT(n_task != NULL); op_id = pcmk__notify_key(op->rsc_id, n_type, n_task); if (op->op_status != PCMK_EXEC_PENDING) { /* Ignore notify errors. * * @TODO It might be better to keep the correct result here, and * ignore it in process_graph_event(). */ lrmd__set_result(op, PCMK_OCF_OK, PCMK_EXEC_DONE, NULL); } /* Migration history is preserved separately, which usually matters for * multiple nodes and is important for future cluster transitions. */ } else if (pcmk__str_any_of(op->op_type, PCMK_ACTION_MIGRATE_TO, PCMK_ACTION_MIGRATE_FROM, NULL)) { op_id = strdup(key); } else if (did_rsc_op_fail(op, target_rc)) { op_id = pcmk__op_key(op->rsc_id, "last_failure", 0); if (op->interval_ms == 0) { /* Ensure 'last' gets updated, in case PCMK_META_RECORD_PENDING is * true */ op_id_additional = pcmk__op_key(op->rsc_id, "last", 0); } exit_reason = op->exit_reason; } else if (op->interval_ms > 0) { op_id = strdup(key); } else { op_id = pcmk__op_key(op->rsc_id, "last", 0); } again: xml_op = pcmk__xe_first_child(parent, PCMK__XE_LRM_RSC_OP, PCMK_XA_ID, op_id); if (xml_op == NULL) { xml_op = pcmk__xe_create(parent, PCMK__XE_LRM_RSC_OP); } if (op->user_data == NULL) { crm_debug("Generating fake transition key for: " PCMK__OP_FMT " %d from %s", op->rsc_id, op->op_type, op->interval_ms, op->call_id, origin); local_user_data = pcmk__transition_key(-1, op->call_id, target_rc, FAKE_TE_ID); op->user_data = local_user_data; } if (magic == NULL) { magic = crm_strdup_printf("%d:%d;%s", op->op_status, op->rc, (const char *) op->user_data); } crm_xml_add(xml_op, PCMK_XA_ID, op_id); crm_xml_add(xml_op, PCMK__XA_OPERATION_KEY, key); crm_xml_add(xml_op, PCMK_XA_OPERATION, task); crm_xml_add(xml_op, PCMK_XA_CRM_DEBUG_ORIGIN, origin); crm_xml_add(xml_op, PCMK_XA_CRM_FEATURE_SET, caller_version); crm_xml_add(xml_op, PCMK__XA_TRANSITION_KEY, op->user_data); crm_xml_add(xml_op, PCMK__XA_TRANSITION_MAGIC, magic); crm_xml_add(xml_op, PCMK_XA_EXIT_REASON, pcmk__s(exit_reason, "")); crm_xml_add(xml_op, PCMK__META_ON_NODE, node); // For context during triage crm_xml_add_int(xml_op, PCMK__XA_CALL_ID, op->call_id); crm_xml_add_int(xml_op, PCMK__XA_RC_CODE, op->rc); crm_xml_add_int(xml_op, PCMK__XA_OP_STATUS, op->op_status); crm_xml_add_ms(xml_op, PCMK_META_INTERVAL, op->interval_ms); if ((op->t_run > 0) || (op->t_rcchange > 0) || (op->exec_time > 0) || (op->queue_time > 0)) { crm_trace("Timing data (" PCMK__OP_FMT "): " "last=%lld change=%lld exec=%u queue=%u", op->rsc_id, op->op_type, op->interval_ms, (long long) op->t_run, (long long) op->t_rcchange, op->exec_time, op->queue_time); if ((op->interval_ms > 0) && (op->t_rcchange > 0)) { // Recurring ops may have changed rc after initial run crm_xml_add_ll(xml_op, PCMK_XA_LAST_RC_CHANGE, (long long) op->t_rcchange); } else { crm_xml_add_ll(xml_op, PCMK_XA_LAST_RC_CHANGE, (long long) op->t_run); } crm_xml_add_int(xml_op, PCMK_XA_EXEC_TIME, op->exec_time); crm_xml_add_int(xml_op, PCMK_XA_QUEUE_TIME, op->queue_time); } if (pcmk__str_any_of(op->op_type, PCMK_ACTION_MIGRATE_TO, PCMK_ACTION_MIGRATE_FROM, NULL)) { /* Record PCMK__META_MIGRATE_SOURCE and PCMK__META_MIGRATE_TARGET always * for migrate ops. */ const char *name = PCMK__META_MIGRATE_SOURCE; crm_xml_add(xml_op, name, crm_meta_value(op->params, name)); name = PCMK__META_MIGRATE_TARGET; crm_xml_add(xml_op, name, crm_meta_value(op->params, name)); } add_op_digest_to_xml(op, xml_op); if (op_id_additional) { free(op_id); op_id = op_id_additional; op_id_additional = NULL; goto again; } if (local_user_data) { free(local_user_data); op->user_data = NULL; } free(magic); free(op_id); free(key); return xml_op; } /*! * \internal * \brief Check whether an action shutdown-locks a resource to a node * * If the PCMK_OPT_SHUTDOWN_LOCK cluster property is set, resources will not be * recovered on a different node if cleanly stopped, and may start only on that * same node. This function checks whether that applies to a given action, so * that the transition graph can be marked appropriately. * * \param[in] action Action to check * * \return true if \p action locks its resource to the action's node, * otherwise false */ bool pcmk__action_locks_rsc_to_node(const pcmk_action_t *action) { // Only resource actions taking place on resource's lock node are locked if ((action == NULL) || (action->rsc == NULL) || !pcmk__same_node(action->node, action->rsc->priv->lock_node)) { return false; } /* During shutdown, only stops are locked (otherwise, another action such as * a demote would cause the controller to clear the lock) */ if (action->node->details->shutdown && (action->task != NULL) && (strcmp(action->task, PCMK_ACTION_STOP) != 0)) { return false; } return true; } /* lowest to highest */ static gint sort_action_id(gconstpointer a, gconstpointer b) { const pcmk__related_action_t *action_wrapper2 = a; const pcmk__related_action_t *action_wrapper1 = b; if (a == NULL) { return 1; } if (b == NULL) { return -1; } if (action_wrapper1->action->id < action_wrapper2->action->id) { return 1; } if (action_wrapper1->action->id > action_wrapper2->action->id) { return -1; } return 0; } /*! * \internal * \brief Remove any duplicate action inputs, merging action flags * * \param[in,out] action Action whose inputs should be checked */ void pcmk__deduplicate_action_inputs(pcmk_action_t *action) { GList *item = NULL; GList *next = NULL; pcmk__related_action_t *last_input = NULL; action->actions_before = g_list_sort(action->actions_before, sort_action_id); for (item = action->actions_before; item != NULL; item = next) { pcmk__related_action_t *input = item->data; next = item->next; if ((last_input != NULL) && (input->action->id == last_input->action->id)) { crm_trace("Input %s (%d) duplicate skipped for action %s (%d)", input->action->uuid, input->action->id, action->uuid, action->id); /* For the purposes of scheduling, the ordering flags no longer * matter, but crm_simulate looks at certain ones when creating a * dot graph. Combining the flags is sufficient for that purpose. */ pcmk__set_relation_flags(last_input->flags, input->flags); if (input->graphed) { last_input->graphed = true; } free(item->data); action->actions_before = g_list_delete_link(action->actions_before, item); } else { last_input = input; input->graphed = false; } } } /*! * \internal * \brief Output all scheduled actions * * \param[in,out] scheduler Scheduler data */ void pcmk__output_actions(pcmk_scheduler_t *scheduler) { pcmk__output_t *out = scheduler->priv->out; // Output node (non-resource) actions for (GList *iter = scheduler->priv->actions; iter != NULL; iter = iter->next) { char *node_name = NULL; char *task = NULL; pcmk_action_t *action = (pcmk_action_t *) iter->data; if (action->rsc != NULL) { continue; // Resource actions will be output later } else if (pcmk_is_set(action->flags, pcmk__action_optional)) { continue; // This action was not scheduled } if (pcmk__str_eq(action->task, PCMK_ACTION_DO_SHUTDOWN, pcmk__str_none)) { task = strdup("Shutdown"); } else if (pcmk__str_eq(action->task, PCMK_ACTION_STONITH, pcmk__str_none)) { const char *op = g_hash_table_lookup(action->meta, PCMK__META_STONITH_ACTION); task = crm_strdup_printf("Fence (%s)", op); } else { continue; // Don't display other node action types } if (pcmk__is_guest_or_bundle_node(action->node)) { const pcmk_resource_t *remote = action->node->priv->remote; node_name = crm_strdup_printf("%s (resource: %s)", pcmk__node_name(action->node), remote->priv->launcher->id); } else if (action->node != NULL) { node_name = crm_strdup_printf("%s", pcmk__node_name(action->node)); } out->message(out, "node-action", task, node_name, action->reason); free(node_name); free(task); } // Output resource actions for (GList *iter = scheduler->priv->resources; iter != NULL; iter = iter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) iter->data; rsc->priv->cmds->output_actions(rsc); } } /*! * \internal * \brief Get action name needed to compare digest for configuration changes * * \param[in] task Action name from history * \param[in] interval_ms Action interval (in milliseconds) * * \return Action name whose digest should be compared */ static const char * task_for_digest(const char *task, guint interval_ms) { /* Certain actions need to be compared against the parameters used to start * the resource. */ if ((interval_ms == 0) && pcmk__str_any_of(task, PCMK_ACTION_MONITOR, PCMK_ACTION_MIGRATE_FROM, PCMK_ACTION_PROMOTE, NULL)) { task = PCMK_ACTION_START; } return task; } /*! * \internal * \brief Check whether only sanitized parameters to an action changed * * When collecting CIB files for troubleshooting, crm_report will mask * sensitive resource parameters. If simulations were run using that, affected * resources would appear to need a restart, which would complicate * troubleshooting. To avoid that, we save a "secure digest" of non-sensitive * parameters. This function used that digest to check whether only masked * parameters are different. * * \param[in] xml_op Resource history entry with secure digest * \param[in] digest_data Operation digest information being compared * \param[in] scheduler Scheduler data * * \return true if only sanitized parameters changed, otherwise false */ static bool only_sanitized_changed(const xmlNode *xml_op, const pcmk__op_digest_t *digest_data, const pcmk_scheduler_t *scheduler) { const char *digest_secure = NULL; if (!pcmk_is_set(scheduler->flags, pcmk__sched_sanitized)) { // The scheduler is not being run as a simulation return false; } digest_secure = crm_element_value(xml_op, PCMK__XA_OP_SECURE_DIGEST); return (digest_data->rc != pcmk__digest_match) && (digest_secure != NULL) && (digest_data->digest_secure_calc != NULL) && (strcmp(digest_data->digest_secure_calc, digest_secure) == 0); } /*! * \internal * \brief Force a restart due to a configuration change * * \param[in,out] rsc Resource that action is for * \param[in] task Name of action whose configuration changed * \param[in] interval_ms Action interval (in milliseconds) * \param[in,out] node Node where resource should be restarted */ static void force_restart(pcmk_resource_t *rsc, const char *task, guint interval_ms, pcmk_node_t *node) { char *key = pcmk__op_key(rsc->id, task, interval_ms); pcmk_action_t *required = custom_action(rsc, key, task, NULL, FALSE, rsc->priv->scheduler); pe_action_set_reason(required, "resource definition change", true); trigger_unfencing(rsc, node, "Device parameters changed", NULL, rsc->priv->scheduler); } /*! * \internal * \brief Schedule a reload of a resource on a node * * \param[in,out] data Resource to reload * \param[in] user_data Where resource should be reloaded */ static void schedule_reload(gpointer data, gpointer user_data) { pcmk_resource_t *rsc = data; const pcmk_node_t *node = user_data; pcmk_action_t *reload = NULL; // For collective resources, just call recursively for children if (rsc->priv->variant > pcmk__rsc_variant_primitive) { g_list_foreach(rsc->priv->children, schedule_reload, user_data); return; } // Skip the reload in certain situations if ((node == NULL) || !pcmk_is_set(rsc->flags, pcmk__rsc_managed) || pcmk_is_set(rsc->flags, pcmk__rsc_failed)) { pcmk__rsc_trace(rsc, "Skip reload of %s:%s%s %s", rsc->id, pcmk_is_set(rsc->flags, pcmk__rsc_managed)? "" : " unmanaged", pcmk_is_set(rsc->flags, pcmk__rsc_failed)? " failed" : "", (node == NULL)? "inactive" : node->priv->name); return; } /* If a resource's configuration changed while a start was pending, * force a full restart instead of a reload. */ if (pcmk_is_set(rsc->flags, pcmk__rsc_start_pending)) { pcmk__rsc_trace(rsc, "%s: preventing agent reload because start pending", rsc->id); custom_action(rsc, stop_key(rsc), PCMK_ACTION_STOP, node, FALSE, rsc->priv->scheduler); return; } // Schedule the reload pcmk__set_rsc_flags(rsc, pcmk__rsc_reload); reload = custom_action(rsc, reload_key(rsc), PCMK_ACTION_RELOAD_AGENT, node, FALSE, rsc->priv->scheduler); pe_action_set_reason(reload, "resource definition change", FALSE); // Set orderings so that a required stop or demote cancels the reload pcmk__new_ordering(NULL, NULL, reload, rsc, stop_key(rsc), NULL, pcmk__ar_ordered|pcmk__ar_then_cancels_first, rsc->priv->scheduler); pcmk__new_ordering(NULL, NULL, reload, rsc, demote_key(rsc), NULL, pcmk__ar_ordered|pcmk__ar_then_cancels_first, rsc->priv->scheduler); } /*! * \internal * \brief Handle any configuration change for an action * * Given an action from resource history, if the resource's configuration * changed since the action was done, schedule any actions needed (restart, * reload, unfencing, rescheduling recurring actions, etc.). * * \param[in,out] rsc Resource that action is for * \param[in,out] node Node that action was on * \param[in] xml_op Action XML from resource history * * \return true if action configuration changed, otherwise false */ bool pcmk__check_action_config(pcmk_resource_t *rsc, pcmk_node_t *node, const xmlNode *xml_op) { guint interval_ms = 0; const char *task = NULL; const pcmk__op_digest_t *digest_data = NULL; CRM_CHECK((rsc != NULL) && (node != NULL) && (xml_op != NULL), return false); task = crm_element_value(xml_op, PCMK_XA_OPERATION); CRM_CHECK(task != NULL, return false); crm_element_value_ms(xml_op, PCMK_META_INTERVAL, &interval_ms); // If this is a recurring action, check whether it has been orphaned if (interval_ms > 0) { if (pcmk__find_action_config(rsc, task, interval_ms, false) != NULL) { pcmk__rsc_trace(rsc, "%s-interval %s for %s on %s is in configuration", pcmk__readable_interval(interval_ms), task, rsc->id, pcmk__node_name(node)); } else if (pcmk_is_set(rsc->priv->scheduler->flags, pcmk__sched_cancel_removed_actions)) { pcmk__schedule_cancel(rsc, crm_element_value(xml_op, PCMK__XA_CALL_ID), task, interval_ms, node, "orphan"); return true; } else { pcmk__rsc_debug(rsc, "%s-interval %s for %s on %s is orphaned", pcmk__readable_interval(interval_ms), task, rsc->id, pcmk__node_name(node)); return true; } } crm_trace("Checking %s-interval %s for %s on %s for configuration changes", pcmk__readable_interval(interval_ms), task, rsc->id, pcmk__node_name(node)); task = task_for_digest(task, interval_ms); digest_data = rsc_action_digest_cmp(rsc, xml_op, node, rsc->priv->scheduler); if (only_sanitized_changed(xml_op, digest_data, rsc->priv->scheduler)) { if (!pcmk__is_daemon && (rsc->priv->scheduler->priv->out != NULL)) { pcmk__output_t *out = rsc->priv->scheduler->priv->out; out->info(out, "Only 'private' parameters to %s-interval %s for %s " "on %s changed: %s", pcmk__readable_interval(interval_ms), task, rsc->id, pcmk__node_name(node), crm_element_value(xml_op, PCMK__XA_TRANSITION_MAGIC)); } return false; } switch (digest_data->rc) { case pcmk__digest_restart: crm_log_xml_debug(digest_data->params_restart, "params:restart"); force_restart(rsc, task, interval_ms, node); return true; case pcmk__digest_unknown: case pcmk__digest_mismatch: // Changes that can potentially be handled by an agent reload if (interval_ms > 0) { /* Recurring actions aren't reloaded per se, they are just * re-scheduled so the next run uses the new parameters. * The old instance will be cancelled automatically. */ crm_log_xml_debug(digest_data->params_all, "params:reschedule"); pcmk__reschedule_recurring(rsc, task, interval_ms, node); } else if (crm_element_value(xml_op, PCMK__XA_OP_RESTART_DIGEST) != NULL) { // Agent supports reload, so use it trigger_unfencing(rsc, node, "Device parameters changed (reload)", NULL, rsc->priv->scheduler); crm_log_xml_debug(digest_data->params_all, "params:reload"); schedule_reload((gpointer) rsc, (gpointer) node); } else { pcmk__rsc_trace(rsc, "Restarting %s " "because agent doesn't support reload", rsc->id); crm_log_xml_debug(digest_data->params_restart, "params:restart"); force_restart(rsc, task, interval_ms, node); } return true; default: break; } return false; } /*! * \internal * \brief Create a list of resource's action history entries, sorted by call ID * * \param[in] rsc_entry Resource's \c PCMK__XE_LRM_RSC_OP status XML * \param[out] start_index Where to store index of start-like action, if any * \param[out] stop_index Where to store index of stop action, if any */ static GList * rsc_history_as_list(const xmlNode *rsc_entry, int *start_index, int *stop_index) { GList *ops = NULL; for (xmlNode *rsc_op = pcmk__xe_first_child(rsc_entry, PCMK__XE_LRM_RSC_OP, NULL, NULL); rsc_op != NULL; rsc_op = pcmk__xe_next(rsc_op, PCMK__XE_LRM_RSC_OP)) { ops = g_list_prepend(ops, rsc_op); } ops = g_list_sort(ops, sort_op_by_callid); calculate_active_ops(ops, start_index, stop_index); return ops; } /*! * \internal * \brief Process a resource's action history from the CIB status * * Given a resource's action history, if the resource's configuration * changed since the actions were done, schedule any actions needed (restart, * reload, unfencing, rescheduling recurring actions, clean-up, etc.). * (This also cancels recurring actions for maintenance mode, which is not * entirely related but convenient to do here.) * * \param[in] rsc_entry Resource's \c PCMK__XE_LRM_RSC_OP status XML * \param[in,out] rsc Resource whose history is being processed * \param[in,out] node Node whose history is being processed */ static void process_rsc_history(const xmlNode *rsc_entry, pcmk_resource_t *rsc, pcmk_node_t *node) { int offset = -1; int stop_index = 0; int start_index = 0; GList *sorted_op_list = NULL; if (pcmk_is_set(rsc->flags, pcmk__rsc_removed)) { if (pcmk__is_anonymous_clone(pe__const_top_resource(rsc, false))) { pcmk__rsc_trace(rsc, "Skipping configuration check " "for orphaned clone instance %s", rsc->id); } else { pcmk__rsc_trace(rsc, "Skipping configuration check and scheduling " "clean-up for orphaned resource %s", rsc->id); pcmk__schedule_cleanup(rsc, node, false); } return; } if (pe_find_node_id(rsc->priv->active_nodes, node->priv->id) == NULL) { if (pcmk__rsc_agent_changed(rsc, node, rsc_entry, false)) { pcmk__schedule_cleanup(rsc, node, false); } pcmk__rsc_trace(rsc, "Skipping configuration check for %s " "because no longer active on %s", rsc->id, pcmk__node_name(node)); return; } pcmk__rsc_trace(rsc, "Checking for configuration changes for %s on %s", rsc->id, pcmk__node_name(node)); if (pcmk__rsc_agent_changed(rsc, node, rsc_entry, true)) { pcmk__schedule_cleanup(rsc, node, false); } sorted_op_list = rsc_history_as_list(rsc_entry, &start_index, &stop_index); if (start_index < stop_index) { return; // Resource is stopped } for (GList *iter = sorted_op_list; iter != NULL; iter = iter->next) { xmlNode *rsc_op = (xmlNode *) iter->data; const char *task = NULL; guint interval_ms = 0; if (++offset < start_index) { // Skip actions that happened before a start continue; } task = crm_element_value(rsc_op, PCMK_XA_OPERATION); crm_element_value_ms(rsc_op, PCMK_META_INTERVAL, &interval_ms); if ((interval_ms > 0) && (pcmk_is_set(rsc->flags, pcmk__rsc_maintenance) || node->details->maintenance)) { // Maintenance mode cancels recurring operations pcmk__schedule_cancel(rsc, crm_element_value(rsc_op, PCMK__XA_CALL_ID), task, interval_ms, node, "maintenance mode"); } else if ((interval_ms > 0) || pcmk__strcase_any_of(task, PCMK_ACTION_MONITOR, PCMK_ACTION_START, PCMK_ACTION_PROMOTE, PCMK_ACTION_MIGRATE_FROM, NULL)) { /* If a resource operation failed, and the operation's definition * has changed, clear any fail count so they can be retried fresh. */ if (pe__bundle_needs_remote_name(rsc)) { /* We haven't assigned resources to nodes yet, so if the * REMOTE_CONTAINER_HACK is used, we may calculate the digest * based on the literal "#uname" value rather than the properly * substituted value. That would mistakenly make the action * definition appear to have been changed. Defer the check until * later in this case. */ - pe__add_param_check(rsc_op, rsc, node, pcmk__check_active, - rsc->priv->scheduler); + pcmk__add_param_check(rsc_op, rsc, node, pcmk__check_active, + rsc->priv->scheduler); } else if (pcmk__check_action_config(rsc, node, rsc_op) && (pe_get_failcount(node, rsc, NULL, pcmk__fc_effective, NULL) != 0)) { pe__clear_failcount(rsc, node, "action definition changed", rsc->priv->scheduler); } } } g_list_free(sorted_op_list); } /*! * \internal * \brief Process a node's action history from the CIB status * * Given a node's resource history, if the resource's configuration changed * since the actions were done, schedule any actions needed (restart, * reload, unfencing, rescheduling recurring actions, clean-up, etc.). * (This also cancels recurring actions for maintenance mode, which is not * entirely related but convenient to do here.) * * \param[in,out] node Node whose history is being processed * \param[in] lrm_rscs Node's \c PCMK__XE_LRM_RESOURCES from CIB status XML */ static void process_node_history(pcmk_node_t *node, const xmlNode *lrm_rscs) { crm_trace("Processing node history for %s", pcmk__node_name(node)); for (const xmlNode *rsc_entry = pcmk__xe_first_child(lrm_rscs, PCMK__XE_LRM_RESOURCE, NULL, NULL); rsc_entry != NULL; rsc_entry = pcmk__xe_next(rsc_entry, PCMK__XE_LRM_RESOURCE)) { if (rsc_entry->children != NULL) { GList *result = pcmk__rscs_matching_id(pcmk__xe_id(rsc_entry), node->priv->scheduler); for (GList *iter = result; iter != NULL; iter = iter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) iter->data; if (pcmk__is_primitive(rsc)) { process_rsc_history(rsc_entry, rsc, node); } } g_list_free(result); } } } // XPath to find a node's resource history #define XPATH_NODE_HISTORY "/" PCMK_XE_CIB "/" PCMK_XE_STATUS \ "/" PCMK__XE_NODE_STATE \ "[@" PCMK_XA_UNAME "='%s']" \ "/" PCMK__XE_LRM "/" PCMK__XE_LRM_RESOURCES /*! * \internal * \brief Process any resource configuration changes in the CIB status * * Go through all nodes' resource history, and if a resource's configuration * changed since its actions were done, schedule any actions needed (restart, * reload, unfencing, rescheduling recurring actions, clean-up, etc.). * (This also cancels recurring actions for maintenance mode, which is not * entirely related but convenient to do here.) * * \param[in,out] scheduler Scheduler data */ void pcmk__handle_rsc_config_changes(pcmk_scheduler_t *scheduler) { crm_trace("Check resource and action configuration for changes"); /* Rather than iterate through the status section, iterate through the nodes * and search for the appropriate status subsection for each. This skips * orphaned nodes and lets us eliminate some cases before searching the XML. */ for (GList *iter = scheduler->nodes; iter != NULL; iter = iter->next) { pcmk_node_t *node = (pcmk_node_t *) iter->data; /* Don't bother checking actions for a node that can't run actions ... * unless it's in maintenance mode, in which case we still need to * cancel any existing recurring monitors. */ if (node->details->maintenance || pcmk__node_available(node, false, false)) { char *xpath = NULL; xmlNode *history = NULL; xpath = crm_strdup_printf(XPATH_NODE_HISTORY, node->priv->name); history = get_xpath_object(xpath, scheduler->input, LOG_NEVER); free(xpath); process_node_history(node, history); } } } diff --git a/lib/pacemaker/pcmk_sched_location.c b/lib/pacemaker/pcmk_sched_location.c index a3915b5435..fe8513dc92 100644 --- a/lib/pacemaker/pcmk_sched_location.c +++ b/lib/pacemaker/pcmk_sched_location.c @@ -1,781 +1,781 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include "libpacemaker_private.h" /*! * \internal * \brief Parse a role configuration for a location constraint * * \param[in] role_spec Role specification * \param[out] role Where to store parsed role * * \return true if role specification is valid, otherwise false */ static bool parse_location_role(const char *role_spec, enum rsc_role_e *role) { if (role_spec == NULL) { *role = pcmk_role_unknown; return true; } *role = pcmk_parse_role(role_spec); switch (*role) { case pcmk_role_unknown: return false; case pcmk_role_started: case pcmk_role_unpromoted: /* Any promotable clone instance cannot be promoted without being in * the unpromoted role first. Therefore, any constraint for the * started or unpromoted role applies to every role. */ *role = pcmk_role_unknown; break; default: break; } return true; } /*! * \internal * \brief Get the score attribute name (if any) used for a rule * * \param[in] rule_xml Rule XML * \param[out] allocated If the score attribute name needs to be allocated, * this will be set to the non-const equivalent of the * return value (should be set to NULL when passed) * \param[in] rule_input Values used to evaluate rule criteria * * \return Score attribute name used for rule, or NULL if none * \note The caller is responsible for freeing \p *allocated if it is non-NULL. */ static const char * score_attribute_name(const xmlNode *rule_xml, char **allocated, const pcmk_rule_input_t *rule_input) { const char *name = NULL; name = crm_element_value(rule_xml, PCMK_XA_SCORE_ATTRIBUTE); if (name == NULL) { return NULL; } /* A score attribute name may use submatches extracted from a * resource ID regular expression. For example, if score-attribute is * "loc-\1", rsc-pattern is "ip-(.*)", and the resource ID is "ip-db", then * the score attribute name is "loc-db". */ if ((rule_input->rsc_id != NULL) && (rule_input->rsc_id_nmatches > 0)) { *allocated = pcmk__replace_submatches(name, rule_input->rsc_id, rule_input->rsc_id_submatches, rule_input->rsc_id_nmatches); if (*allocated != NULL) { name = *allocated; } } return name; } /*! * \internal * \brief Parse a score from a rule without a score attribute * * \param[in] rule_xml Rule XML * \param[out] score Where to store parsed score * * \return Standard Pacemaker return code */ static int score_from_rule(const xmlNode *rule_xml, int *score) { int rc = pcmk_rc_ok; const char *score_s = crm_element_value(rule_xml, PCMK_XA_SCORE); if (score_s == NULL) { // Not possible with schema validation enabled pcmk__config_err("Ignoring location constraint rule %s because " "neither " PCMK_XA_SCORE " nor " PCMK_XA_SCORE_ATTRIBUTE " was specified", pcmk__xe_id(rule_xml)); return pcmk_rc_unpack_error; } rc = pcmk_parse_score(score_s, score, 0); if (rc != pcmk_rc_ok) { // Not possible with schema validation enabled pcmk__config_err("Ignoring location constraint rule %s because " "'%s' is not a valid " PCMK_XA_SCORE ": %s", pcmk__xe_id(rule_xml), score_s, pcmk_rc_str(rc)); return pcmk_rc_unpack_error; } return pcmk_rc_ok; } /*! * \internal * \brief Get a rule score from a node attribute * * \param[in] constraint_id Location constraint ID (for logging only) * \param[in] attr_name Name of node attribute with score * \param[in] node Node to get attribute for * \param[in] rsc Resource being located * \param[out] score Where to store parsed score * * \return Standard Pacemaker return code (pcmk_rc_ok if a valid score was * parsed, ENXIO if the node attribute was unset, and some other value * if the node attribute value was invalid) */ static int score_from_attr(const char *constraint_id, const char *attr_name, const pcmk_node_t *node, const pcmk_resource_t *rsc, int *score) { int rc = pcmk_rc_ok; const char *target = NULL; const char *score_s = NULL; target = g_hash_table_lookup(rsc->priv->meta, PCMK_META_CONTAINER_ATTRIBUTE_TARGET); score_s = pcmk__node_attr(node, attr_name, target, pcmk__rsc_node_current); if (pcmk__str_empty(score_s)) { crm_info("Ignoring location %s for %s on %s " "because it has no node attribute %s", constraint_id, rsc->id, pcmk__node_name(node), attr_name); return ENXIO; } rc = pcmk_parse_score(score_s, score, 0); if (rc != pcmk_rc_ok) { crm_warn("Ignoring location %s for node %s because node " "attribute %s value '%s' is not a valid score: %s", constraint_id, pcmk__node_name(node), attr_name, score_s, pcmk_rc_str(rc)); return rc; } return pcmk_rc_ok; } /*! * \internal * \brief Generate a location constraint from a rule * * \param[in,out] rsc Resource that constraint is for * \param[in] rule_xml Rule XML (sub-element of location constraint) * \param[in] discovery Value of \c PCMK_XA_RESOURCE_DISCOVERY for * constraint * \param[out] next_change Where to set when rule evaluation will change * \param[in,out] rule_input Values used to evaluate rule criteria * (node-specific values will be overwritten by * this function) * \param[in] constraint_id ID of location constraint (for logging only) * * \return true if rule is valid, otherwise false */ static bool generate_location_rule(pcmk_resource_t *rsc, xmlNode *rule_xml, const char *discovery, crm_time_t *next_change, pcmk_rule_input_t *rule_input, const char *constraint_id) { const char *rule_id = NULL; const char *score_attr = NULL; const char *boolean = NULL; const char *role_spec = NULL; GList *iter = NULL; int score = 0; char *local_score_attr = NULL; pcmk__location_t *location_rule = NULL; enum rsc_role_e role = pcmk_role_unknown; enum pcmk__combine combine = pcmk__combine_unknown; rule_xml = pcmk__xe_resolve_idref(rule_xml, rsc->priv->scheduler->input); if (rule_xml == NULL) { return false; // Error already logged } rule_id = crm_element_value(rule_xml, PCMK_XA_ID); if (rule_id == NULL) { pcmk__config_err("Ignoring location constraint '%s' because its rule " "has no " PCMK_XA_ID, constraint_id); return false; } boolean = crm_element_value(rule_xml, PCMK_XA_BOOLEAN_OP); role_spec = crm_element_value(rule_xml, PCMK_XA_ROLE); if (parse_location_role(role_spec, &role)) { crm_trace("Setting rule %s role filter to %s", rule_id, role_spec); } else { pcmk__config_err("Ignoring location constraint '%s' because rule '%s' " "has invalid " PCMK_XA_ROLE " '%s'", constraint_id, rule_id, role_spec); return false; } combine = pcmk__parse_combine(boolean); switch (combine) { case pcmk__combine_and: case pcmk__combine_or: break; default: // Not possible with schema validation enabled pcmk__config_err("Ignoring location constraint '%s' because rule " "'%s' has invalid " PCMK_XA_BOOLEAN_OP " '%s'", constraint_id, rule_id, boolean); return false; } /* Users may configure the rule with either a score or the name of a * node attribute whose value should be used as the constraint score for * that node. */ score_attr = score_attribute_name(rule_xml, &local_score_attr, rule_input); if ((score_attr == NULL) && (score_from_rule(rule_xml, &score) != pcmk_rc_ok)) { return false; } location_rule = pcmk__new_location(rule_id, rsc, 0, discovery, NULL); CRM_CHECK(location_rule != NULL, return NULL); location_rule->role_filter = role; for (iter = rsc->priv->scheduler->nodes; iter != NULL; iter = iter->next) { pcmk_node_t *node = iter->data; pcmk_node_t *local = NULL; rule_input->node_attrs = node->priv->attrs; rule_input->rsc_params = pe_rsc_params(rsc, node, rsc->priv->scheduler); if (pcmk_evaluate_rule(rule_xml, rule_input, next_change) != pcmk_rc_ok) { continue; } if ((score_attr != NULL) && (score_from_attr(constraint_id, score_attr, node, rsc, &score) != pcmk_rc_ok)) { continue; // Message already logged } local = pe__copy_node(node); location_rule->nodes = g_list_prepend(location_rule->nodes, local); local->assign->score = score; pcmk__rsc_trace(rsc, "Location %s score for %s on %s is %s via rule %s", constraint_id, rsc->id, pcmk__node_name(node), pcmk_readable_score(score), rule_id); } free(local_score_attr); if (location_rule->nodes == NULL) { crm_trace("No matching nodes for location constraint rule %s", rule_id); } else { crm_trace("Location constraint rule %s matched %d nodes", rule_id, g_list_length(location_rule->nodes)); } return true; } static void unpack_rsc_location(xmlNode *xml_obj, pcmk_resource_t *rsc, const char *role_spec, const char *score, char *rsc_id_match, int rsc_id_nmatches, regmatch_t *rsc_id_submatches) { const char *rsc_id = crm_element_value(xml_obj, PCMK_XA_RSC); const char *id = crm_element_value(xml_obj, PCMK_XA_ID); const char *node = crm_element_value(xml_obj, PCMK_XA_NODE); const char *discovery = crm_element_value(xml_obj, PCMK_XA_RESOURCE_DISCOVERY); if (rsc == NULL) { pcmk__config_warn("Ignoring constraint '%s' because resource '%s' " "does not exist", id, rsc_id); return; } if (score == NULL) { score = crm_element_value(xml_obj, PCMK_XA_SCORE); } if ((node != NULL) && (score != NULL)) { int score_i = 0; int rc = pcmk_rc_ok; pcmk_node_t *match = pcmk_find_node(rsc->priv->scheduler, node); enum rsc_role_e role = pcmk_role_unknown; pcmk__location_t *location = NULL; if (match == NULL) { crm_info("Ignoring location constraint %s " "because '%s' is not a known node", pcmk__s(id, "without ID"), node); return; } rc = pcmk_parse_score(score, &score_i, 0); if (rc != pcmk_rc_ok) { // Not possible with schema validation enabled pcmk__config_err("Ignoring location constraint %s " "because '%s' is not a valid score", id, score); return; } if (role_spec == NULL) { role_spec = crm_element_value(xml_obj, PCMK_XA_ROLE); } if (parse_location_role(role_spec, &role)) { crm_trace("Setting location constraint %s role filter: %s", id, role_spec); } else { // Not possible with schema validation enabled pcmk__config_err("Ignoring location constraint %s " "because '%s' is not a valid " PCMK_XA_ROLE, id, role_spec); return; } location = pcmk__new_location(id, rsc, score_i, discovery, match); if (location == NULL) { return; // Error already logged } location->role_filter = role; } else { crm_time_t *next_change = crm_time_new_undefined(); xmlNode *rule_xml = pcmk__xe_first_child(xml_obj, PCMK_XE_RULE, NULL, NULL); pcmk_rule_input_t rule_input = { .now = rsc->priv->scheduler->priv->now, .rsc_meta = rsc->priv->meta, .rsc_id = rsc_id_match, .rsc_id_submatches = rsc_id_submatches, .rsc_id_nmatches = rsc_id_nmatches, }; generate_location_rule(rsc, rule_xml, discovery, next_change, &rule_input, id); /* If there is a point in the future when the evaluation of a rule will * change, make sure the scheduler is re-run by that time. */ if (crm_time_is_defined(next_change)) { time_t t = (time_t) crm_time_get_seconds_since_epoch(next_change); - pe__update_recheck_time(t, rsc->priv->scheduler, - "location rule evaluation"); + pcmk__update_recheck_time(t, rsc->priv->scheduler, + "location rule evaluation"); } crm_time_free(next_change); } } static void unpack_simple_location(xmlNode *xml_obj, pcmk_scheduler_t *scheduler) { const char *id = crm_element_value(xml_obj, PCMK_XA_ID); const char *value = crm_element_value(xml_obj, PCMK_XA_RSC); if (value) { pcmk_resource_t *rsc; rsc = pcmk__find_constraint_resource(scheduler->priv->resources, value); unpack_rsc_location(xml_obj, rsc, NULL, NULL, NULL, 0, NULL); } value = crm_element_value(xml_obj, PCMK_XA_RSC_PATTERN); if (value) { regex_t regex; bool invert = false; if (value[0] == '!') { value++; invert = true; } if (regcomp(®ex, value, REG_EXTENDED) != 0) { pcmk__config_err("Ignoring constraint '%s' because " PCMK_XA_RSC_PATTERN " has invalid value '%s'", id, value); return; } for (GList *iter = scheduler->priv->resources; iter != NULL; iter = iter->next) { pcmk_resource_t *r = iter->data; int nregs = 0; regmatch_t *pmatch = NULL; int status; if (regex.re_nsub > 0) { nregs = regex.re_nsub + 1; } else { nregs = 1; } pmatch = pcmk__assert_alloc(nregs, sizeof(regmatch_t)); status = regexec(®ex, r->id, nregs, pmatch, 0); if (!invert && (status == 0)) { crm_debug("'%s' matched '%s' for %s", r->id, value, id); unpack_rsc_location(xml_obj, r, NULL, NULL, r->id, nregs, pmatch); } else if (invert && (status != 0)) { crm_debug("'%s' is an inverted match of '%s' for %s", r->id, value, id); unpack_rsc_location(xml_obj, r, NULL, NULL, NULL, 0, NULL); } else { crm_trace("'%s' does not match '%s' for %s", r->id, value, id); } free(pmatch); } regfree(®ex); } } // \return Standard Pacemaker return code static int unpack_location_tags(xmlNode *xml_obj, xmlNode **expanded_xml, pcmk_scheduler_t *scheduler) { const char *id = NULL; const char *rsc_id = NULL; const char *state = NULL; pcmk_resource_t *rsc = NULL; pcmk__idref_t *tag = NULL; xmlNode *rsc_set = NULL; *expanded_xml = NULL; CRM_CHECK(xml_obj != NULL, return EINVAL); id = pcmk__xe_id(xml_obj); if (id == NULL) { pcmk__config_err("Ignoring <%s> constraint without " PCMK_XA_ID, xml_obj->name); return pcmk_rc_unpack_error; } // Check whether there are any resource sets with template or tag references *expanded_xml = pcmk__expand_tags_in_sets(xml_obj, scheduler); if (*expanded_xml != NULL) { crm_log_xml_trace(*expanded_xml, "Expanded " PCMK_XE_RSC_LOCATION); return pcmk_rc_ok; } rsc_id = crm_element_value(xml_obj, PCMK_XA_RSC); if (rsc_id == NULL) { return pcmk_rc_ok; } if (!pcmk__valid_resource_or_tag(scheduler, rsc_id, &rsc, &tag)) { pcmk__config_err("Ignoring constraint '%s' because '%s' is not a " "valid resource or tag", id, rsc_id); return pcmk_rc_unpack_error; } else if (rsc != NULL) { // No template is referenced return pcmk_rc_ok; } state = crm_element_value(xml_obj, PCMK_XA_ROLE); *expanded_xml = pcmk__xml_copy(NULL, xml_obj); /* Convert any template or tag reference into constraint * PCMK_XE_RESOURCE_SET */ if (!pcmk__tag_to_set(*expanded_xml, &rsc_set, PCMK_XA_RSC, false, scheduler)) { pcmk__xml_free(*expanded_xml); *expanded_xml = NULL; return pcmk_rc_unpack_error; } if (rsc_set != NULL) { if (state != NULL) { /* Move PCMK_XA_RSC_ROLE into converted PCMK_XE_RESOURCE_SET as * PCMK_XA_ROLE attribute */ crm_xml_add(rsc_set, PCMK_XA_ROLE, state); pcmk__xe_remove_attr(*expanded_xml, PCMK_XA_ROLE); } crm_log_xml_trace(*expanded_xml, "Expanded " PCMK_XE_RSC_LOCATION); } else { // No sets pcmk__xml_free(*expanded_xml); *expanded_xml = NULL; } return pcmk_rc_ok; } // \return Standard Pacemaker return code static int unpack_location_set(xmlNode *location, xmlNode *set, pcmk_scheduler_t *scheduler) { xmlNode *xml_rsc = NULL; pcmk_resource_t *resource = NULL; const char *set_id; const char *role; const char *local_score; CRM_CHECK(set != NULL, return EINVAL); set_id = pcmk__xe_id(set); if (set_id == NULL) { pcmk__config_err("Ignoring " PCMK_XE_RESOURCE_SET " without " PCMK_XA_ID " in constraint '%s'", pcmk__s(pcmk__xe_id(location), "(missing ID)")); return pcmk_rc_unpack_error; } role = crm_element_value(set, PCMK_XA_ROLE); local_score = crm_element_value(set, PCMK_XA_SCORE); for (xml_rsc = pcmk__xe_first_child(set, PCMK_XE_RESOURCE_REF, NULL, NULL); xml_rsc != NULL; xml_rsc = pcmk__xe_next(xml_rsc, PCMK_XE_RESOURCE_REF)) { resource = pcmk__find_constraint_resource(scheduler->priv->resources, pcmk__xe_id(xml_rsc)); if (resource == NULL) { pcmk__config_err("%s: No resource found for %s", set_id, pcmk__xe_id(xml_rsc)); return pcmk_rc_unpack_error; } unpack_rsc_location(location, resource, role, local_score, NULL, 0, NULL); } return pcmk_rc_ok; } void pcmk__unpack_location(xmlNode *xml_obj, pcmk_scheduler_t *scheduler) { xmlNode *set = NULL; bool any_sets = false; xmlNode *orig_xml = NULL; xmlNode *expanded_xml = NULL; if (unpack_location_tags(xml_obj, &expanded_xml, scheduler) != pcmk_rc_ok) { return; } if (expanded_xml) { orig_xml = xml_obj; xml_obj = expanded_xml; } for (set = pcmk__xe_first_child(xml_obj, PCMK_XE_RESOURCE_SET, NULL, NULL); set != NULL; set = pcmk__xe_next(set, PCMK_XE_RESOURCE_SET)) { any_sets = true; set = pcmk__xe_resolve_idref(set, scheduler->input); if ((set == NULL) // Configuration error, message already logged || (unpack_location_set(xml_obj, set, scheduler) != pcmk_rc_ok)) { if (expanded_xml) { pcmk__xml_free(expanded_xml); } return; } } if (expanded_xml) { pcmk__xml_free(expanded_xml); xml_obj = orig_xml; } if (!any_sets) { unpack_simple_location(xml_obj, scheduler); } } /*! * \internal * \brief Add a new location constraint to scheduler data * * \param[in] id XML ID of location constraint * \param[in,out] rsc Resource in location constraint * \param[in] node_score Constraint score * \param[in] probe_mode When resource should be probed on node * \param[in] node Node in constraint (or NULL if rule-based) * * \return Newly allocated location constraint on success, otherwise NULL * \note The result will be added to the cluster (via \p rsc) and should not be * freed separately. */ pcmk__location_t * pcmk__new_location(const char *id, pcmk_resource_t *rsc, int node_score, const char *probe_mode, pcmk_node_t *node) { pcmk__location_t *new_con = NULL; CRM_CHECK((node != NULL) || (node_score == 0), return NULL); if (id == NULL) { pcmk__config_err("Invalid constraint: no ID specified"); return NULL; } if (rsc == NULL) { pcmk__config_err("Invalid constraint %s: no resource specified", id); return NULL; } new_con = pcmk__assert_alloc(1, sizeof(pcmk__location_t)); new_con->id = pcmk__str_copy(id); new_con->rsc = rsc; new_con->nodes = NULL; new_con->role_filter = pcmk_role_unknown; if (pcmk__str_eq(probe_mode, PCMK_VALUE_ALWAYS, pcmk__str_null_matches|pcmk__str_casei)) { new_con->probe_mode = pcmk__probe_always; } else if (pcmk__str_eq(probe_mode, PCMK_VALUE_NEVER, pcmk__str_casei)) { new_con->probe_mode = pcmk__probe_never; } else if (pcmk__str_eq(probe_mode, PCMK_VALUE_EXCLUSIVE, pcmk__str_casei)) { new_con->probe_mode = pcmk__probe_exclusive; pcmk__set_rsc_flags(rsc, pcmk__rsc_exclusive_probes); } else { pcmk__config_err("Invalid " PCMK_XA_RESOURCE_DISCOVERY " value %s " "in location constraint", probe_mode); } if (node != NULL) { pcmk_node_t *copy = pe__copy_node(node); copy->assign->score = node_score; new_con->nodes = g_list_prepend(NULL, copy); } rsc->priv->scheduler->priv->location_constraints = g_list_prepend(rsc->priv->scheduler->priv->location_constraints, new_con); rsc->priv->location_constraints = g_list_prepend(rsc->priv->location_constraints, new_con); return new_con; } /*! * \internal * \brief Apply all location constraints * * \param[in,out] scheduler Scheduler data */ void pcmk__apply_locations(pcmk_scheduler_t *scheduler) { for (GList *iter = scheduler->priv->location_constraints; iter != NULL; iter = iter->next) { pcmk__location_t *location = iter->data; location->rsc->priv->cmds->apply_location(location->rsc, location); } } /*! * \internal * \brief Apply a location constraint to a resource's allowed node scores * * \param[in,out] rsc Resource to apply constraint to * \param[in,out] location Location constraint to apply * * \note This does not consider the resource's children, so the resource's * apply_location() method should be used instead in most cases. */ void pcmk__apply_location(pcmk_resource_t *rsc, pcmk__location_t *location) { bool need_role = false; pcmk__assert((rsc != NULL) && (location != NULL)); // If a role was specified, ensure constraint is applicable need_role = (location->role_filter > pcmk_role_unknown); if (need_role && (location->role_filter != rsc->priv->next_role)) { pcmk__rsc_trace(rsc, "Not applying %s to %s because role will be %s not %s", location->id, rsc->id, pcmk_role_text(rsc->priv->next_role), pcmk_role_text(location->role_filter)); return; } if (location->nodes == NULL) { pcmk__rsc_trace(rsc, "Not applying %s to %s because no nodes match", location->id, rsc->id); return; } for (GList *iter = location->nodes; iter != NULL; iter = iter->next) { pcmk_node_t *node = iter->data; pcmk_node_t *allowed_node = NULL; allowed_node = g_hash_table_lookup(rsc->priv->allowed_nodes, node->priv->id); pcmk__rsc_trace(rsc, "Applying %s%s%s to %s score on %s: %c %s", location->id, (need_role? " for role " : ""), (need_role? pcmk_role_text(location->role_filter) : ""), rsc->id, pcmk__node_name(node), ((allowed_node == NULL)? '=' : '+'), pcmk_readable_score(node->assign->score)); if (allowed_node == NULL) { allowed_node = pe__copy_node(node); g_hash_table_insert(rsc->priv->allowed_nodes, (gpointer) allowed_node->priv->id, allowed_node); } else { allowed_node->assign->score = pcmk__add_scores(allowed_node->assign->score, node->assign->score); } if (allowed_node->assign->probe_mode < location->probe_mode) { if (location->probe_mode == pcmk__probe_exclusive) { pcmk__set_rsc_flags(rsc, pcmk__rsc_exclusive_probes); } /* exclusive > never > always... always is default */ allowed_node->assign->probe_mode = location->probe_mode; } } } diff --git a/lib/pacemaker/pcmk_sched_primitive.c b/lib/pacemaker/pcmk_sched_primitive.c index ad68c6b49f..b1c112abed 100644 --- a/lib/pacemaker/pcmk_sched_primitive.c +++ b/lib/pacemaker/pcmk_sched_primitive.c @@ -1,1716 +1,1719 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include // uint8_t, uint32_t #include #include #include "libpacemaker_private.h" static void stop_resource(pcmk_resource_t *rsc, pcmk_node_t *node, bool optional); static void start_resource(pcmk_resource_t *rsc, pcmk_node_t *node, bool optional); static void demote_resource(pcmk_resource_t *rsc, pcmk_node_t *node, bool optional); static void promote_resource(pcmk_resource_t *rsc, pcmk_node_t *node, bool optional); static void assert_role_error(pcmk_resource_t *rsc, pcmk_node_t *node, bool optional); #define RSC_ROLE_MAX (pcmk_role_promoted + 1) static enum rsc_role_e rsc_state_matrix[RSC_ROLE_MAX][RSC_ROLE_MAX] = { /* This array lists the immediate next role when transitioning from one role * to a target role. For example, when going from Stopped to Promoted, the * next role is Unpromoted, because the resource must be started before it * can be promoted. The current state then becomes Started, which is fed * into this array again, giving a next role of Promoted. * * Current role Immediate next role Final target role * ------------ ------------------- ----------------- */ /* Unknown */ { pcmk_role_unknown, /* Unknown */ pcmk_role_stopped, /* Stopped */ pcmk_role_stopped, /* Started */ pcmk_role_stopped, /* Unpromoted */ pcmk_role_stopped, /* Promoted */ }, /* Stopped */ { pcmk_role_stopped, /* Unknown */ pcmk_role_stopped, /* Stopped */ pcmk_role_started, /* Started */ pcmk_role_unpromoted, /* Unpromoted */ pcmk_role_unpromoted, /* Promoted */ }, /* Started */ { pcmk_role_stopped, /* Unknown */ pcmk_role_stopped, /* Stopped */ pcmk_role_started, /* Started */ pcmk_role_unpromoted, /* Unpromoted */ pcmk_role_promoted, /* Promoted */ }, /* Unpromoted */ { pcmk_role_stopped, /* Unknown */ pcmk_role_stopped, /* Stopped */ pcmk_role_stopped, /* Started */ pcmk_role_unpromoted, /* Unpromoted */ pcmk_role_promoted, /* Promoted */ }, /* Promoted */ { pcmk_role_stopped, /* Unknown */ pcmk_role_unpromoted, /* Stopped */ pcmk_role_unpromoted, /* Started */ pcmk_role_unpromoted, /* Unpromoted */ pcmk_role_promoted, /* Promoted */ }, }; /*! * \internal * \brief Function to schedule actions needed for a role change * * \param[in,out] rsc Resource whose role is changing * \param[in,out] node Node where resource will be in its next role * \param[in] optional Whether scheduled actions should be optional */ typedef void (*rsc_transition_fn)(pcmk_resource_t *rsc, pcmk_node_t *node, bool optional); static rsc_transition_fn rsc_action_matrix[RSC_ROLE_MAX][RSC_ROLE_MAX] = { /* This array lists the function needed to transition directly from one role * to another. NULL indicates that nothing is needed. * * Current role Transition function Next role * ------------ ------------------- ---------- */ /* Unknown */ { assert_role_error, /* Unknown */ stop_resource, /* Stopped */ assert_role_error, /* Started */ assert_role_error, /* Unpromoted */ assert_role_error, /* Promoted */ }, /* Stopped */ { assert_role_error, /* Unknown */ NULL, /* Stopped */ start_resource, /* Started */ start_resource, /* Unpromoted */ assert_role_error, /* Promoted */ }, /* Started */ { assert_role_error, /* Unknown */ stop_resource, /* Stopped */ NULL, /* Started */ NULL, /* Unpromoted */ promote_resource, /* Promoted */ }, /* Unpromoted */ { assert_role_error, /* Unknown */ stop_resource, /* Stopped */ stop_resource, /* Started */ NULL, /* Unpromoted */ promote_resource, /* Promoted */ }, /* Promoted */ { assert_role_error, /* Unknown */ demote_resource, /* Stopped */ demote_resource, /* Started */ demote_resource, /* Unpromoted */ NULL, /* Promoted */ }, }; /*! * \internal * \brief Get a list of a resource's allowed nodes sorted by node score * * \param[in] rsc Resource to check * * \return List of allowed nodes sorted by node score */ static GList * sorted_allowed_nodes(const pcmk_resource_t *rsc) { if (rsc->priv->allowed_nodes != NULL) { GList *nodes = g_hash_table_get_values(rsc->priv->allowed_nodes); if (nodes != NULL) { return pcmk__sort_nodes(nodes, pcmk__current_node(rsc)); } } return NULL; } /*! * \internal * \brief Assign a resource to its best allowed node, if possible * * \param[in,out] rsc Resource to choose a node for * \param[in] prefer If not \c NULL, prefer this node when all else * equal * \param[in] stop_if_fail If \c true and \p rsc can't be assigned to a * node, set next role to stopped and update * existing actions * * \return true if \p rsc could be assigned to a node, otherwise false * * \note If \p stop_if_fail is \c false, then \c pcmk__unassign_resource() can * completely undo the assignment. A successful assignment can be either * undone or left alone as final. A failed assignment has the same effect * as calling pcmk__unassign_resource(); there are no side effects on * roles or actions. */ static bool assign_best_node(pcmk_resource_t *rsc, const pcmk_node_t *prefer, bool stop_if_fail) { GList *nodes = NULL; pcmk_node_t *chosen = NULL; pcmk_node_t *best = NULL; const pcmk_node_t *most_free_node = pcmk__ban_insufficient_capacity(rsc); if (prefer == NULL) { prefer = most_free_node; } if (!pcmk_is_set(rsc->flags, pcmk__rsc_unassigned)) { // We've already finished assignment of resources to nodes return rsc->priv->assigned_node != NULL; } // Sort allowed nodes by score nodes = sorted_allowed_nodes(rsc); if (nodes != NULL) { best = (pcmk_node_t *) nodes->data; // First node has best score } if ((prefer != NULL) && (nodes != NULL)) { // Get the allowed node version of prefer chosen = g_hash_table_lookup(rsc->priv->allowed_nodes, prefer->priv->id); if (chosen == NULL) { pcmk__rsc_trace(rsc, "Preferred node %s for %s was unknown", pcmk__node_name(prefer), rsc->id); /* Favor the preferred node as long as its score is at least as good as * the best allowed node's. * * An alternative would be to favor the preferred node even if the best * node is better, when the best node's score is less than INFINITY. */ } else if (chosen->assign->score < best->assign->score) { pcmk__rsc_trace(rsc, "Preferred node %s for %s was unsuitable", pcmk__node_name(chosen), rsc->id); chosen = NULL; } else if (!pcmk__node_available(chosen, true, false)) { pcmk__rsc_trace(rsc, "Preferred node %s for %s was unavailable", pcmk__node_name(chosen), rsc->id); chosen = NULL; } else { pcmk__rsc_trace(rsc, "Chose preferred node %s for %s " "(ignoring %d candidates)", pcmk__node_name(chosen), rsc->id, g_list_length(nodes)); } } if ((chosen == NULL) && (best != NULL)) { /* Either there is no preferred node, or the preferred node is not * suitable, but another node is allowed to run the resource. */ chosen = best; if (!pcmk__is_unique_clone(rsc->priv->parent) && (chosen->assign->score > 0) // Zero not acceptable && pcmk__node_available(chosen, false, false)) { /* If the resource is already running on a node, prefer that node if * it is just as good as the chosen node. * * We don't do this for unique clone instances, because * pcmk__assign_instances() has already assigned instances to their * running nodes when appropriate, and if we get here, we don't want * remaining unassigned instances to prefer a node that's already * running another instance. */ pcmk_node_t *running = pcmk__current_node(rsc); if (running == NULL) { // Nothing to do } else if (!pcmk__node_available(running, true, false)) { pcmk__rsc_trace(rsc, "Current node for %s (%s) can't run resources", rsc->id, pcmk__node_name(running)); } else { int nodes_with_best_score = 1; for (GList *iter = nodes->next; iter; iter = iter->next) { pcmk_node_t *allowed = (pcmk_node_t *) iter->data; if (allowed->assign->score != chosen->assign->score) { // The nodes are sorted by score, so no more are equal break; } if (pcmk__same_node(allowed, running)) { // Scores are equal, so prefer the current node chosen = allowed; } nodes_with_best_score++; } if (nodes_with_best_score > 1) { uint8_t log_level = LOG_INFO; if (chosen->assign->score >= PCMK_SCORE_INFINITY) { log_level = LOG_WARNING; } do_crm_log(log_level, "Chose %s for %s from %d nodes with score %s", pcmk__node_name(chosen), rsc->id, nodes_with_best_score, pcmk_readable_score(chosen->assign->score)); } } } pcmk__rsc_trace(rsc, "Chose %s for %s from %d candidates", pcmk__node_name(chosen), rsc->id, g_list_length(nodes)); } pcmk__assign_resource(rsc, chosen, false, stop_if_fail); g_list_free(nodes); return rsc->priv->assigned_node != NULL; } /*! * \internal * \brief Apply a "this with" colocation to a node's allowed node scores * * \param[in,out] colocation Colocation to apply * \param[in,out] rsc Resource being assigned */ static void apply_this_with(pcmk__colocation_t *colocation, pcmk_resource_t *rsc) { GHashTable *archive = NULL; pcmk_resource_t *other = colocation->primary; // In certain cases, we will need to revert the node scores if ((colocation->dependent_role >= pcmk_role_promoted) || ((colocation->score < 0) && (colocation->score > -PCMK_SCORE_INFINITY))) { archive = pcmk__copy_node_table(rsc->priv->allowed_nodes); } if (pcmk_is_set(other->flags, pcmk__rsc_unassigned)) { pcmk__rsc_trace(rsc, "%s: Assigning colocation %s primary %s first" "(score=%d role=%s)", rsc->id, colocation->id, other->id, colocation->score, pcmk_role_text(colocation->dependent_role)); other->priv->cmds->assign(other, NULL, true); } // Apply the colocation score to this resource's allowed node scores rsc->priv->cmds->apply_coloc_score(rsc, other, colocation, true); if ((archive != NULL) && !pcmk__any_node_available(rsc->priv->allowed_nodes)) { pcmk__rsc_info(rsc, "%s: Reverting scores from colocation with %s " "because no nodes allowed", rsc->id, other->id); g_hash_table_destroy(rsc->priv->allowed_nodes); rsc->priv->allowed_nodes = archive; archive = NULL; } if (archive != NULL) { g_hash_table_destroy(archive); } } /*! * \internal * \brief Update a Pacemaker Remote node once its connection has been assigned * * \param[in] connection Connection resource that has been assigned */ static void remote_connection_assigned(const pcmk_resource_t *connection) { pcmk_node_t *remote_node = pcmk_find_node(connection->priv->scheduler, connection->id); CRM_CHECK(remote_node != NULL, return); if ((connection->priv->assigned_node != NULL) && (connection->priv->next_role != pcmk_role_stopped)) { crm_trace("Pacemaker Remote node %s will be online", remote_node->priv->id); remote_node->details->online = TRUE; if (!pcmk_is_set(remote_node->priv->flags, pcmk__node_seen)) { // Avoid unnecessary fence, since we will attempt connection remote_node->details->unclean = FALSE; } } else { crm_trace("Pacemaker Remote node %s will be shut down " "(%sassigned connection's next role is %s)", remote_node->priv->id, ((connection->priv->assigned_node == NULL)? "un" : ""), pcmk_role_text(connection->priv->next_role)); remote_node->details->shutdown = TRUE; } } /*! * \internal * \brief Assign a primitive resource to a node * * \param[in,out] rsc Resource to assign to a node * \param[in] prefer Node to prefer, if all else is equal * \param[in] stop_if_fail If \c true and \p rsc can't be assigned to a * node, set next role to stopped and update * existing actions * * \return Node that \p rsc is assigned to, if assigned entirely to one node * * \note If \p stop_if_fail is \c false, then \c pcmk__unassign_resource() can * completely undo the assignment. A successful assignment can be either * undone or left alone as final. A failed assignment has the same effect * as calling pcmk__unassign_resource(); there are no side effects on * roles or actions. */ pcmk_node_t * pcmk__primitive_assign(pcmk_resource_t *rsc, const pcmk_node_t *prefer, bool stop_if_fail) { GList *this_with_colocations = NULL; GList *with_this_colocations = NULL; GList *iter = NULL; pcmk_resource_t *parent = NULL; pcmk__colocation_t *colocation = NULL; pcmk_scheduler_t *scheduler = NULL; pcmk__assert(pcmk__is_primitive(rsc)); scheduler = rsc->priv->scheduler; parent = rsc->priv->parent; // Never assign a child without parent being assigned first if ((parent != NULL) && !pcmk_is_set(parent->flags, pcmk__rsc_assigning)) { pcmk__rsc_debug(rsc, "%s: Assigning parent %s first", rsc->id, parent->id); parent->priv->cmds->assign(parent, prefer, stop_if_fail); } if (!pcmk_is_set(rsc->flags, pcmk__rsc_unassigned)) { // Assignment has already been done const char *node_name = "no node"; if (rsc->priv->assigned_node != NULL) { node_name = pcmk__node_name(rsc->priv->assigned_node); } pcmk__rsc_debug(rsc, "%s: pre-assigned to %s", rsc->id, node_name); return rsc->priv->assigned_node; } // Ensure we detect assignment loops if (pcmk_is_set(rsc->flags, pcmk__rsc_assigning)) { pcmk__rsc_debug(rsc, "Breaking assignment loop involving %s", rsc->id); return NULL; } pcmk__set_rsc_flags(rsc, pcmk__rsc_assigning); pe__show_node_scores(true, rsc, "Pre-assignment", rsc->priv->allowed_nodes, scheduler); this_with_colocations = pcmk__this_with_colocations(rsc); with_this_colocations = pcmk__with_this_colocations(rsc); // Apply mandatory colocations first, to satisfy as many as possible for (iter = this_with_colocations; iter != NULL; iter = iter->next) { colocation = iter->data; if ((colocation->score <= -PCMK_SCORE_INFINITY) || (colocation->score >= PCMK_SCORE_INFINITY)) { apply_this_with(colocation, rsc); } } for (iter = with_this_colocations; iter != NULL; iter = iter->next) { colocation = iter->data; if ((colocation->score <= -PCMK_SCORE_INFINITY) || (colocation->score >= PCMK_SCORE_INFINITY)) { pcmk__add_dependent_scores(colocation, rsc); } } pe__show_node_scores(true, rsc, "Mandatory-colocations", rsc->priv->allowed_nodes, scheduler); // Then apply optional colocations for (iter = this_with_colocations; iter != NULL; iter = iter->next) { colocation = iter->data; if ((colocation->score > -PCMK_SCORE_INFINITY) && (colocation->score < PCMK_SCORE_INFINITY)) { apply_this_with(colocation, rsc); } } for (iter = with_this_colocations; iter != NULL; iter = iter->next) { colocation = iter->data; if ((colocation->score > -PCMK_SCORE_INFINITY) && (colocation->score < PCMK_SCORE_INFINITY)) { pcmk__add_dependent_scores(colocation, rsc); } } g_list_free(this_with_colocations); g_list_free(with_this_colocations); if (rsc->priv->next_role == pcmk_role_stopped) { pcmk__rsc_trace(rsc, "Banning %s from all nodes because it will be stopped", rsc->id); resource_location(rsc, NULL, -PCMK_SCORE_INFINITY, PCMK_META_TARGET_ROLE, scheduler); } else if ((rsc->priv->next_role > rsc->priv->orig_role) && !pcmk_is_set(scheduler->flags, pcmk__sched_quorate) && (scheduler->no_quorum_policy == pcmk_no_quorum_freeze)) { crm_notice("Resource %s cannot be elevated from %s to %s due to " PCMK_OPT_NO_QUORUM_POLICY "=" PCMK_VALUE_FREEZE, rsc->id, pcmk_role_text(rsc->priv->orig_role), pcmk_role_text(rsc->priv->next_role)); pe__set_next_role(rsc, rsc->priv->orig_role, PCMK_OPT_NO_QUORUM_POLICY "=" PCMK_VALUE_FREEZE); } pe__show_node_scores(!pcmk_is_set(scheduler->flags, pcmk__sched_output_scores), rsc, __func__, rsc->priv->allowed_nodes, scheduler); // Unmanage resource if fencing is enabled but no device is configured if (pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled) && !pcmk_is_set(scheduler->flags, pcmk__sched_have_fencing)) { pcmk__clear_rsc_flags(rsc, pcmk__rsc_managed); } if (!pcmk_is_set(rsc->flags, pcmk__rsc_managed)) { // Unmanaged resources stay on their current node const char *reason = NULL; pcmk_node_t *assign_to = NULL; pe__set_next_role(rsc, rsc->priv->orig_role, "unmanaged"); assign_to = pcmk__current_node(rsc); if (assign_to == NULL) { reason = "inactive"; } else if (rsc->priv->orig_role == pcmk_role_promoted) { reason = "promoted"; } else if (pcmk_is_set(rsc->flags, pcmk__rsc_failed)) { reason = "failed"; } else { reason = "active"; } pcmk__rsc_info(rsc, "Unmanaged resource %s assigned to %s: %s", rsc->id, (assign_to? assign_to->priv->name : "no node"), reason); pcmk__assign_resource(rsc, assign_to, true, stop_if_fail); } else if (pcmk_is_set(scheduler->flags, pcmk__sched_stop_all)) { // Must stop at some point, but be consistent with stop_if_fail if (stop_if_fail) { pcmk__rsc_debug(rsc, "Forcing %s to stop: " PCMK_OPT_STOP_ALL_RESOURCES, rsc->id); } pcmk__assign_resource(rsc, NULL, true, stop_if_fail); } else if (!assign_best_node(rsc, prefer, stop_if_fail)) { // Assignment failed if (!pcmk_is_set(rsc->flags, pcmk__rsc_removed)) { pcmk__rsc_info(rsc, "Resource %s cannot run anywhere", rsc->id); } else if ((rsc->priv->active_nodes != NULL) && stop_if_fail) { pcmk__rsc_info(rsc, "Stopping removed resource %s", rsc->id); } } pcmk__clear_rsc_flags(rsc, pcmk__rsc_assigning); if (pcmk_is_set(rsc->flags, pcmk__rsc_is_remote_connection)) { remote_connection_assigned(rsc); } return rsc->priv->assigned_node; } /*! * \internal * \brief Schedule actions to bring resource down and back to current role * * \param[in,out] rsc Resource to restart * \param[in,out] current Node that resource should be brought down on * \param[in] need_stop Whether the resource must be stopped * \param[in] need_promote Whether the resource must be promoted * * \return Role that resource would have after scheduled actions are taken */ static void schedule_restart_actions(pcmk_resource_t *rsc, pcmk_node_t *current, bool need_stop, bool need_promote) { enum rsc_role_e role = rsc->priv->orig_role; enum rsc_role_e next_role; rsc_transition_fn fn = NULL; pcmk__set_rsc_flags(rsc, pcmk__rsc_restarting); // Bring resource down to a stop on its current node while (role != pcmk_role_stopped) { next_role = rsc_state_matrix[role][pcmk_role_stopped]; pcmk__rsc_trace(rsc, "Creating %s action to take %s down from %s to %s", (need_stop? "required" : "optional"), rsc->id, pcmk_role_text(role), pcmk_role_text(next_role)); fn = rsc_action_matrix[role][next_role]; if (fn == NULL) { break; } fn(rsc, current, !need_stop); role = next_role; } // Bring resource up to its next role on its next node while ((rsc->priv->orig_role <= rsc->priv->next_role) && (role != rsc->priv->orig_role) && !pcmk_is_set(rsc->flags, pcmk__rsc_blocked)) { bool required = need_stop; next_role = rsc_state_matrix[role][rsc->priv->orig_role]; if ((next_role == pcmk_role_promoted) && need_promote) { required = true; } pcmk__rsc_trace(rsc, "Creating %s action to take %s up from %s to %s", (required? "required" : "optional"), rsc->id, pcmk_role_text(role), pcmk_role_text(next_role)); fn = rsc_action_matrix[role][next_role]; if (fn == NULL) { break; } fn(rsc, rsc->priv->assigned_node, !required); role = next_role; } pcmk__clear_rsc_flags(rsc, pcmk__rsc_restarting); } /*! * \internal * \brief If a resource's next role is not explicitly specified, set a default * * \param[in,out] rsc Resource to set next role for * * \return "explicit" if next role was explicitly set, otherwise "implicit" */ static const char * set_default_next_role(pcmk_resource_t *rsc) { if (rsc->priv->next_role != pcmk_role_unknown) { return "explicit"; } if (rsc->priv->assigned_node == NULL) { pe__set_next_role(rsc, pcmk_role_stopped, "assignment"); } else { pe__set_next_role(rsc, pcmk_role_started, "assignment"); } return "implicit"; } /*! * \internal * \brief Create an action to represent an already pending start * * \param[in,out] rsc Resource to create start action for */ static void create_pending_start(pcmk_resource_t *rsc) { pcmk_action_t *start = NULL; pcmk__rsc_trace(rsc, "Creating action for %s to represent already pending start", rsc->id); start = start_action(rsc, rsc->priv->assigned_node, TRUE); pcmk__set_action_flags(start, pcmk__action_always_in_graph); } /*! * \internal * \brief Schedule actions needed to take a resource to its next role * * \param[in,out] rsc Resource to schedule actions for */ static void schedule_role_transition_actions(pcmk_resource_t *rsc) { enum rsc_role_e role = rsc->priv->orig_role; while (role != rsc->priv->next_role) { enum rsc_role_e next_role = rsc_state_matrix[role][rsc->priv->next_role]; rsc_transition_fn fn = NULL; pcmk__rsc_trace(rsc, "Creating action to take %s from %s to %s " "(ending at %s)", rsc->id, pcmk_role_text(role), pcmk_role_text(next_role), pcmk_role_text(rsc->priv->next_role)); fn = rsc_action_matrix[role][next_role]; if (fn == NULL) { break; } fn(rsc, rsc->priv->assigned_node, false); role = next_role; } } /*! * \internal * \brief Create all actions needed for a given primitive resource * * \param[in,out] rsc Primitive resource to create actions for */ void pcmk__primitive_create_actions(pcmk_resource_t *rsc) { bool need_stop = false; bool need_promote = false; bool is_moving = false; bool allow_migrate = false; bool multiply_active = false; pcmk_node_t *current = NULL; pcmk_node_t *migration_target = NULL; unsigned int num_all_active = 0; unsigned int num_clean_active = 0; const char *next_role_source = NULL; pcmk__assert(pcmk__is_primitive(rsc)); next_role_source = set_default_next_role(rsc); pcmk__rsc_trace(rsc, "Creating all actions for %s transition from %s to %s " "(%s) on %s", rsc->id, pcmk_role_text(rsc->priv->orig_role), pcmk_role_text(rsc->priv->next_role), next_role_source, pcmk__node_name(rsc->priv->assigned_node)); current = rsc->priv->fns->active_node(rsc, &num_all_active, &num_clean_active); g_list_foreach(rsc->priv->dangling_migration_sources, pcmk__abort_dangling_migration, rsc); if ((current != NULL) && (rsc->priv->assigned_node != NULL) && !pcmk__same_node(current, rsc->priv->assigned_node) && (rsc->priv->next_role >= pcmk_role_started)) { pcmk__rsc_trace(rsc, "Moving %s from %s to %s", rsc->id, pcmk__node_name(current), pcmk__node_name(rsc->priv->assigned_node)); is_moving = true; allow_migrate = pcmk__rsc_can_migrate(rsc, current); // This is needed even if migrating (though I'm not sure why ...) need_stop = true; } // Check whether resource is partially migrated and/or multiply active migration_target = rsc->priv->partial_migration_target; if ((rsc->priv->partial_migration_source != NULL) && (migration_target != NULL) && allow_migrate && (num_all_active == 2) && pcmk__same_node(current, rsc->priv->partial_migration_source) && pcmk__same_node(rsc->priv->assigned_node, migration_target)) { /* A partial migration is in progress, and the migration target remains * the same as when the migration began. */ pcmk__rsc_trace(rsc, "Partial migration of %s from %s to %s will continue", rsc->id, pcmk__node_name(rsc->priv->partial_migration_source), pcmk__node_name(migration_target)); } else if ((rsc->priv->partial_migration_source != NULL) || (migration_target != NULL)) { // A partial migration is in progress but can't be continued if (num_all_active > 2) { // The resource is migrating *and* multiply active! crm_notice("Forcing recovery of %s because it is migrating " "from %s to %s and possibly active elsewhere", rsc->id, pcmk__node_name(rsc->priv->partial_migration_source), pcmk__node_name(migration_target)); } else { // The migration source or target isn't available crm_notice("Forcing recovery of %s because it can no longer " "migrate from %s to %s", rsc->id, pcmk__node_name(rsc->priv->partial_migration_source), pcmk__node_name(migration_target)); } need_stop = true; rsc->priv->partial_migration_source = NULL; rsc->priv->partial_migration_target = NULL; allow_migrate = false; } else if (pcmk_is_set(rsc->flags, pcmk__rsc_needs_fencing)) { multiply_active = (num_all_active > 1); } else { /* If a resource has PCMK_META_REQUIRES set to PCMK_VALUE_NOTHING or * PCMK_VALUE_QUORUM, don't consider it active on unclean nodes (similar * to how all resources behave when PCMK_OPT_STONITH_ENABLED is false). * We can start such resources elsewhere before fencing completes, and * if we considered the resource active on the failed node, we would * attempt recovery for being active on multiple nodes. */ multiply_active = (num_clean_active > 1); } if (multiply_active) { const char *class = crm_element_value(rsc->priv->xml, PCMK_XA_CLASS); // Resource was (possibly) incorrectly multiply active pcmk__sched_err(rsc->priv->scheduler, "%s resource %s might be active on %u nodes (%s)", pcmk__s(class, "Untyped"), rsc->id, num_all_active, pcmk__multiply_active_text(rsc)); crm_notice("For more information, see \"What are multiply active " "resources?\" at " "https://projects.clusterlabs.org/w/clusterlabs/faq/"); switch (rsc->priv->multiply_active_policy) { case pcmk__multiply_active_restart: need_stop = true; break; case pcmk__multiply_active_unexpected: need_stop = true; // stop_resource() will skip expected node pcmk__set_rsc_flags(rsc, pcmk__rsc_stop_unexpected); break; default: break; } } else { pcmk__clear_rsc_flags(rsc, pcmk__rsc_stop_unexpected); } if (pcmk_is_set(rsc->flags, pcmk__rsc_start_pending)) { create_pending_start(rsc); } if (is_moving) { // Remaining tests are only for resources staying where they are } else if (pcmk_is_set(rsc->flags, pcmk__rsc_failed)) { if (pcmk_is_set(rsc->flags, pcmk__rsc_stop_if_failed)) { need_stop = true; pcmk__rsc_trace(rsc, "Recovering %s", rsc->id); } else { pcmk__rsc_trace(rsc, "Recovering %s by demotion", rsc->id); if (rsc->priv->next_role == pcmk_role_promoted) { need_promote = true; } } } else if (pcmk_is_set(rsc->flags, pcmk__rsc_blocked)) { pcmk__rsc_trace(rsc, "Blocking further actions on %s", rsc->id); need_stop = true; } else if ((rsc->priv->orig_role > pcmk_role_started) && (current != NULL) && (rsc->priv->assigned_node != NULL)) { pcmk_action_t *start = NULL; pcmk__rsc_trace(rsc, "Creating start action for promoted resource %s", rsc->id); start = start_action(rsc, rsc->priv->assigned_node, TRUE); if (!pcmk_is_set(start->flags, pcmk__action_optional)) { // Recovery of a promoted resource pcmk__rsc_trace(rsc, "%s restart is required for recovery", rsc->id); need_stop = true; } } // Create any actions needed to bring resource down and back up to same role schedule_restart_actions(rsc, current, need_stop, need_promote); // Create any actions needed to take resource from this role to the next schedule_role_transition_actions(rsc); pcmk__create_recurring_actions(rsc); if (allow_migrate) { pcmk__create_migration_actions(rsc, current); } } /*! * \internal * \brief Ban a resource from any allowed nodes that are Pacemaker Remote nodes * * \param[in] rsc Resource to check */ static void rsc_avoids_remote_nodes(const pcmk_resource_t *rsc) { GHashTableIter iter; pcmk_node_t *node = NULL; g_hash_table_iter_init(&iter, rsc->priv->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (void **) &node)) { if (node->priv->remote != NULL) { node->assign->score = -PCMK_SCORE_INFINITY; } } } /*! * \internal * \brief Return allowed nodes as (possibly sorted) list * * Convert a resource's hash table of allowed nodes to a list. If printing to * stdout, sort the list, to keep action ID numbers consistent for regression * test output (while avoiding the performance hit on a live cluster). * * \param[in] rsc Resource to check for allowed nodes * * \return List of resource's allowed nodes * \note Callers should take care not to rely on the list being sorted. */ static GList * allowed_nodes_as_list(const pcmk_resource_t *rsc) { GList *allowed_nodes = NULL; if (rsc->priv->allowed_nodes != NULL) { allowed_nodes = g_hash_table_get_values(rsc->priv->allowed_nodes); } if (!pcmk__is_daemon) { allowed_nodes = g_list_sort(allowed_nodes, pe__cmp_node_name); } return allowed_nodes; } /*! * \internal * \brief Create implicit constraints needed for a primitive resource * * \param[in,out] rsc Primitive resource to create implicit constraints for */ void pcmk__primitive_internal_constraints(pcmk_resource_t *rsc) { GList *allowed_nodes = NULL; bool check_unfencing = false; bool check_utilization = false; pcmk_scheduler_t *scheduler = NULL; pcmk__assert(pcmk__is_primitive(rsc)); scheduler = rsc->priv->scheduler; if (!pcmk_is_set(rsc->flags, pcmk__rsc_managed)) { pcmk__rsc_trace(rsc, "Skipping implicit constraints for unmanaged resource " "%s", rsc->id); return; } // Whether resource requires unfencing check_unfencing = !pcmk_is_set(rsc->flags, pcmk__rsc_fence_device) && pcmk_is_set(scheduler->flags, pcmk__sched_enable_unfencing) && pcmk_is_set(rsc->flags, pcmk__rsc_needs_unfencing); // Whether a non-default placement strategy is used check_utilization = (g_hash_table_size(rsc->priv->utilization) > 0) && !pcmk__str_eq(scheduler->priv->placement_strategy, PCMK_VALUE_DEFAULT, pcmk__str_casei); // Order stops before starts (i.e. restart) pcmk__new_ordering(rsc, pcmk__op_key(rsc->id, PCMK_ACTION_STOP, 0), NULL, rsc, pcmk__op_key(rsc->id, PCMK_ACTION_START, 0), NULL, pcmk__ar_ordered |pcmk__ar_first_implies_then |pcmk__ar_intermediate_stop, scheduler); // Promotable ordering: demote before stop, start before promote if (pcmk_is_set(pe__const_top_resource(rsc, false)->flags, pcmk__rsc_promotable) || (rsc->priv->orig_role > pcmk_role_unpromoted)) { pcmk__new_ordering(rsc, pcmk__op_key(rsc->id, PCMK_ACTION_DEMOTE, 0), NULL, rsc, pcmk__op_key(rsc->id, PCMK_ACTION_STOP, 0), NULL, pcmk__ar_promoted_then_implies_first, scheduler); pcmk__new_ordering(rsc, pcmk__op_key(rsc->id, PCMK_ACTION_START, 0), NULL, rsc, pcmk__op_key(rsc->id, PCMK_ACTION_PROMOTE, 0), NULL, pcmk__ar_unrunnable_first_blocks, scheduler); } // Don't clear resource history if probing on same node pcmk__new_ordering(rsc, pcmk__op_key(rsc->id, PCMK_ACTION_LRM_DELETE, 0), NULL, rsc, pcmk__op_key(rsc->id, PCMK_ACTION_MONITOR, 0), NULL, pcmk__ar_if_on_same_node|pcmk__ar_then_cancels_first, scheduler); // Certain checks need allowed nodes if (check_unfencing || check_utilization || (rsc->priv->launcher != NULL)) { allowed_nodes = allowed_nodes_as_list(rsc); } if (check_unfencing) { g_list_foreach(allowed_nodes, pcmk__order_restart_vs_unfence, rsc); } if (check_utilization) { pcmk__create_utilization_constraints(rsc, allowed_nodes); } if (rsc->priv->launcher != NULL) { pcmk_resource_t *remote_rsc = NULL; if (pcmk_is_set(rsc->flags, pcmk__rsc_is_remote_connection)) { // rsc is the implicit remote connection for a guest or bundle node /* Guest resources are not allowed to run on Pacemaker Remote nodes, * to avoid nesting remotes. However, bundles are allowed. */ if (!pcmk_is_set(rsc->flags, pcmk__rsc_remote_nesting_allowed)) { rsc_avoids_remote_nodes(rsc->priv->launcher); } /* If someone cleans up a guest or bundle node's launcher, we will * likely schedule a (re-)probe of the launcher and recovery of the * connection. Order the connection stop after the launcher probe, * so that if we detect the launcher running, we will trigger a new * transition and avoid the unnecessary recovery. */ pcmk__order_resource_actions(rsc->priv->launcher, PCMK_ACTION_MONITOR, rsc, PCMK_ACTION_STOP, pcmk__ar_ordered); /* A user can specify that a resource must start on a Pacemaker Remote * node by explicitly configuring it with the PCMK__META_CONTAINER * meta-attribute. This is of questionable merit, since location * constraints can accomplish the same thing. But we support it, so here * we check whether a resource (that is not itself a remote connection) * has PCMK__META_CONTAINER set to a remote node or guest node resource. */ } else if (pcmk_is_set(rsc->priv->launcher->flags, pcmk__rsc_is_remote_connection)) { remote_rsc = rsc->priv->launcher; } else { remote_rsc = pe__resource_contains_guest_node(scheduler, rsc->priv->launcher); } if (remote_rsc != NULL) { /* Force the resource on the Pacemaker Remote node instead of * colocating the resource with the launcher. */ for (GList *item = allowed_nodes; item; item = item->next) { pcmk_node_t *node = item->data; if (node->priv->remote != remote_rsc) { node->assign->score = -PCMK_SCORE_INFINITY; } } } else { /* This resource is either launched by a resource that does NOT * represent a Pacemaker Remote node, or a Pacemaker Remote * connection resource for a guest node or bundle. */ int score; crm_trace("Order and colocate %s relative to its launcher %s", rsc->id, rsc->priv->launcher->id); pcmk__new_ordering(rsc->priv->launcher, pcmk__op_key(rsc->priv->launcher->id, PCMK_ACTION_START, 0), NULL, rsc, pcmk__op_key(rsc->id, PCMK_ACTION_START, 0), NULL, pcmk__ar_first_implies_then |pcmk__ar_unrunnable_first_blocks, scheduler); pcmk__new_ordering(rsc, pcmk__op_key(rsc->id, PCMK_ACTION_STOP, 0), NULL, rsc->priv->launcher, pcmk__op_key(rsc->priv->launcher->id, PCMK_ACTION_STOP, 0), NULL, pcmk__ar_then_implies_first, scheduler); if (pcmk_is_set(rsc->flags, pcmk__rsc_remote_nesting_allowed)) { score = 10000; /* Highly preferred but not essential */ } else { score = PCMK_SCORE_INFINITY; // Force to run on same host } pcmk__new_colocation("#resource-with-container", NULL, score, rsc, rsc->priv->launcher, NULL, NULL, pcmk__coloc_influence); } } if (pcmk_is_set(rsc->flags, pcmk__rsc_is_remote_connection) || pcmk_is_set(rsc->flags, pcmk__rsc_fence_device)) { /* Remote connections and fencing devices are not allowed to run on * Pacemaker Remote nodes */ rsc_avoids_remote_nodes(rsc); } g_list_free(allowed_nodes); } /*! * \internal * \brief Apply a colocation's score to node scores or resource priority * * Given a colocation constraint, apply its score to the dependent's * allowed node scores (if we are still placing resources) or priority (if * we are choosing promotable clone instance roles). * * \param[in,out] dependent Dependent resource in colocation * \param[in] primary Primary resource in colocation * \param[in] colocation Colocation constraint to apply * \param[in] for_dependent true if called on behalf of dependent * * \return The score added to the dependent's priority */ int pcmk__primitive_apply_coloc_score(pcmk_resource_t *dependent, const pcmk_resource_t *primary, const pcmk__colocation_t *colocation, bool for_dependent) { enum pcmk__coloc_affects filter_results; pcmk__assert((dependent != NULL) && (primary != NULL) && (colocation != NULL)); if (for_dependent) { // Always process on behalf of primary resource return primary->priv->cmds->apply_coloc_score(dependent, primary, colocation, false); } filter_results = pcmk__colocation_affects(dependent, primary, colocation, false); pcmk__rsc_trace(dependent, "%s %s with %s (%s, score=%d, filter=%d)", ((colocation->score > 0)? "Colocating" : "Anti-colocating"), dependent->id, primary->id, colocation->id, colocation->score, filter_results); switch (filter_results) { case pcmk__coloc_affects_role: return pcmk__apply_coloc_to_priority(dependent, primary, colocation); case pcmk__coloc_affects_location: pcmk__apply_coloc_to_scores(dependent, primary, colocation); return 0; default: // pcmk__coloc_affects_nothing return 0; } } /* Primitive implementation of * pcmk__assignment_methods_t:with_this_colocations() */ void pcmk__with_primitive_colocations(const pcmk_resource_t *rsc, const pcmk_resource_t *orig_rsc, GList **list) { const pcmk_resource_t *parent = NULL; pcmk__assert(pcmk__is_primitive(rsc) && (list != NULL)); parent = rsc->priv->parent; if (rsc == orig_rsc) { /* For the resource itself, add all of its own colocations and relevant * colocations from its parent (if any). */ pcmk__add_with_this_list(list, rsc->priv->with_this_colocations, orig_rsc); if (parent != NULL) { parent->priv->cmds->with_this_colocations(parent, orig_rsc, list); } } else { // For an ancestor, add only explicitly configured constraints for (GList *iter = rsc->priv->with_this_colocations; iter != NULL; iter = iter->next) { pcmk__colocation_t *colocation = iter->data; if (pcmk_is_set(colocation->flags, pcmk__coloc_explicit)) { pcmk__add_with_this(list, colocation, orig_rsc); } } } } /* Primitive implementation of * pcmk__assignment_methods_t:this_with_colocations() */ void pcmk__primitive_with_colocations(const pcmk_resource_t *rsc, const pcmk_resource_t *orig_rsc, GList **list) { const pcmk_resource_t *parent = NULL; pcmk__assert(pcmk__is_primitive(rsc) && (list != NULL)); parent = rsc->priv->parent; if (rsc == orig_rsc) { /* For the resource itself, add all of its own colocations and relevant * colocations from its parent (if any). */ pcmk__add_this_with_list(list, rsc->priv->this_with_colocations, orig_rsc); if (parent != NULL) { parent->priv->cmds->this_with_colocations(parent, orig_rsc, list); } } else { // For an ancestor, add only explicitly configured constraints for (GList *iter = rsc->priv->this_with_colocations; iter != NULL; iter = iter->next) { pcmk__colocation_t *colocation = iter->data; if (pcmk_is_set(colocation->flags, pcmk__coloc_explicit)) { pcmk__add_this_with(list, colocation, orig_rsc); } } } } /*! * \internal * \brief Return action flags for a given primitive resource action * * \param[in,out] action Action to get flags for * \param[in] node If not NULL, limit effects to this node (ignored) * * \return Flags appropriate to \p action on \p node */ uint32_t pcmk__primitive_action_flags(pcmk_action_t *action, const pcmk_node_t *node) { pcmk__assert(action != NULL); return (uint32_t) action->flags; } /*! * \internal * \brief Check whether a node is a multiply active resource's expected node * * \param[in] rsc Resource to check * \param[in] node Node to check * * \return \c true if \p rsc is multiply active with * \c PCMK_META_MULTIPLE_ACTIVE set to \c PCMK_VALUE_STOP_UNEXPECTED, * and \p node is the node where it will remain active * \note This assumes that the resource's next role cannot be changed to stopped * after this is called, which should be reasonable if status has already * been unpacked and resources have been assigned to nodes. */ static bool is_expected_node(const pcmk_resource_t *rsc, const pcmk_node_t *node) { return pcmk_all_flags_set(rsc->flags, pcmk__rsc_stop_unexpected|pcmk__rsc_restarting) && (rsc->priv->next_role > pcmk_role_stopped) && pcmk__same_node(rsc->priv->assigned_node, node); } /*! * \internal * \brief Schedule actions needed to stop a resource wherever it is active * * \param[in,out] rsc Resource being stopped * \param[in] node Node where resource is being stopped (ignored) * \param[in] optional Whether actions should be optional */ static void stop_resource(pcmk_resource_t *rsc, pcmk_node_t *node, bool optional) { for (GList *iter = rsc->priv->active_nodes; iter != NULL; iter = iter->next) { pcmk_node_t *current = (pcmk_node_t *) iter->data; pcmk_action_t *stop = NULL; if (is_expected_node(rsc, current)) { /* We are scheduling restart actions for a multiply active resource * with PCMK_META_MULTIPLE_ACTIVE=PCMK_VALUE_STOP_UNEXPECTED, and * this is where it should not be stopped. */ pcmk__rsc_trace(rsc, "Skipping stop of multiply active resource %s " "on expected node %s", rsc->id, pcmk__node_name(current)); continue; } if (rsc->priv->partial_migration_target != NULL) { // Continue migration if node originally was and remains target if (pcmk__same_node(current, rsc->priv->partial_migration_target) && pcmk__same_node(current, rsc->priv->assigned_node)) { pcmk__rsc_trace(rsc, "Skipping stop of %s on %s " "because partial migration there will continue", rsc->id, pcmk__node_name(current)); continue; } else { pcmk__rsc_trace(rsc, "Forcing stop of %s on %s " "because migration target changed", rsc->id, pcmk__node_name(current)); optional = false; } } pcmk__rsc_trace(rsc, "Scheduling stop of %s on %s", rsc->id, pcmk__node_name(current)); stop = stop_action(rsc, current, optional); if (rsc->priv->assigned_node == NULL) { pe_action_set_reason(stop, "node availability", true); } else if (pcmk_all_flags_set(rsc->flags, pcmk__rsc_restarting |pcmk__rsc_stop_unexpected)) { /* We are stopping a multiply active resource on a node that is * not its expected node, and we are still scheduling restart * actions, so the stop is for being multiply active. */ pe_action_set_reason(stop, "being multiply active", true); } if (!pcmk_is_set(rsc->flags, pcmk__rsc_managed)) { pcmk__clear_action_flags(stop, pcmk__action_runnable); } if (pcmk_is_set(rsc->flags, pcmk__rsc_needs_unfencing)) { pcmk_action_t *unfence = pe_fence_op(current, PCMK_ACTION_ON, true, NULL, false, rsc->priv->scheduler); order_actions(stop, unfence, pcmk__ar_then_implies_first); if (!pcmk__node_unfenced(current)) { pcmk__sched_err(rsc->priv->scheduler, "Stopping %s until %s can be unfenced", rsc->id, pcmk__node_name(current)); } } } } /*! * \internal * \brief Schedule actions needed to start a resource on a node * * \param[in,out] rsc Resource being started * \param[in,out] node Node where resource should be started * \param[in] optional Whether actions should be optional */ static void start_resource(pcmk_resource_t *rsc, pcmk_node_t *node, bool optional) { pcmk_action_t *start = NULL; pcmk__assert(node != NULL); pcmk__rsc_trace(rsc, "Scheduling %s start of %s on %s (score %d)", (optional? "optional" : "required"), rsc->id, pcmk__node_name(node), node->assign->score); start = start_action(rsc, node, TRUE); pcmk__order_vs_unfence(rsc, node, start, pcmk__ar_first_implies_then); if (pcmk_is_set(start->flags, pcmk__action_runnable) && !optional) { pcmk__clear_action_flags(start, pcmk__action_optional); } if (is_expected_node(rsc, node)) { /* This could be a problem if the start becomes necessary for other * reasons later. */ pcmk__rsc_trace(rsc, "Start of multiply active resouce %s " "on expected node %s will be a pseudo-action", rsc->id, pcmk__node_name(node)); pcmk__set_action_flags(start, pcmk__action_pseudo); } } /*! * \internal * \brief Schedule actions needed to promote a resource on a node * * \param[in,out] rsc Resource being promoted * \param[in] node Node where resource should be promoted * \param[in] optional Whether actions should be optional */ static void promote_resource(pcmk_resource_t *rsc, pcmk_node_t *node, bool optional) { GList *iter = NULL; GList *action_list = NULL; bool runnable = true; pcmk__assert(node != NULL); // Any start must be runnable for promotion to be runnable action_list = pe__resource_actions(rsc, node, PCMK_ACTION_START, true); for (iter = action_list; iter != NULL; iter = iter->next) { pcmk_action_t *start = (pcmk_action_t *) iter->data; if (!pcmk_is_set(start->flags, pcmk__action_runnable)) { runnable = false; } } g_list_free(action_list); if (runnable) { pcmk_action_t *promote = promote_action(rsc, node, optional); pcmk__rsc_trace(rsc, "Scheduling %s promotion of %s on %s", (optional? "optional" : "required"), rsc->id, pcmk__node_name(node)); if (is_expected_node(rsc, node)) { /* This could be a problem if the promote becomes necessary for * other reasons later. */ pcmk__rsc_trace(rsc, "Promotion of multiply active resouce %s " "on expected node %s will be a pseudo-action", rsc->id, pcmk__node_name(node)); pcmk__set_action_flags(promote, pcmk__action_pseudo); } } else { pcmk__rsc_trace(rsc, "Not promoting %s on %s: start unrunnable", rsc->id, pcmk__node_name(node)); action_list = pe__resource_actions(rsc, node, PCMK_ACTION_PROMOTE, true); for (iter = action_list; iter != NULL; iter = iter->next) { pcmk_action_t *promote = (pcmk_action_t *) iter->data; pcmk__clear_action_flags(promote, pcmk__action_runnable); } g_list_free(action_list); } } /*! * \internal * \brief Schedule actions needed to demote a resource wherever it is active * * \param[in,out] rsc Resource being demoted * \param[in] node Node where resource should be demoted (ignored) * \param[in] optional Whether actions should be optional */ static void demote_resource(pcmk_resource_t *rsc, pcmk_node_t *node, bool optional) { /* Since this will only be called for a primitive (possibly as an instance * of a collective resource), the resource is multiply active if it is * running on more than one node, so we want to demote on all of them as * part of recovery, regardless of which one is the desired node. */ for (GList *iter = rsc->priv->active_nodes; iter != NULL; iter = iter->next) { pcmk_node_t *current = (pcmk_node_t *) iter->data; if (is_expected_node(rsc, current)) { pcmk__rsc_trace(rsc, "Skipping demote of multiply active resource %s " "on expected node %s", rsc->id, pcmk__node_name(current)); } else { pcmk__rsc_trace(rsc, "Scheduling %s demotion of %s on %s", (optional? "optional" : "required"), rsc->id, pcmk__node_name(current)); demote_action(rsc, current, optional); } } } static void assert_role_error(pcmk_resource_t *rsc, pcmk_node_t *node, bool optional) { pcmk__assert(false); } /*! * \internal * \brief Schedule cleanup of a resource * * \param[in,out] rsc Resource to clean up * \param[in] node Node to clean up on * \param[in] optional Whether clean-up should be optional */ void pcmk__schedule_cleanup(pcmk_resource_t *rsc, const pcmk_node_t *node, bool optional) { /* If the cleanup is required, its orderings are optional, because they're * relevant only if both actions are required. Conversely, if the cleanup is * optional, the orderings make the then action required if the first action * becomes required. */ uint32_t flag = optional? pcmk__ar_first_implies_then : pcmk__ar_ordered; CRM_CHECK((rsc != NULL) && (node != NULL), return); if (pcmk_is_set(rsc->flags, pcmk__rsc_failed)) { pcmk__rsc_trace(rsc, "Skipping clean-up of %s on %s: resource failed", rsc->id, pcmk__node_name(node)); return; } if (node->details->unclean || !node->details->online) { pcmk__rsc_trace(rsc, "Skipping clean-up of %s on %s: node unavailable", rsc->id, pcmk__node_name(node)); return; } crm_notice("Scheduling clean-up of %s on %s", rsc->id, pcmk__node_name(node)); delete_action(rsc, node, optional); // stop -> clean-up -> start pcmk__order_resource_actions(rsc, PCMK_ACTION_STOP, rsc, PCMK_ACTION_DELETE, flag); pcmk__order_resource_actions(rsc, PCMK_ACTION_DELETE, rsc, PCMK_ACTION_START, flag); } /*! * \internal * \brief Add primitive meta-attributes relevant to graph actions to XML * * \param[in] rsc Primitive resource whose meta-attributes should be added * \param[in,out] xml Transition graph action attributes XML to add to */ void pcmk__primitive_add_graph_meta(const pcmk_resource_t *rsc, xmlNode *xml) { char *name = NULL; char *value = NULL; const pcmk_resource_t *parent = NULL; pcmk__assert(pcmk__is_primitive(rsc) && (xml != NULL)); /* Clone instance numbers get set internally as meta-attributes, and are * needed in the transition graph (for example, to tell unique clone * instances apart). */ value = g_hash_table_lookup(rsc->priv->meta, PCMK__META_CLONE); if (value != NULL) { name = crm_meta_name(PCMK__META_CLONE); crm_xml_add(xml, name, value); free(name); } // Not sure if this one is really needed ... value = g_hash_table_lookup(rsc->priv->meta, PCMK_META_REMOTE_NODE); if (value != NULL) { name = crm_meta_name(PCMK_META_REMOTE_NODE); crm_xml_add(xml, name, value); free(name); } /* The PCMK__META_CONTAINER meta-attribute can be set on the primitive * itself or one of its ancestors, so check them all and keep the highest. */ for (parent = rsc; parent != NULL; parent = parent->priv->parent) { if (parent->priv->launcher != NULL) { crm_xml_add(xml, CRM_META "_" PCMK__META_CONTAINER, parent->priv->launcher->id); } } /* Bundle replica children will get their external-ip set internally as a * meta-attribute. The graph action needs it, but under a different naming * convention than other meta-attributes. */ value = g_hash_table_lookup(rsc->priv->meta, "external-ip"); if (value != NULL) { crm_xml_add(xml, "pcmk_external_ip", value); } } // Primitive implementation of pcmk__assignment_methods_t:add_utilization() void pcmk__primitive_add_utilization(const pcmk_resource_t *rsc, const pcmk_resource_t *orig_rsc, GList *all_rscs, GHashTable *utilization) { pcmk__assert(pcmk__is_primitive(rsc) && (orig_rsc != NULL) && (utilization != NULL)); if (!pcmk_is_set(rsc->flags, pcmk__rsc_unassigned)) { return; } pcmk__rsc_trace(orig_rsc, "%s: Adding primitive %s as colocated utilization", orig_rsc->id, rsc->id); pcmk__release_node_capacity(utilization, rsc); } /*! * \internal * \brief Get epoch time of node's shutdown attribute (or now if none) * * \param[in,out] node Node to check * * \return Epoch time corresponding to shutdown attribute if set or now if not */ static time_t shutdown_time(pcmk_node_t *node) { const char *shutdown = pcmk__node_attr(node, PCMK__NODE_ATTR_SHUTDOWN, NULL, pcmk__rsc_node_current); time_t result = 0; if (shutdown != NULL) { long long result_ll; int rc = pcmk__scan_ll(shutdown, &result_ll, 0LL); if (rc == pcmk_rc_ok) { result = (time_t) result_ll; } else { crm_warn("Ignoring invalid value '%s' for %s " PCMK__NODE_ATTR_SHUTDOWN " attribute: %s", shutdown, pcmk__node_name(node), pcmk_rc_str(rc)); } } - return (result == 0)? get_effective_time(node->priv->scheduler) : result; + if (result == 0) { + result = pcmk__scheduler_epoch_time(node->priv->scheduler); + } + return result; } /*! * \internal * \brief Ban a resource from a node if it's not locked to the node * * \param[in] data Node to check * \param[in,out] user_data Resource to check */ static void ban_if_not_locked(gpointer data, gpointer user_data) { const pcmk_node_t *node = (const pcmk_node_t *) data; pcmk_resource_t *rsc = (pcmk_resource_t *) user_data; if (!pcmk__same_node(node, rsc->priv->lock_node)) { resource_location(rsc, node, -PCMK_SCORE_INFINITY, PCMK_OPT_SHUTDOWN_LOCK, rsc->priv->scheduler); } } // Primitive implementation of pcmk__assignment_methods_t:shutdown_lock() void pcmk__primitive_shutdown_lock(pcmk_resource_t *rsc) { pcmk_scheduler_t *scheduler = NULL; pcmk__assert(pcmk__is_primitive(rsc)); scheduler = rsc->priv->scheduler; // Fence devices and remote connections can't be locked if (pcmk_any_flags_set(rsc->flags, pcmk__rsc_fence_device |pcmk__rsc_is_remote_connection)) { return; } if (rsc->priv->lock_node != NULL) { // The lock was obtained from resource history if (rsc->priv->active_nodes != NULL) { /* The resource was started elsewhere even though it is now * considered locked. This shouldn't be possible, but as a * failsafe, we don't want to disturb the resource now. */ pcmk__rsc_info(rsc, "Cancelling shutdown lock " "because %s is already active", rsc->id); pe__clear_resource_history(rsc, rsc->priv->lock_node); rsc->priv->lock_node = NULL; rsc->priv->lock_time = 0; } // Only a resource active on exactly one node can be locked } else if (pcmk__list_of_1(rsc->priv->active_nodes)) { pcmk_node_t *node = rsc->priv->active_nodes->data; if (node->details->shutdown) { if (node->details->unclean) { pcmk__rsc_debug(rsc, "Not locking %s to unclean %s for shutdown", rsc->id, pcmk__node_name(node)); } else { rsc->priv->lock_node = node; rsc->priv->lock_time = shutdown_time(node); } } } if (rsc->priv->lock_node == NULL) { // No lock needed return; } if (scheduler->priv->shutdown_lock_ms > 0U) { time_t lock_expiration = rsc->priv->lock_time + pcmk__timeout_ms2s(scheduler->priv->shutdown_lock_ms); pcmk__rsc_info(rsc, "Locking %s to %s due to shutdown (expires @%lld)", rsc->id, pcmk__node_name(rsc->priv->lock_node), (long long) lock_expiration); - pe__update_recheck_time(++lock_expiration, scheduler, - "shutdown lock expiration"); + pcmk__update_recheck_time(++lock_expiration, scheduler, + "shutdown lock expiration"); } else { pcmk__rsc_info(rsc, "Locking %s to %s due to shutdown", rsc->id, pcmk__node_name(rsc->priv->lock_node)); } // If resource is locked to one node, ban it from all other nodes g_list_foreach(scheduler->nodes, ban_if_not_locked, rsc); } diff --git a/lib/pacemaker/pcmk_scheduler.c b/lib/pacemaker/pcmk_scheduler.c index 8960c298f5..4debba82b5 100644 --- a/lib/pacemaker/pcmk_scheduler.c +++ b/lib/pacemaker/pcmk_scheduler.c @@ -1,880 +1,872 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include #include "libpacemaker_private.h" CRM_TRACE_INIT_DATA(pacemaker); /*! * \internal * \brief Do deferred action checks after assignment * * When unpacking the resource history, the scheduler checks for resource * configurations that have changed since an action was run. However, at that * time, bundles using the REMOTE_CONTAINER_HACK don't have their final * parameter information, so instead they add a deferred check to a list. This * function processes one entry in that list. * * \param[in,out] rsc Resource that action history is for * \param[in,out] node Node that action history is for * \param[in] rsc_op Action history entry * \param[in] check Type of deferred check to do */ static void check_params(pcmk_resource_t *rsc, pcmk_node_t *node, const xmlNode *rsc_op, enum pcmk__check_parameters check) { const char *reason = NULL; pcmk__op_digest_t *digest_data = NULL; switch (check) { case pcmk__check_active: if (pcmk__check_action_config(rsc, node, rsc_op) && pe_get_failcount(node, rsc, NULL, pcmk__fc_effective, NULL)) { reason = "action definition changed"; } break; case pcmk__check_last_failure: digest_data = rsc_action_digest_cmp(rsc, rsc_op, node, rsc->priv->scheduler); switch (digest_data->rc) { case pcmk__digest_unknown: crm_trace("Resource %s history entry %s on %s has " "no digest to compare", rsc->id, pcmk__xe_id(rsc_op), node->priv->id); break; case pcmk__digest_match: break; default: reason = "resource parameters have changed"; break; } break; } if (reason != NULL) { pe__clear_failcount(rsc, node, reason, rsc->priv->scheduler); } } /*! * \internal * \brief Check whether a resource has failcount clearing scheduled on a node * * \param[in] node Node to check * \param[in] rsc Resource to check * * \return true if \p rsc has failcount clearing scheduled on \p node, * otherwise false */ static bool failcount_clear_action_exists(const pcmk_node_t *node, const pcmk_resource_t *rsc) { GList *list = pe__resource_actions(rsc, node, PCMK_ACTION_CLEAR_FAILCOUNT, TRUE); if (list != NULL) { g_list_free(list); return true; } return false; } /*! * \internal * \brief Ban a resource from a node if it reached its failure threshold there * * \param[in,out] data Resource to check failure threshold for * \param[in] user_data Node to check resource on */ static void check_failure_threshold(gpointer data, gpointer user_data) { pcmk_resource_t *rsc = data; const pcmk_node_t *node = user_data; // If this is a collective resource, apply recursively to children instead if (rsc->priv->children != NULL) { g_list_foreach(rsc->priv->children, check_failure_threshold, user_data); return; } if (!failcount_clear_action_exists(node, rsc)) { /* Don't force the resource away from this node due to a failcount * that's going to be cleared. * * @TODO Failcount clearing can be scheduled in * pcmk__handle_rsc_config_changes() via process_rsc_history(), or in * schedule_resource_actions() via check_params(). This runs well before * then, so it cannot detect those, meaning we might check the migration * threshold when we shouldn't. Worst case, we stop or move the * resource, then move it back in the next transition. */ pcmk_resource_t *failed = NULL; if (pcmk__threshold_reached(rsc, node, &failed)) { resource_location(failed, node, -PCMK_SCORE_INFINITY, "__fail_limit__", rsc->priv->scheduler); } } } /*! * \internal * \brief If resource has exclusive discovery, ban node if not allowed * * Location constraints have a PCMK_XA_RESOURCE_DISCOVERY option that allows * users to specify where probes are done for the affected resource. If this is * set to \c exclusive, probes will only be done on nodes listed in exclusive * constraints. This function bans the resource from the node if the node is not * listed. * * \param[in,out] data Resource to check * \param[in] user_data Node to check resource on */ static void apply_exclusive_discovery(gpointer data, gpointer user_data) { pcmk_resource_t *rsc = data; const pcmk_node_t *node = user_data; if (pcmk_is_set(rsc->flags, pcmk__rsc_exclusive_probes) || pcmk_is_set(pe__const_top_resource(rsc, false)->flags, pcmk__rsc_exclusive_probes)) { pcmk_node_t *match = NULL; // If this is a collective resource, apply recursively to children g_list_foreach(rsc->priv->children, apply_exclusive_discovery, user_data); match = g_hash_table_lookup(rsc->priv->allowed_nodes, node->priv->id); if ((match != NULL) && (match->assign->probe_mode != pcmk__probe_exclusive)) { match->assign->score = -PCMK_SCORE_INFINITY; } } } /*! * \internal * \brief Apply stickiness to a resource if appropriate * * \param[in,out] data Resource to check for stickiness * \param[in] user_data Ignored */ static void apply_stickiness(gpointer data, gpointer user_data) { pcmk_resource_t *rsc = data; pcmk_node_t *node = NULL; // If this is a collective resource, apply recursively to children instead if (rsc->priv->children != NULL) { g_list_foreach(rsc->priv->children, apply_stickiness, NULL); return; } /* A resource is sticky if it is managed, has stickiness configured, and is * active on a single node. */ if (!pcmk_is_set(rsc->flags, pcmk__rsc_managed) || (rsc->priv->stickiness < 1) || !pcmk__list_of_1(rsc->priv->active_nodes)) { return; } node = rsc->priv->active_nodes->data; /* In a symmetric cluster, stickiness can always be used. In an * asymmetric cluster, we have to check whether the resource is still * allowed on the node, so we don't keep the resource somewhere it is no * longer explicitly enabled. */ if (!pcmk_is_set(rsc->priv->scheduler->flags, pcmk__sched_symmetric_cluster) && (g_hash_table_lookup(rsc->priv->allowed_nodes, node->priv->id) == NULL)) { pcmk__rsc_debug(rsc, "Ignoring %s stickiness because the cluster is " "asymmetric and %s is not explicitly allowed", rsc->id, pcmk__node_name(node)); return; } pcmk__rsc_debug(rsc, "Resource %s has %d stickiness on %s", rsc->id, rsc->priv->stickiness, pcmk__node_name(node)); resource_location(rsc, node, rsc->priv->stickiness, "stickiness", rsc->priv->scheduler); } /*! * \internal * \brief Apply shutdown locks for all resources as appropriate * * \param[in,out] scheduler Scheduler data */ static void apply_shutdown_locks(pcmk_scheduler_t *scheduler) { if (!pcmk_is_set(scheduler->flags, pcmk__sched_shutdown_lock)) { return; } for (GList *iter = scheduler->priv->resources; iter != NULL; iter = iter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) iter->data; rsc->priv->cmds->shutdown_lock(rsc); } } /* * \internal * \brief Apply node-specific scheduling criteria * * After the CIB has been unpacked, process node-specific scheduling criteria * including shutdown locks, location constraints, resource stickiness, * migration thresholds, and exclusive resource discovery. */ static void apply_node_criteria(pcmk_scheduler_t *scheduler) { crm_trace("Applying node-specific scheduling criteria"); apply_shutdown_locks(scheduler); pcmk__apply_locations(scheduler); g_list_foreach(scheduler->priv->resources, apply_stickiness, NULL); for (GList *node_iter = scheduler->nodes; node_iter != NULL; node_iter = node_iter->next) { for (GList *rsc_iter = scheduler->priv->resources; rsc_iter != NULL; rsc_iter = rsc_iter->next) { check_failure_threshold(rsc_iter->data, node_iter->data); apply_exclusive_discovery(rsc_iter->data, node_iter->data); } } } /*! * \internal * \brief Assign resources to nodes * * \param[in,out] scheduler Scheduler data */ static void assign_resources(pcmk_scheduler_t *scheduler) { GList *iter = NULL; crm_trace("Assigning resources to nodes"); if (!pcmk__str_eq(scheduler->priv->placement_strategy, PCMK_VALUE_DEFAULT, pcmk__str_casei)) { pcmk__sort_resources(scheduler); } pcmk__show_node_capacities("Original", scheduler); if (pcmk_is_set(scheduler->flags, pcmk__sched_have_remote_nodes)) { /* Assign remote connection resources first (which will also assign any * colocation dependencies). If the connection is migrating, always * prefer the partial migration target. */ for (iter = scheduler->priv->resources; iter != NULL; iter = iter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) iter->data; const pcmk_node_t *target = rsc->priv->partial_migration_target; if (pcmk_is_set(rsc->flags, pcmk__rsc_is_remote_connection)) { pcmk__rsc_trace(rsc, "Assigning remote connection resource '%s'", rsc->id); rsc->priv->cmds->assign(rsc, target, true); } } } /* now do the rest of the resources */ for (iter = scheduler->priv->resources; iter != NULL; iter = iter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) iter->data; if (!pcmk_is_set(rsc->flags, pcmk__rsc_is_remote_connection)) { pcmk__rsc_trace(rsc, "Assigning %s resource '%s'", rsc->priv->xml->name, rsc->id); rsc->priv->cmds->assign(rsc, NULL, true); } } pcmk__show_node_capacities("Remaining", scheduler); } /*! * \internal * \brief Schedule fail count clearing on online nodes if resource is orphaned * * \param[in,out] data Resource to check * \param[in] user_data Ignored */ static void clear_failcounts_if_orphaned(gpointer data, gpointer user_data) { pcmk_resource_t *rsc = data; if (!pcmk_is_set(rsc->flags, pcmk__rsc_removed)) { return; } crm_trace("Clear fail counts for orphaned resource %s", rsc->id); /* There's no need to recurse into rsc->private->children because those * should just be unassigned clone instances. */ for (GList *iter = rsc->priv->scheduler->nodes; iter != NULL; iter = iter->next) { pcmk_node_t *node = (pcmk_node_t *) iter->data; pcmk_action_t *clear_op = NULL; if (!node->details->online) { continue; } if (pe_get_failcount(node, rsc, NULL, pcmk__fc_effective, NULL) == 0) { continue; } clear_op = pe__clear_failcount(rsc, node, "it is orphaned", rsc->priv->scheduler); /* We can't use order_action_then_stop() here because its * pcmk__ar_guest_allowed breaks things */ pcmk__new_ordering(clear_op->rsc, NULL, clear_op, rsc, stop_key(rsc), NULL, pcmk__ar_ordered, rsc->priv->scheduler); } } /*! * \internal * \brief Schedule any resource actions needed * * \param[in,out] scheduler Scheduler data */ static void schedule_resource_actions(pcmk_scheduler_t *scheduler) { // Process deferred action checks - pe__foreach_param_check(scheduler, check_params); - pe__free_param_checks(scheduler); + pcmk__foreach_param_check(scheduler, check_params); + pcmk__free_param_checks(scheduler); if (pcmk_is_set(scheduler->flags, pcmk__sched_probe_resources)) { crm_trace("Scheduling probes"); pcmk__schedule_probes(scheduler); } if (pcmk_is_set(scheduler->flags, pcmk__sched_stop_removed_resources)) { g_list_foreach(scheduler->priv->resources, clear_failcounts_if_orphaned, NULL); } crm_trace("Scheduling resource actions"); for (GList *iter = scheduler->priv->resources; iter != NULL; iter = iter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) iter->data; rsc->priv->cmds->create_actions(rsc); } } /*! * \internal * \brief Check whether a resource or any of its descendants are managed * * \param[in] rsc Resource to check * * \return true if resource or any descendant is managed, otherwise false */ static bool is_managed(const pcmk_resource_t *rsc) { if (pcmk_is_set(rsc->flags, pcmk__rsc_managed)) { return true; } for (GList *iter = rsc->priv->children; iter != NULL; iter = iter->next) { if (is_managed((pcmk_resource_t *) iter->data)) { return true; } } return false; } /*! * \internal * \brief Check whether any resources in the cluster are managed * * \param[in] scheduler Scheduler data * * \return true if any resource is managed, otherwise false */ static bool any_managed_resources(const pcmk_scheduler_t *scheduler) { for (const GList *iter = scheduler->priv->resources; iter != NULL; iter = iter->next) { if (is_managed((const pcmk_resource_t *) iter->data)) { return true; } } return false; } /*! * \internal * \brief Check whether a node requires fencing * * \param[in] node Node to check * \param[in] have_managed Whether any resource in cluster is managed * * \return true if \p node should be fenced, otherwise false */ static bool needs_fencing(const pcmk_node_t *node, bool have_managed) { return have_managed && node->details->unclean && pe_can_fence(node->priv->scheduler, node); } /*! * \internal * \brief Check whether a node requires shutdown * * \param[in] node Node to check * * \return true if \p node should be shut down, otherwise false */ static bool needs_shutdown(const pcmk_node_t *node) { if (pcmk__is_pacemaker_remote_node(node)) { /* Do not send shutdown actions for Pacemaker Remote nodes. * @TODO We might come up with a good use for this in the future. */ return false; } return node->details->online && node->details->shutdown; } /*! * \internal * \brief Track and order non-DC fencing * * \param[in,out] list List of existing non-DC fencing actions * \param[in,out] action Fencing action to prepend to \p list * \param[in] scheduler Scheduler data * * \return (Possibly new) head of \p list */ static GList * add_nondc_fencing(GList *list, pcmk_action_t *action, const pcmk_scheduler_t *scheduler) { if (!pcmk_is_set(scheduler->flags, pcmk__sched_concurrent_fencing) && (list != NULL)) { /* Concurrent fencing is disabled, so order each non-DC * fencing in a chain. If there is any DC fencing or * shutdown, it will be ordered after the last action in the * chain later. */ order_actions((pcmk_action_t *) list->data, action, pcmk__ar_ordered); } return g_list_prepend(list, action); } /*! * \internal * \brief Schedule a node for fencing * * \param[in,out] node Node that requires fencing */ static pcmk_action_t * schedule_fencing(pcmk_node_t *node) { pcmk_action_t *fencing = pe_fence_op(node, NULL, FALSE, "node is unclean", FALSE, node->priv->scheduler); pcmk__sched_warn(node->priv->scheduler, "Scheduling node %s for fencing", pcmk__node_name(node)); pcmk__order_vs_fence(fencing, node->priv->scheduler); return fencing; } /*! * \internal * \brief Create and order node fencing and shutdown actions * * \param[in,out] scheduler Scheduler data */ static void schedule_fencing_and_shutdowns(pcmk_scheduler_t *scheduler) { pcmk_action_t *dc_down = NULL; bool integrity_lost = false; bool have_managed = any_managed_resources(scheduler); GList *fencing_ops = NULL; GList *shutdown_ops = NULL; crm_trace("Scheduling fencing and shutdowns as needed"); if (!have_managed) { crm_notice("No fencing will be done until there are resources " "to manage"); } // Check each node for whether it needs fencing or shutdown for (GList *iter = scheduler->nodes; iter != NULL; iter = iter->next) { pcmk_node_t *node = (pcmk_node_t *) iter->data; pcmk_action_t *fencing = NULL; const bool is_dc = pcmk__same_node(node, scheduler->dc_node); /* Guest nodes are "fenced" by recovering their container resource, * so handle them separately. */ if (pcmk__is_guest_or_bundle_node(node)) { if (pcmk_is_set(node->priv->flags, pcmk__node_remote_reset) && have_managed && pe_can_fence(scheduler, node)) { pcmk__fence_guest(node); } continue; } if (needs_fencing(node, have_managed)) { fencing = schedule_fencing(node); // Track DC and non-DC fence actions separately if (is_dc) { dc_down = fencing; } else { fencing_ops = add_nondc_fencing(fencing_ops, fencing, scheduler); } } else if (needs_shutdown(node)) { pcmk_action_t *down_op = pcmk__new_shutdown_action(node); // Track DC and non-DC shutdown actions separately if (is_dc) { dc_down = down_op; } else { shutdown_ops = g_list_prepend(shutdown_ops, down_op); } } if ((fencing == NULL) && node->details->unclean) { integrity_lost = true; pcmk__config_warn("Node %s is unclean but cannot be fenced", pcmk__node_name(node)); } } if (integrity_lost) { if (!pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled)) { pcmk__config_warn("Resource functionality and data integrity " "cannot be guaranteed (configure, enable, " "and test fencing to correct this)"); } else if (!pcmk_is_set(scheduler->flags, pcmk__sched_quorate)) { crm_notice("Unclean nodes will not be fenced until quorum is " "attained or " PCMK_OPT_NO_QUORUM_POLICY " is set to " PCMK_VALUE_IGNORE); } } if (dc_down != NULL) { /* Order any non-DC shutdowns before any DC shutdown, to avoid repeated * DC elections. However, we don't want to order non-DC shutdowns before * a DC *fencing*, because even though we don't want a node that's * shutting down to become DC, the DC fencing could be ordered before a * clone stop that's also ordered before the shutdowns, thus leading to * a graph loop. */ if (pcmk__str_eq(dc_down->task, PCMK_ACTION_DO_SHUTDOWN, pcmk__str_none)) { pcmk__order_after_each(dc_down, shutdown_ops); } // Order any non-DC fencing before any DC fencing or shutdown if (pcmk_is_set(scheduler->flags, pcmk__sched_concurrent_fencing)) { /* With concurrent fencing, order each non-DC fencing action * separately before any DC fencing or shutdown. */ pcmk__order_after_each(dc_down, fencing_ops); } else if (fencing_ops != NULL) { /* Without concurrent fencing, the non-DC fencing actions are * already ordered relative to each other, so we just need to order * the DC fencing after the last action in the chain (which is the * first item in the list). */ order_actions((pcmk_action_t *) fencing_ops->data, dc_down, pcmk__ar_ordered); } } g_list_free(fencing_ops); g_list_free(shutdown_ops); } static void log_resource_details(pcmk_scheduler_t *scheduler) { pcmk__output_t *out = scheduler->priv->out; GList *all = NULL; /* Due to the `crm_mon --node=` feature, out->message() for all the * resource-related messages expects a list of nodes that we are allowed to * output information for. Here, we create a wildcard to match all nodes. */ all = g_list_prepend(all, (gpointer) "*"); for (GList *item = scheduler->priv->resources; item != NULL; item = item->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) item->data; // Log all resources except inactive orphans if (!pcmk_is_set(rsc->flags, pcmk__rsc_removed) || (rsc->priv->orig_role != pcmk_role_stopped)) { out->message(out, (const char *) rsc->priv->xml->name, 0UL, rsc, all, all); } } g_list_free(all); } static void log_all_actions(pcmk_scheduler_t *scheduler) { /* This only ever outputs to the log, so ignore whatever output object was * previously set and just log instead. */ pcmk__output_t *prev_out = scheduler->priv->out; pcmk__output_t *out = NULL; if (pcmk__log_output_new(&out) != pcmk_rc_ok) { return; } pe__register_messages(out); pcmk__register_lib_messages(out); pcmk__output_set_log_level(out, LOG_NOTICE); scheduler->priv->out = out; out->begin_list(out, NULL, NULL, "Actions"); pcmk__output_actions(scheduler); out->end_list(out); out->finish(out, CRM_EX_OK, true, NULL); pcmk__output_free(out); scheduler->priv->out = prev_out; } /*! * \internal * \brief Log all required but unrunnable actions at trace level * * \param[in] scheduler Scheduler data */ static void log_unrunnable_actions(const pcmk_scheduler_t *scheduler) { const uint64_t flags = pcmk__action_optional |pcmk__action_runnable |pcmk__action_pseudo; crm_trace("Required but unrunnable actions:"); for (const GList *iter = scheduler->priv->actions; iter != NULL; iter = iter->next) { const pcmk_action_t *action = (const pcmk_action_t *) iter->data; if (!pcmk_any_flags_set(action->flags, flags)) { pcmk__log_action("\t", action, true); } } } /*! * \internal * \brief Unpack the CIB for scheduling * * \param[in,out] cib CIB XML to unpack (may be NULL if already unpacked) * \param[in] flags Scheduler flags to set in addition to defaults * \param[in,out] scheduler Scheduler data */ static void unpack_cib(xmlNode *cib, unsigned long long flags, pcmk_scheduler_t *scheduler) { if (pcmk_is_set(scheduler->flags, pcmk__sched_have_status)) { crm_trace("Reusing previously calculated cluster status"); pcmk__set_scheduler_flags(scheduler, flags); return; } - pcmk__assert(cib != NULL); crm_trace("Calculating cluster status"); - - /* This will zero the entire struct without freeing anything first, so - * callers should never call pcmk__schedule_actions() with a populated data - * set unless pcmk__sched_have_status is set (i.e. cluster_status() was - * previously called, whether directly or via pcmk__schedule_actions()). - */ - set_working_set_defaults(scheduler); - + pcmk_reset_scheduler(scheduler); pcmk__set_scheduler_flags(scheduler, flags); scheduler->input = cib; cluster_status(scheduler); // Sets pcmk__sched_have_status } /*! * \internal * \brief Run the scheduler for a given CIB * * \param[in,out] cib CIB XML to use as scheduler input * \param[in] flags Scheduler flags to set in addition to defaults * \param[in,out] scheduler Scheduler data */ void pcmk__schedule_actions(xmlNode *cib, unsigned long long flags, pcmk_scheduler_t *scheduler) { unpack_cib(cib, flags, scheduler); pcmk__set_assignment_methods(scheduler); pcmk__apply_node_health(scheduler); pcmk__unpack_constraints(scheduler); if (pcmk_is_set(scheduler->flags, pcmk__sched_validate_only)) { return; } if (!pcmk_is_set(scheduler->flags, pcmk__sched_location_only) && pcmk__is_daemon) { log_resource_details(scheduler); } apply_node_criteria(scheduler); if (pcmk_is_set(scheduler->flags, pcmk__sched_location_only)) { return; } pcmk__create_internal_constraints(scheduler); pcmk__handle_rsc_config_changes(scheduler); assign_resources(scheduler); schedule_resource_actions(scheduler); /* Remote ordering constraints need to happen prior to calculating fencing * because it is one more place we can mark nodes as needing fencing. */ pcmk__order_remote_connection_actions(scheduler); schedule_fencing_and_shutdowns(scheduler); pcmk__apply_orderings(scheduler); log_all_actions(scheduler); pcmk__create_graph(scheduler); if (get_crm_log_level() == LOG_TRACE) { log_unrunnable_actions(scheduler); } } /*! * \internal * \brief Initialize scheduler data * * Make our own copies of the CIB XML and date/time object, if they're not * \c NULL. This way we don't have to take ownership of the objects passed via * the API. * * This function is most useful for public API functions that want the caller * to retain ownership of the CIB object * * \param[in,out] out Output object * \param[in] input The CIB XML to check (if \c NULL, use current CIB) * \param[in] date Date and time to use in the scheduler (if \c NULL, * use current date and time). This can be used for * checking whether a rule is in effect at a certa * date and time. * \param[out] scheduler Where to store initialized scheduler data * * \return Standard Pacemaker return code */ int pcmk__init_scheduler(pcmk__output_t *out, xmlNodePtr input, const crm_time_t *date, pcmk_scheduler_t **scheduler) { // Allows for cleaner syntax than dereferencing the scheduler argument pcmk_scheduler_t *new_scheduler = NULL; - new_scheduler = pe_new_working_set(); + new_scheduler = pcmk_new_scheduler(); if (new_scheduler == NULL) { return ENOMEM; } pcmk__set_scheduler_flags(new_scheduler, pcmk__sched_no_counts); // Populate the scheduler data // Make our own copy of the given input or fetch the CIB and use that if (input != NULL) { new_scheduler->input = pcmk__xml_copy(NULL, input); if (new_scheduler->input == NULL) { out->err(out, "Failed to copy input XML"); - pe_free_working_set(new_scheduler); + pcmk_free_scheduler(new_scheduler); return ENOMEM; } } else { int rc = cib__signon_query(out, NULL, &(new_scheduler->input)); if (rc != pcmk_rc_ok) { - pe_free_working_set(new_scheduler); + pcmk_free_scheduler(new_scheduler); return rc; } } // Make our own copy of the given crm_time_t object; otherwise // cluster_status() populates with the current time if (date != NULL) { // pcmk_copy_time() guarantees non-NULL new_scheduler->priv->now = pcmk_copy_time(date); } // Unpack everything cluster_status(new_scheduler); *scheduler = new_scheduler; return pcmk_rc_ok; } diff --git a/lib/pacemaker/pcmk_simulate.c b/lib/pacemaker/pcmk_simulate.c index dadf52c1e3..d076aa88a9 100644 --- a/lib/pacemaker/pcmk_simulate.c +++ b/lib/pacemaker/pcmk_simulate.c @@ -1,1017 +1,1017 @@ /* * Copyright 2021-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include #include #include "libpacemaker_private.h" static pcmk__output_t *out = NULL; static cib_t *fake_cib = NULL; static GList *fake_resource_list = NULL; static const GList *fake_op_fail_list = NULL; static void set_effective_date(pcmk_scheduler_t *scheduler, bool print_original, const char *use_date); /*! * \internal * \brief Create an action name for use in a dot graph * * \param[in] action Action to create name for * \param[in] verbose If true, add action ID to name * * \return Newly allocated string with action name * \note It is the caller's responsibility to free the result. */ static char * create_action_name(const pcmk_action_t *action, bool verbose) { char *action_name = NULL; const char *prefix = ""; const char *action_host = NULL; const char *history_id = NULL; const char *task = action->task; if (action->node != NULL) { action_host = action->node->priv->name; } else if (!pcmk_is_set(action->flags, pcmk__action_pseudo)) { action_host = ""; } if (pcmk__str_eq(action->task, PCMK_ACTION_CANCEL, pcmk__str_none)) { prefix = "Cancel "; task = action->cancel_task; } if (action->rsc != NULL) { history_id = action->rsc->priv->history_id; } if (history_id != NULL) { char *key = NULL; guint interval_ms = 0; if (pcmk__guint_from_hash(action->meta, PCMK_META_INTERVAL, 0, &interval_ms) != pcmk_rc_ok) { interval_ms = 0; } if (pcmk__strcase_any_of(action->task, PCMK_ACTION_NOTIFY, PCMK_ACTION_NOTIFIED, NULL)) { const char *n_type = g_hash_table_lookup(action->meta, "notify_key_type"); const char *n_task = g_hash_table_lookup(action->meta, "notify_key_operation"); pcmk__assert((n_type != NULL) && (n_task != NULL)); key = pcmk__notify_key(history_id, n_type, n_task); } else { key = pcmk__op_key(history_id, task, interval_ms); } if (action_host != NULL) { action_name = crm_strdup_printf("%s%s %s", prefix, key, action_host); } else { action_name = crm_strdup_printf("%s%s", prefix, key); } free(key); } else if (pcmk__str_eq(action->task, PCMK_ACTION_STONITH, pcmk__str_none)) { const char *op = g_hash_table_lookup(action->meta, PCMK__META_STONITH_ACTION); action_name = crm_strdup_printf("%s%s '%s' %s", prefix, action->task, op, action_host); } else if (action->rsc && action_host) { action_name = crm_strdup_printf("%s%s %s", prefix, action->uuid, action_host); } else if (action_host) { action_name = crm_strdup_printf("%s%s %s", prefix, action->task, action_host); } else { action_name = crm_strdup_printf("%s", action->uuid); } if (verbose) { char *with_id = crm_strdup_printf("%s (%d)", action_name, action->id); free(action_name); action_name = with_id; } return action_name; } /*! * \internal * \brief Display the status of a cluster * * \param[in,out] scheduler Scheduler data * \param[in] show_opts How to modify display (as pcmk_show_opt_e flags) * \param[in] section_opts Sections to display (as pcmk_section_e flags) * \param[in] title What to use as list title * \param[in] print_spacer Whether to display a spacer first */ static void print_cluster_status(pcmk_scheduler_t *scheduler, uint32_t show_opts, uint32_t section_opts, const char *title, bool print_spacer) { pcmk__output_t *out = scheduler->priv->out; GList *all = NULL; crm_exit_t stonith_rc = 0; enum pcmk_pacemakerd_state state = pcmk_pacemakerd_state_invalid; section_opts |= pcmk_section_nodes | pcmk_section_resources; show_opts |= pcmk_show_inactive_rscs | pcmk_show_failed_detail; all = g_list_prepend(all, (gpointer) "*"); PCMK__OUTPUT_SPACER_IF(out, print_spacer); out->begin_list(out, NULL, NULL, "%s", title); out->message(out, "cluster-status", scheduler, state, stonith_rc, NULL, pcmk__fence_history_none, section_opts, show_opts, NULL, all, all); out->end_list(out); g_list_free(all); } /*! * \internal * \brief Display a summary of all actions scheduled in a transition * * \param[in,out] scheduler Scheduler data (fully scheduled) * \param[in] print_spacer Whether to display a spacer first */ static void print_transition_summary(pcmk_scheduler_t *scheduler, bool print_spacer) { pcmk__output_t *out = scheduler->priv->out; PCMK__OUTPUT_SPACER_IF(out, print_spacer); out->begin_list(out, NULL, NULL, "Transition Summary"); pcmk__output_actions(scheduler); out->end_list(out); } /*! * \internal * \brief Reset scheduler input, output, date, and flags * * \param[in,out] scheduler Scheduler data * \param[in] input What to set as cluster input * \param[in] out What to set as cluster output object * \param[in] use_date What to set as cluster's current timestamp * \param[in] flags Group of enum pcmk__scheduler_flags to set */ static void reset(pcmk_scheduler_t *scheduler, xmlNodePtr input, pcmk__output_t *out, const char *use_date, unsigned int flags) { scheduler->input = input; scheduler->priv->out = out; set_effective_date(scheduler, true, use_date); if (pcmk_is_set(flags, pcmk_sim_sanitized)) { pcmk__set_scheduler_flags(scheduler, pcmk__sched_sanitized); } if (pcmk_is_set(flags, pcmk_sim_show_scores)) { pcmk__set_scheduler_flags(scheduler, pcmk__sched_output_scores); } if (pcmk_is_set(flags, pcmk_sim_show_utilization)) { pcmk__set_scheduler_flags(scheduler, pcmk__sched_show_utilization); } } /*! * \brief Write out a file in dot(1) format describing the actions that will * be taken by the scheduler in response to an input CIB file. * * \param[in,out] scheduler Scheduler data * \param[in] dot_file The filename to write * \param[in] all_actions Write all actions, even those that are optional * or are on unmanaged resources * \param[in] verbose Add extra information, such as action IDs, to the * output * * \return Standard Pacemaker return code */ static int write_sim_dotfile(pcmk_scheduler_t *scheduler, const char *dot_file, bool all_actions, bool verbose) { GList *iter = NULL; FILE *dot_strm = fopen(dot_file, "w"); if (dot_strm == NULL) { return errno; } fprintf(dot_strm, " digraph \"g\" {\n"); for (iter = scheduler->priv->actions; iter != NULL; iter = iter->next) { pcmk_action_t *action = (pcmk_action_t *) iter->data; const char *style = "dashed"; const char *font = "black"; const char *color = "black"; char *action_name = create_action_name(action, verbose); if (pcmk_is_set(action->flags, pcmk__action_pseudo)) { font = "orange"; } if (pcmk_is_set(action->flags, pcmk__action_added_to_graph)) { style = PCMK__VALUE_BOLD; color = "green"; } else if ((action->rsc != NULL) && !pcmk_is_set(action->rsc->flags, pcmk__rsc_managed)) { color = "red"; font = "purple"; if (!all_actions) { goto do_not_write; } } else if (pcmk_is_set(action->flags, pcmk__action_optional)) { color = "blue"; if (!all_actions) { goto do_not_write; } } else { color = "red"; CRM_LOG_ASSERT(!pcmk_is_set(action->flags, pcmk__action_runnable)); } pcmk__set_action_flags(action, pcmk__action_added_to_graph); fprintf(dot_strm, "\"%s\" [ style=%s color=\"%s\" fontcolor=\"%s\"]\n", action_name, style, color, font); do_not_write: free(action_name); } for (iter = scheduler->priv->actions; iter != NULL; iter = iter->next) { pcmk_action_t *action = (pcmk_action_t *) iter->data; for (GList *before_iter = action->actions_before; before_iter != NULL; before_iter = before_iter->next) { pcmk__related_action_t *before = before_iter->data; char *before_name = NULL; char *after_name = NULL; const char *style = "dashed"; bool optional = true; if (before->graphed) { optional = false; style = PCMK__VALUE_BOLD; } else if (before->flags == pcmk__ar_none) { continue; } else if (pcmk_is_set(before->action->flags, pcmk__action_added_to_graph) && pcmk_is_set(action->flags, pcmk__action_added_to_graph) && before->flags != pcmk__ar_if_on_same_node_or_target) { optional = false; } if (all_actions || !optional) { before_name = create_action_name(before->action, verbose); after_name = create_action_name(action, verbose); fprintf(dot_strm, "\"%s\" -> \"%s\" [ style = %s]\n", before_name, after_name, style); free(before_name); free(after_name); } } } fprintf(dot_strm, "}\n"); fflush(dot_strm); fclose(dot_strm); return pcmk_rc_ok; } /*! * \brief Profile the configuration updates and scheduler actions in a single * CIB file, printing the profiling timings. * * \note \p scheduler->priv->out must have been set to a valid \p pcmk__output_t * object before this function is called. * * \param[in] xml_file The CIB file to profile * \param[in] repeat Number of times to run * \param[in,out] scheduler Scheduler data * \param[in] use_date The date to set the cluster's time to (may be NULL) */ static void profile_file(const char *xml_file, long long repeat, pcmk_scheduler_t *scheduler, const char *use_date) { pcmk__output_t *out = scheduler->priv->out; xmlNode *cib_object = NULL; clock_t start = 0; clock_t end; unsigned long long scheduler_flags = pcmk__sched_none; pcmk__assert(out != NULL); cib_object = pcmk__xml_read(xml_file); start = clock(); if (pcmk_find_cib_element(cib_object, PCMK_XE_STATUS) == NULL) { pcmk__xe_create(cib_object, PCMK_XE_STATUS); } if (pcmk__update_configured_schema(&cib_object, false) != pcmk_rc_ok) { pcmk__xml_free(cib_object); return; } if (!pcmk__validate_xml(cib_object, NULL, NULL, NULL)) { pcmk__xml_free(cib_object); return; } if (pcmk_is_set(scheduler->flags, pcmk__sched_output_scores)) { scheduler_flags |= pcmk__sched_output_scores; } if (pcmk_is_set(scheduler->flags, pcmk__sched_show_utilization)) { scheduler_flags |= pcmk__sched_show_utilization; } for (int i = 0; i < repeat; ++i) { xmlNode *input = cib_object; if (repeat > 1) { input = pcmk__xml_copy(NULL, cib_object); } scheduler->input = input; set_effective_date(scheduler, false, use_date); pcmk__schedule_actions(input, scheduler_flags, scheduler); - pe_reset_working_set(scheduler); + pcmk_reset_scheduler(scheduler); } end = clock(); out->message(out, "profile", xml_file, start, end); } void pcmk__profile_dir(const char *dir, long long repeat, pcmk_scheduler_t *scheduler, const char *use_date) { pcmk__output_t *out = scheduler->priv->out; struct dirent **namelist; int file_num = scandir(dir, &namelist, 0, alphasort); pcmk__assert(out != NULL); if (file_num > 0) { struct stat prop; char buffer[FILENAME_MAX]; out->begin_list(out, NULL, NULL, "Timings"); while (file_num--) { if ('.' == namelist[file_num]->d_name[0]) { free(namelist[file_num]); continue; } else if (!pcmk__ends_with_ext(namelist[file_num]->d_name, ".xml")) { free(namelist[file_num]); continue; } snprintf(buffer, sizeof(buffer), "%s/%s", dir, namelist[file_num]->d_name); if (stat(buffer, &prop) == 0 && S_ISREG(prop.st_mode)) { profile_file(buffer, repeat, scheduler, use_date); } free(namelist[file_num]); } free(namelist); out->end_list(out); } } /*! * \brief Set the date of the cluster, either to the value given by * \p use_date, or to the \c PCMK_XA_EXECUTION_DATE value in the CIB. * * \note \p scheduler->priv->out must have been set to a valid \p pcmk__output_t * object before this function is called. * * \param[in,out] scheduler Scheduler data * \param[in] print_original If \p true, the \c PCMK_XA_EXECUTION_DATE * should also be printed * \param[in] use_date The date to set the cluster's time to * (may be NULL) */ static void set_effective_date(pcmk_scheduler_t *scheduler, bool print_original, const char *use_date) { pcmk__output_t *out = scheduler->priv->out; time_t original_date = 0; pcmk__assert(out != NULL); crm_element_value_epoch(scheduler->input, PCMK_XA_EXECUTION_DATE, &original_date); if (use_date) { scheduler->priv->now = crm_time_new(use_date); out->info(out, "Setting effective cluster time: %s", use_date); crm_time_log(LOG_NOTICE, "Pretending 'now' is", scheduler->priv->now, crm_time_log_date | crm_time_log_timeofday); } else if (original_date != 0) { scheduler->priv->now = pcmk__copy_timet(original_date); if (print_original) { char *when = crm_time_as_string(scheduler->priv->now, crm_time_log_date |crm_time_log_timeofday); out->info(out, "Using the original execution date of: %s", when); free(when); } } } /*! * \internal * \brief Simulate successfully executing a pseudo-action in a graph * * \param[in,out] graph Graph to update with pseudo-action result * \param[in,out] action Pseudo-action to simulate executing * * \return Standard Pacemaker return code */ static int simulate_pseudo_action(pcmk__graph_t *graph, pcmk__graph_action_t *action) { const char *node = crm_element_value(action->xml, PCMK__META_ON_NODE); const char *task = crm_element_value(action->xml, PCMK__XA_OPERATION_KEY); pcmk__set_graph_action_flags(action, pcmk__graph_action_confirmed); out->message(out, "inject-pseudo-action", node, task); pcmk__update_graph(graph, action); return pcmk_rc_ok; } /*! * \internal * \brief Simulate executing a resource action in a graph * * \param[in,out] graph Graph to update with resource action result * \param[in,out] action Resource action to simulate executing * * \return Standard Pacemaker return code */ static int simulate_resource_action(pcmk__graph_t *graph, pcmk__graph_action_t *action) { int rc; lrmd_event_data_t *op = NULL; int target_outcome = PCMK_OCF_OK; const char *rtype = NULL; const char *rclass = NULL; const char *resource = NULL; const char *rprovider = NULL; const char *resource_config_name = NULL; const char *operation = crm_element_value(action->xml, PCMK_XA_OPERATION); const char *target_rc_s = crm_meta_value(action->params, PCMK__META_OP_TARGET_RC); xmlNode *cib_node = NULL; xmlNode *cib_resource = NULL; xmlNode *action_rsc = pcmk__xe_first_child(action->xml, PCMK_XE_PRIMITIVE, NULL, NULL); char *node = crm_element_value_copy(action->xml, PCMK__META_ON_NODE); char *uuid = NULL; const char *router_node = crm_element_value(action->xml, PCMK__XA_ROUTER_NODE); // Certain actions don't need to be displayed or history entries if (pcmk__str_eq(operation, CRM_OP_REPROBE, pcmk__str_none)) { crm_debug("No history injection for %s op on %s", operation, node); goto done; // Confirm action and update graph } if (action_rsc == NULL) { // Shouldn't be possible crm_log_xml_err(action->xml, "Bad"); free(node); return EPROTO; } /* A resource might be known by different names in the configuration and in * the action (for example, a clone instance). Grab the configuration name * (which is preferred when writing history), and if necessary, the instance * name. */ resource_config_name = crm_element_value(action_rsc, PCMK_XA_ID); if (resource_config_name == NULL) { // Shouldn't be possible crm_log_xml_err(action->xml, "No ID"); free(node); return EPROTO; } resource = resource_config_name; if (pe_find_resource(fake_resource_list, resource) == NULL) { const char *longname = crm_element_value(action_rsc, PCMK__XA_LONG_ID); if ((longname != NULL) && (pe_find_resource(fake_resource_list, longname) != NULL)) { resource = longname; } } // Certain actions need to be displayed but don't need history entries if (pcmk__strcase_any_of(operation, PCMK_ACTION_DELETE, PCMK_ACTION_META_DATA, NULL)) { out->message(out, "inject-rsc-action", resource, operation, node, (guint) 0); goto done; // Confirm action and update graph } rclass = crm_element_value(action_rsc, PCMK_XA_CLASS); rtype = crm_element_value(action_rsc, PCMK_XA_TYPE); rprovider = crm_element_value(action_rsc, PCMK_XA_PROVIDER); pcmk__scan_min_int(target_rc_s, &target_outcome, 0); pcmk__assert(fake_cib->cmds->query(fake_cib, NULL, NULL, cib_sync_call) == pcmk_ok); // Ensure the action node is in the CIB uuid = crm_element_value_copy(action->xml, PCMK__META_ON_NODE_UUID); cib_node = pcmk__inject_node(fake_cib, node, ((router_node == NULL)? uuid: node)); free(uuid); pcmk__assert(cib_node != NULL); // Add a history entry for the action cib_resource = pcmk__inject_resource_history(out, cib_node, resource, resource_config_name, rclass, rtype, rprovider); if (cib_resource == NULL) { crm_err("Could not simulate action %d history for resource %s", action->id, resource); free(node); pcmk__xml_free(cib_node); return EINVAL; } // Simulate and display an executor event for the action result op = pcmk__event_from_graph_action(cib_resource, action, PCMK_EXEC_DONE, target_outcome, "User-injected result"); out->message(out, "inject-rsc-action", resource, op->op_type, node, op->interval_ms); // Check whether action is in a list of desired simulated failures for (const GList *iter = fake_op_fail_list; iter != NULL; iter = iter->next) { const char *spec = (const char *) iter->data; char *key = NULL; const char *match_name = NULL; const char *offset = NULL; // Allow user to specify anonymous clone with or without instance number key = crm_strdup_printf(PCMK__OP_FMT "@%s=", resource, op->op_type, op->interval_ms, node); if (strncasecmp(key, spec, strlen(key)) == 0) { match_name = resource; } free(key); // If not found, try the resource's name in the configuration if ((match_name == NULL) && (strcmp(resource, resource_config_name) != 0)) { key = crm_strdup_printf(PCMK__OP_FMT "@%s=", resource_config_name, op->op_type, op->interval_ms, node); if (strncasecmp(key, spec, strlen(key)) == 0) { match_name = resource_config_name; } free(key); } if (match_name == NULL) { continue; // This failed action entry doesn't match } // ${match_name}_${task}_${interval_in_ms}@${node}=${rc} rc = sscanf(spec, "%*[^=]=%d", (int *) &op->rc); if (rc != 1) { out->err(out, "Invalid failed operation '%s' " "(result code must be integer)", spec); continue; // Keep checking other list entries } out->info(out, "Pretending action %d failed with rc=%d", action->id, op->rc); pcmk__set_graph_action_flags(action, pcmk__graph_action_failed); graph->abort_priority = PCMK_SCORE_INFINITY; if (pcmk__str_eq(op->op_type, PCMK_ACTION_START, pcmk__str_none)) { offset = pcmk__s(graph->failed_start_offset, PCMK_VALUE_INFINITY); } else if (pcmk__str_eq(op->op_type, PCMK_ACTION_STOP, pcmk__str_none)) { offset = pcmk__s(graph->failed_stop_offset, PCMK_VALUE_INFINITY); } pcmk__inject_failcount(out, fake_cib, cib_node, match_name, op->op_type, op->interval_ms, op->rc, pcmk_str_is_infinity(offset)); break; } pcmk__inject_action_result(cib_resource, op, node, target_outcome); lrmd_free_event(op); rc = fake_cib->cmds->modify(fake_cib, PCMK_XE_STATUS, cib_node, cib_sync_call); pcmk__assert(rc == pcmk_ok); done: free(node); pcmk__xml_free(cib_node); pcmk__set_graph_action_flags(action, pcmk__graph_action_confirmed); pcmk__update_graph(graph, action); return pcmk_rc_ok; } /*! * \internal * \brief Simulate successfully executing a cluster action * * \param[in,out] graph Graph to update with action result * \param[in,out] action Cluster action to simulate * * \return Standard Pacemaker return code */ static int simulate_cluster_action(pcmk__graph_t *graph, pcmk__graph_action_t *action) { const char *node = crm_element_value(action->xml, PCMK__META_ON_NODE); const char *task = crm_element_value(action->xml, PCMK_XA_OPERATION); xmlNode *rsc = pcmk__xe_first_child(action->xml, PCMK_XE_PRIMITIVE, NULL, NULL); pcmk__set_graph_action_flags(action, pcmk__graph_action_confirmed); out->message(out, "inject-cluster-action", node, task, rsc); pcmk__update_graph(graph, action); return pcmk_rc_ok; } /*! * \internal * \brief Simulate successfully executing a fencing action * * \param[in,out] graph Graph to update with action result * \param[in,out] action Fencing action to simulate * * \return Standard Pacemaker return code */ static int simulate_fencing_action(pcmk__graph_t *graph, pcmk__graph_action_t *action) { const char *op = crm_meta_value(action->params, PCMK__META_STONITH_ACTION); char *target = crm_element_value_copy(action->xml, PCMK__META_ON_NODE); out->message(out, "inject-fencing-action", target, op); if (!pcmk__str_eq(op, PCMK_ACTION_ON, pcmk__str_casei)) { int rc = pcmk_ok; GString *xpath = g_string_sized_new(512); // Set node state to offline xmlNode *cib_node = pcmk__inject_node_state_change(fake_cib, target, false); pcmk__assert(cib_node != NULL); crm_xml_add(cib_node, PCMK_XA_CRM_DEBUG_ORIGIN, __func__); rc = fake_cib->cmds->replace(fake_cib, PCMK_XE_STATUS, cib_node, cib_sync_call); pcmk__assert(rc == pcmk_ok); // Simulate controller clearing node's resource history and attributes pcmk__g_strcat(xpath, "//" PCMK__XE_NODE_STATE "[@" PCMK_XA_UNAME "='", target, "']/" PCMK__XE_LRM, NULL); fake_cib->cmds->remove(fake_cib, (const char *) xpath->str, NULL, cib_xpath|cib_sync_call); g_string_truncate(xpath, 0); pcmk__g_strcat(xpath, "//" PCMK__XE_NODE_STATE "[@" PCMK_XA_UNAME "='", target, "']" "/" PCMK__XE_TRANSIENT_ATTRIBUTES, NULL); fake_cib->cmds->remove(fake_cib, (const char *) xpath->str, NULL, cib_xpath|cib_sync_call); pcmk__xml_free(cib_node); g_string_free(xpath, TRUE); } pcmk__set_graph_action_flags(action, pcmk__graph_action_confirmed); pcmk__update_graph(graph, action); free(target); return pcmk_rc_ok; } enum pcmk__graph_status pcmk__simulate_transition(pcmk_scheduler_t *scheduler, cib_t *cib, const GList *op_fail_list) { pcmk__graph_t *transition = NULL; enum pcmk__graph_status graph_rc; pcmk__graph_functions_t simulation_fns = { simulate_pseudo_action, simulate_resource_action, simulate_cluster_action, simulate_fencing_action, }; out = scheduler->priv->out; fake_cib = cib; fake_op_fail_list = op_fail_list; if (!out->is_quiet(out)) { out->begin_list(out, NULL, NULL, "Executing Cluster Transition"); } pcmk__set_graph_functions(&simulation_fns); transition = pcmk__unpack_graph(scheduler->priv->graph, crm_system_name); pcmk__log_graph(LOG_DEBUG, transition); fake_resource_list = scheduler->priv->resources; do { graph_rc = pcmk__execute_graph(transition); } while (graph_rc == pcmk__graph_active); fake_resource_list = NULL; if (graph_rc != pcmk__graph_complete) { out->err(out, "Transition failed: %s", pcmk__graph_status2text(graph_rc)); pcmk__log_graph(LOG_ERR, transition); out->err(out, "An invalid transition was produced"); } pcmk__free_graph(transition); if (!out->is_quiet(out)) { // If not quiet, we'll need the resulting CIB for later display xmlNode *cib_object = NULL; int rc = fake_cib->cmds->query(fake_cib, NULL, &cib_object, cib_sync_call); pcmk__assert(rc == pcmk_ok); - pe_reset_working_set(scheduler); + pcmk_reset_scheduler(scheduler); scheduler->input = cib_object; out->end_list(out); } return graph_rc; } int pcmk__simulate(pcmk_scheduler_t *scheduler, pcmk__output_t *out, const pcmk_injections_t *injections, unsigned int flags, uint32_t section_opts, const char *use_date, const char *input_file, const char *graph_file, const char *dot_file) { int printed = pcmk_rc_no_output; int rc = pcmk_rc_ok; xmlNodePtr input = NULL; cib_t *cib = NULL; rc = cib__signon_query(out, &cib, &input); if (rc != pcmk_rc_ok) { goto simulate_done; } reset(scheduler, input, out, use_date, flags); cluster_status(scheduler); if (!out->is_quiet(out)) { const bool show_pending = pcmk_is_set(flags, pcmk_sim_show_pending); if (pcmk_is_set(scheduler->flags, pcmk__sched_in_maintenance)) { printed = out->message(out, "maint-mode", scheduler->flags); } if ((scheduler->priv->disabled_resources > 0) || (scheduler->priv->blocked_resources > 0)) { PCMK__OUTPUT_SPACER_IF(out, printed == pcmk_rc_ok); printed = out->info(out, "%d of %d resource instances DISABLED and " "%d BLOCKED from further action due to failure", scheduler->priv->disabled_resources, scheduler->priv->ninstances, scheduler->priv->blocked_resources); } /* Most formatted output headers use caps for each word, but this one * only has the first word capitalized for compatibility with pcs. */ print_cluster_status(scheduler, (show_pending? pcmk_show_pending : 0), section_opts, "Current cluster status", (printed == pcmk_rc_ok)); printed = pcmk_rc_ok; } // If the user requested any injections, handle them if ((injections->node_down != NULL) || (injections->node_fail != NULL) || (injections->node_up != NULL) || (injections->op_inject != NULL) || (injections->ticket_activate != NULL) || (injections->ticket_grant != NULL) || (injections->ticket_revoke != NULL) || (injections->ticket_standby != NULL) || (injections->watchdog != NULL)) { PCMK__OUTPUT_SPACER_IF(out, printed == pcmk_rc_ok); pcmk__inject_scheduler_input(scheduler, cib, injections); printed = pcmk_rc_ok; rc = cib->cmds->query(cib, NULL, &input, cib_sync_call); if (rc != pcmk_rc_ok) { rc = pcmk_legacy2rc(rc); goto simulate_done; } - cleanup_calculations(scheduler); + pcmk_reset_scheduler(scheduler); reset(scheduler, input, out, use_date, flags); cluster_status(scheduler); } if (input_file != NULL) { rc = pcmk__xml_write_file(input, input_file, false); if (rc != pcmk_rc_ok) { goto simulate_done; } } if (pcmk_any_flags_set(flags, pcmk_sim_process | pcmk_sim_simulate)) { pcmk__output_t *logger_out = NULL; unsigned long long scheduler_flags = pcmk__sched_none; if (pcmk_is_set(scheduler->flags, pcmk__sched_output_scores)) { scheduler_flags |= pcmk__sched_output_scores; } if (pcmk_is_set(scheduler->flags, pcmk__sched_show_utilization)) { scheduler_flags |= pcmk__sched_show_utilization; } if (pcmk_all_flags_set(scheduler->flags, pcmk__sched_output_scores |pcmk__sched_show_utilization)) { PCMK__OUTPUT_SPACER_IF(out, printed == pcmk_rc_ok); out->begin_list(out, NULL, NULL, "Assignment Scores and Utilization Information"); printed = pcmk_rc_ok; } else if (pcmk_is_set(scheduler->flags, pcmk__sched_output_scores)) { PCMK__OUTPUT_SPACER_IF(out, printed == pcmk_rc_ok); out->begin_list(out, NULL, NULL, "Assignment Scores"); printed = pcmk_rc_ok; } else if (pcmk_is_set(scheduler->flags, pcmk__sched_show_utilization)) { PCMK__OUTPUT_SPACER_IF(out, printed == pcmk_rc_ok); out->begin_list(out, NULL, NULL, "Utilization Information"); printed = pcmk_rc_ok; } else { rc = pcmk__log_output_new(&logger_out); if (rc != pcmk_rc_ok) { goto simulate_done; } pe__register_messages(logger_out); pcmk__register_lib_messages(logger_out); scheduler->priv->out = logger_out; } pcmk__schedule_actions(input, scheduler_flags, scheduler); if (logger_out == NULL) { out->end_list(out); } else { logger_out->finish(logger_out, CRM_EX_OK, true, NULL); pcmk__output_free(logger_out); scheduler->priv->out = out; } input = NULL; /* Don't try and free it twice */ if (graph_file != NULL) { rc = pcmk__xml_write_file(scheduler->priv->graph, graph_file, false); if (rc != pcmk_rc_ok) { rc = pcmk_rc_graph_error; goto simulate_done; } } if (dot_file != NULL) { rc = write_sim_dotfile(scheduler, dot_file, pcmk_is_set(flags, pcmk_sim_all_actions), pcmk_is_set(flags, pcmk_sim_verbose)); if (rc != pcmk_rc_ok) { rc = pcmk_rc_dot_error; goto simulate_done; } } if (!out->is_quiet(out)) { print_transition_summary(scheduler, printed == pcmk_rc_ok); } } rc = pcmk_rc_ok; if (!pcmk_is_set(flags, pcmk_sim_simulate)) { goto simulate_done; } PCMK__OUTPUT_SPACER_IF(out, printed == pcmk_rc_ok); if (pcmk__simulate_transition(scheduler, cib, injections->op_fail) != pcmk__graph_complete) { rc = pcmk_rc_invalid_transition; } if (out->is_quiet(out)) { goto simulate_done; } set_effective_date(scheduler, true, use_date); if (pcmk_is_set(flags, pcmk_sim_show_scores)) { pcmk__set_scheduler_flags(scheduler, pcmk__sched_output_scores); } if (pcmk_is_set(flags, pcmk_sim_show_utilization)) { pcmk__set_scheduler_flags(scheduler, pcmk__sched_show_utilization); } cluster_status(scheduler); print_cluster_status(scheduler, 0, section_opts, "Revised Cluster Status", true); simulate_done: cib__clean_up_connection(&cib); return rc; } int pcmk_simulate(xmlNodePtr *xml, pcmk_scheduler_t *scheduler, const pcmk_injections_t *injections, unsigned int flags, unsigned int section_opts, const char *use_date, const char *input_file, const char *graph_file, const char *dot_file) { pcmk__output_t *out = NULL; int rc = pcmk_rc_ok; rc = pcmk__xml_output_new(&out, xml); if (rc != pcmk_rc_ok) { return rc; } pe__register_messages(out); pcmk__register_lib_messages(out); rc = pcmk__simulate(scheduler, out, injections, flags, section_opts, use_date, input_file, graph_file, dot_file); pcmk__xml_output_finish(out, pcmk_rc2exitc(rc), xml); return rc; } diff --git a/lib/pacemaker/pcmk_status.c b/lib/pacemaker/pcmk_status.c index 002817c06a..e111456279 100644 --- a/lib/pacemaker/pcmk_status.c +++ b/lib/pacemaker/pcmk_status.c @@ -1,283 +1,283 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include // stonith__register_messages() #include #include static stonith_t * fencing_connect(void) { stonith_t *st = stonith_api_new(); int rc = pcmk_rc_ok; if (st == NULL) { return NULL; } rc = st->cmds->connect(st, crm_system_name, NULL); if (rc == pcmk_rc_ok) { return st; } else { stonith_api_delete(st); return NULL; } } /*! * \internal * \brief Output the cluster status given a fencer and CIB connection * * \param[in,out] scheduler Scheduler object (will be reset) * \param[in,out] stonith Fencer connection * \param[in,out] cib CIB connection * \param[in] current_cib Current CIB XML * \param[in] pcmkd_state \p pacemakerd state * \param[in] fence_history How much of the fencing history to output * \param[in] show Group of \p pcmk_section_e flags * \param[in] show_opts Group of \p pcmk_show_opt_e flags * \param[in] only_node If a node name or tag, include only the * matching node(s) (if any) in the output. * If \p "*" or \p NULL, include all nodes * in the output. * \param[in] only_rsc If a resource ID or tag, include only the * matching resource(s) (if any) in the * output. If \p "*" or \p NULL, include all * resources in the output. * \param[in] neg_location_prefix Prefix denoting a ban in a constraint ID * * \return Standard Pacemaker return code */ int pcmk__output_cluster_status(pcmk_scheduler_t *scheduler, stonith_t *stonith, cib_t *cib, xmlNode *current_cib, enum pcmk_pacemakerd_state pcmkd_state, enum pcmk__fence_history fence_history, uint32_t show, uint32_t show_opts, const char *only_node, const char *only_rsc, const char *neg_location_prefix) { xmlNode *cib_copy = pcmk__xml_copy(NULL, current_cib); stonith_history_t *stonith_history = NULL; int history_rc = 0; GList *unames = NULL; GList *resources = NULL; pcmk__output_t *out = NULL; int rc = pcmk_rc_ok; if ((scheduler == NULL) || (scheduler->priv->out == NULL)) { return EINVAL; } out = scheduler->priv->out; rc = pcmk__update_configured_schema(&cib_copy, false); if (rc != pcmk_rc_ok) { cib__clean_up_connection(&cib); pcmk__xml_free(cib_copy); out->err(out, "Upgrade failed: %s", pcmk_rc_str(rc)); return rc; } /* get the stonith-history if there is evidence we need it */ if (fence_history != pcmk__fence_history_none) { history_rc = pcmk__get_fencing_history(stonith, &stonith_history, fence_history); } - pe_reset_working_set(scheduler); + pcmk_reset_scheduler(scheduler); scheduler->input = cib_copy; cluster_status(scheduler); /* Unpack constraints if any section will need them * (tickets may be referenced in constraints but not granted yet, * and bans need negative location constraints) */ if (pcmk_is_set(show, pcmk_section_bans) || pcmk_is_set(show, pcmk_section_tickets)) { pcmk__unpack_constraints(scheduler); } unames = pe__build_node_name_list(scheduler, only_node); resources = pe__build_rsc_list(scheduler, only_rsc); /* Always print DC if NULL. */ if (scheduler->dc_node == NULL) { show |= pcmk_section_dc; } out->message(out, "cluster-status", scheduler, pcmkd_state, pcmk_rc2exitc(history_rc), stonith_history, fence_history, show, show_opts, neg_location_prefix, unames, resources); g_list_free_full(unames, free); g_list_free_full(resources, free); stonith_history_free(stonith_history); stonith_history = NULL; return rc; } int pcmk_status(xmlNodePtr *xml) { cib_t *cib = NULL; pcmk__output_t *out = NULL; int rc = pcmk_rc_ok; uint32_t show_opts = pcmk_show_pending |pcmk_show_inactive_rscs |pcmk_show_timing; cib = cib_new(); if (cib == NULL) { return pcmk_rc_cib_corrupt; } rc = pcmk__xml_output_new(&out, xml); if (rc != pcmk_rc_ok) { cib_delete(cib); return rc; } pcmk__register_lib_messages(out); pe__register_messages(out); stonith__register_messages(out); rc = pcmk__status(out, cib, pcmk__fence_history_full, pcmk_section_all, show_opts, NULL, NULL, NULL, 0); pcmk__xml_output_finish(out, pcmk_rc2exitc(rc), xml); cib_delete(cib); return rc; } /*! * \internal * \brief Query and output the cluster status * * The operation is considered a success if we're able to get the \p pacemakerd * state. If possible, we'll also try to connect to the fencer and CIB and * output their respective status information. * * \param[in,out] out Output object * \param[in,out] cib CIB connection * \param[in] fence_history How much of the fencing history to output * \param[in] show Group of \p pcmk_section_e flags * \param[in] show_opts Group of \p pcmk_show_opt_e flags * \param[in] only_node If a node name or tag, include only the * matching node(s) (if any) in the output. * If \p "*" or \p NULL, include all nodes * in the output. * \param[in] only_rsc If a resource ID or tag, include only the * matching resource(s) (if any) in the * output. If \p "*" or \p NULL, include all * resources in the output. * \param[in] neg_location_prefix Prefix denoting a ban in a constraint ID * \param[in] timeout_ms How long to wait for a reply from the * \p pacemakerd API. If 0, * \p pcmk_ipc_dispatch_sync will be used. * If positive, \p pcmk_ipc_dispatch_main * will be used, and a new mainloop will be * created for this purpose (freed before * return). * * \return Standard Pacemaker return code */ int pcmk__status(pcmk__output_t *out, cib_t *cib, enum pcmk__fence_history fence_history, uint32_t show, uint32_t show_opts, const char *only_node, const char *only_rsc, const char *neg_location_prefix, unsigned int timeout_ms) { xmlNode *current_cib = NULL; int rc = pcmk_rc_ok; stonith_t *stonith = NULL; enum pcmk_pacemakerd_state pcmkd_state = pcmk_pacemakerd_state_invalid; time_t last_updated = 0; pcmk_scheduler_t *scheduler = NULL; if (cib == NULL) { return ENOTCONN; } if (cib->variant == cib_native) { rc = pcmk__pacemakerd_status(out, crm_system_name, timeout_ms, false, &pcmkd_state); if (rc != pcmk_rc_ok) { return rc; } last_updated = time(NULL); switch (pcmkd_state) { case pcmk_pacemakerd_state_running: case pcmk_pacemakerd_state_shutting_down: case pcmk_pacemakerd_state_remote: /* Fencer and CIB may still be available while shutting down or * running on a Pacemaker Remote node */ break; default: // Fencer and CIB are definitely unavailable out->message(out, "pacemakerd-health", NULL, pcmkd_state, NULL, last_updated); return rc; } if (fence_history != pcmk__fence_history_none) { stonith = fencing_connect(); } } rc = cib__signon_query(out, &cib, ¤t_cib); if (rc != pcmk_rc_ok) { if (pcmkd_state != pcmk_pacemakerd_state_invalid) { // Invalid at this point means we didn't query the pcmkd state out->message(out, "pacemakerd-health", NULL, pcmkd_state, NULL, last_updated); } goto done; } - scheduler = pe_new_working_set(); + scheduler = pcmk_new_scheduler(); pcmk__mem_assert(scheduler); scheduler->priv->out = out; if ((cib->variant == cib_native) && pcmk_is_set(show, pcmk_section_times)) { // Currently used only in the times section pcmk__query_node_name(out, 0, &(scheduler->priv->local_node_name), 0); } rc = pcmk__output_cluster_status(scheduler, stonith, cib, current_cib, pcmkd_state, fence_history, show, show_opts, only_node, only_rsc, neg_location_prefix); if (rc != pcmk_rc_ok) { out->err(out, "Error outputting status info from the fencer or CIB"); } done: - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); stonith_api_delete(stonith); pcmk__xml_free(current_cib); return pcmk_rc_ok; } diff --git a/lib/pacemaker/pcmk_ticket.c b/lib/pacemaker/pcmk_ticket.c index 44b9fa56f6..6e9d297938 100644 --- a/lib/pacemaker/pcmk_ticket.c +++ b/lib/pacemaker/pcmk_ticket.c @@ -1,559 +1,559 @@ /* * Copyright 2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include "libpacemaker_private.h" static int build_ticket_modify_xml(cib_t *cib, const char *ticket_id, xmlNode **ticket_state_xml, xmlNode **xml_top) { int rc = pcmk__get_ticket_state(cib, ticket_id, ticket_state_xml); if (rc == pcmk_rc_ok || rc == pcmk_rc_duplicate_id) { /* Ticket(s) found - return their state */ *xml_top = *ticket_state_xml; } else if (rc == ENXIO) { /* No ticket found - build the XML needed to create it */ xmlNode *xml_obj = NULL; *xml_top = pcmk__xe_create(NULL, PCMK_XE_STATUS); xml_obj = pcmk__xe_create(*xml_top, PCMK_XE_TICKETS); *ticket_state_xml = pcmk__xe_create(xml_obj, PCMK__XE_TICKET_STATE); crm_xml_add(*ticket_state_xml, PCMK_XA_ID, ticket_id); rc = pcmk_rc_ok; } else { /* Some other error occurred - clean up and return */ pcmk__xml_free(*ticket_state_xml); } return rc; } static void add_attribute_xml(pcmk_scheduler_t *scheduler, const char *ticket_id, GHashTable *attr_set, xmlNode **ticket_state_xml) { GHashTableIter hash_iter; char *key = NULL; char *value = NULL; pcmk__ticket_t *ticket = NULL; ticket = g_hash_table_lookup(scheduler->priv->ticket_constraints, ticket_id); g_hash_table_iter_init(&hash_iter, attr_set); while (g_hash_table_iter_next(&hash_iter, (gpointer *) & key, (gpointer *) & value)) { crm_xml_add(*ticket_state_xml, key, value); if (pcmk__str_eq(key, PCMK__XA_GRANTED, pcmk__str_none) && ((ticket == NULL) || !pcmk_is_set(ticket->flags, pcmk__ticket_granted)) && crm_is_true(value)) { char *now = pcmk__ttoa(time(NULL)); crm_xml_add(*ticket_state_xml, PCMK_XA_LAST_GRANTED, now); free(now); } } } int pcmk__get_ticket_state(cib_t *cib, const char *ticket_id, xmlNode **state) { int rc = pcmk_rc_ok; xmlNode *xml_search = NULL; char *xpath = NULL; pcmk__assert((cib != NULL) && (state != NULL)); *state = NULL; if (ticket_id != NULL) { xpath = crm_strdup_printf("/" PCMK_XE_CIB "/" PCMK_XE_STATUS "/" PCMK_XE_TICKETS "/" PCMK__XE_TICKET_STATE "[@" PCMK_XA_ID "=\"%s\"]", ticket_id); } else { xpath = crm_strdup_printf("/" PCMK_XE_CIB "/" PCMK_XE_STATUS "/" PCMK_XE_TICKETS); } rc = cib->cmds->query(cib, xpath, &xml_search, cib_sync_call|cib_xpath); rc = pcmk_legacy2rc(rc); if (rc == pcmk_rc_ok) { crm_log_xml_debug(xml_search, "Match"); if (xml_search->children != NULL && ticket_id != NULL) { rc = pcmk_rc_duplicate_id; } } free(xpath); *state = xml_search; return rc; } int pcmk__ticket_constraints(pcmk__output_t *out, cib_t *cib, const char *ticket_id) { int rc = pcmk_rc_ok; xmlNode *result = NULL; const char *xpath_base = NULL; char *xpath = NULL; pcmk__assert((out != NULL) && (cib != NULL)); xpath_base = pcmk_cib_xpath_for(PCMK_XE_CONSTRAINTS); pcmk__assert(xpath_base != NULL); if (ticket_id != NULL) { xpath = crm_strdup_printf("%s/" PCMK_XE_RSC_TICKET "[@" PCMK_XA_TICKET "=\"%s\"]", xpath_base, ticket_id); } else { xpath = crm_strdup_printf("%s/" PCMK_XE_RSC_TICKET, xpath_base); } rc = cib->cmds->query(cib, xpath, &result, cib_sync_call|cib_xpath); rc = pcmk_legacy2rc(rc); if (result != NULL) { out->message(out, "ticket-constraints", result); pcmk__xml_free(result); } free(xpath); return rc; } int pcmk_ticket_constraints(xmlNodePtr *xml, const char *ticket_id) { pcmk__output_t *out = NULL; int rc = pcmk_rc_ok; cib_t *cib = NULL; rc = pcmk__setup_output_cib_sched(&out, &cib, NULL, xml); if (rc != pcmk_rc_ok) { goto done; } rc = pcmk__ticket_constraints(out, cib, ticket_id); done: if (cib != NULL) { cib__clean_up_connection(&cib); } pcmk__xml_output_finish(out, pcmk_rc2exitc(rc), xml); return rc; } static int delete_single_ticket(xmlNode *child, void *userdata) { int rc = pcmk_rc_ok; cib_t *cib = (cib_t *) userdata; rc = cib->cmds->remove(cib, PCMK_XE_STATUS, child, cib_sync_call); rc = pcmk_legacy2rc(rc); return rc; } int pcmk__ticket_delete(pcmk__output_t *out, cib_t *cib, pcmk_scheduler_t *scheduler, const char *ticket_id, bool force) { int rc = pcmk_rc_ok; xmlNode *state = NULL; pcmk__assert((cib != NULL) && (scheduler != NULL)); if (ticket_id == NULL) { return EINVAL; } if (!force) { pcmk__ticket_t *ticket = NULL; ticket = g_hash_table_lookup(scheduler->priv->ticket_constraints, ticket_id); if (ticket == NULL) { return ENXIO; } if (pcmk_is_set(ticket->flags, pcmk__ticket_granted)) { return EACCES; } } rc = pcmk__get_ticket_state(cib, ticket_id, &state); if (rc == pcmk_rc_duplicate_id) { out->info(out, "Multiple " PCMK__XE_TICKET_STATE "s match ticket=%s", ticket_id); } else if (rc == ENXIO) { return pcmk_rc_ok; } else if (rc != pcmk_rc_ok) { return rc; } crm_log_xml_debug(state, "Delete"); if (rc == pcmk_rc_duplicate_id) { rc = pcmk__xe_foreach_child(state, NULL, delete_single_ticket, cib); } else { rc = delete_single_ticket(state, cib); } if (rc == pcmk_rc_ok) { out->info(out, "Cleaned up %s", ticket_id); } pcmk__xml_free(state); return rc; } int pcmk_ticket_delete(xmlNodePtr *xml, const char *ticket_id, bool force) { pcmk_scheduler_t *scheduler = NULL; pcmk__output_t *out = NULL; cib_t *cib = NULL; int rc = pcmk_rc_ok; rc = pcmk__setup_output_cib_sched(&out, &cib, &scheduler, xml); if (rc != pcmk_rc_ok) { goto done; } rc = pcmk__ticket_delete(out, cib, scheduler, ticket_id, force); done: if (cib != NULL) { cib__clean_up_connection(&cib); } pcmk__xml_output_finish(out, pcmk_rc2exitc(rc), xml); - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); return rc; } int pcmk__ticket_get_attr(pcmk__output_t *out, pcmk_scheduler_t *scheduler, const char *ticket_id, const char *attr_name, const char *attr_default) { int rc = pcmk_rc_ok; const char *attr_value = NULL; pcmk__ticket_t *ticket = NULL; pcmk__assert((out != NULL) && (scheduler != NULL)); if (ticket_id == NULL || attr_name == NULL) { return EINVAL; } ticket = g_hash_table_lookup(scheduler->priv->ticket_constraints, ticket_id); if (ticket != NULL) { attr_value = g_hash_table_lookup(ticket->state, attr_name); } if (attr_value != NULL) { out->message(out, "ticket-attribute", ticket_id, attr_name, attr_value); } else if (attr_default != NULL) { out->message(out, "ticket-attribute", ticket_id, attr_name, attr_default); } else { rc = ENXIO; } return rc; } int pcmk_ticket_get_attr(xmlNodePtr *xml, const char *ticket_id, const char *attr_name, const char *attr_default) { pcmk_scheduler_t *scheduler = NULL; pcmk__output_t *out = NULL; int rc = pcmk_rc_ok; rc = pcmk__setup_output_cib_sched(&out, NULL, &scheduler, xml); if (rc != pcmk_rc_ok) { goto done; } rc = pcmk__ticket_get_attr(out, scheduler, ticket_id, attr_name, attr_default); done: pcmk__xml_output_finish(out, pcmk_rc2exitc(rc), xml); - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); return rc; } int pcmk__ticket_info(pcmk__output_t *out, pcmk_scheduler_t *scheduler, const char *ticket_id, bool details, bool raw) { int rc = pcmk_rc_ok; pcmk__assert((out != NULL) && (scheduler != NULL)); if (ticket_id != NULL) { GHashTable *tickets = NULL; pcmk__ticket_t *ticket = NULL; ticket = g_hash_table_lookup(scheduler->priv->ticket_constraints, ticket_id); if (ticket == NULL) { return ENXIO; } /* The ticket-list message expects a GHashTable, so we'll construct * one with just this single item. */ tickets = pcmk__strkey_table(free, NULL); g_hash_table_insert(tickets, strdup(ticket->id), ticket); out->message(out, "ticket-list", tickets, false, raw, details); g_hash_table_destroy(tickets); } else { out->message(out, "ticket-list", scheduler->priv->ticket_constraints, false, raw, details); } return rc; } int pcmk_ticket_info(xmlNodePtr *xml, const char *ticket_id) { pcmk_scheduler_t *scheduler = NULL; pcmk__output_t *out = NULL; int rc = pcmk_rc_ok; rc = pcmk__setup_output_cib_sched(&out, NULL, &scheduler, xml); if (rc != pcmk_rc_ok) { goto done; } pe__register_messages(out); /* XML output (which is the only format supported by public API functions * due to the use of pcmk__xml_output_new above) always prints all details, * so just pass false for the last two arguments. */ rc = pcmk__ticket_info(out, scheduler, ticket_id, false, false); done: pcmk__xml_output_finish(out, pcmk_rc2exitc(rc), xml); - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); return rc; } int pcmk__ticket_remove_attr(pcmk__output_t *out, cib_t *cib, pcmk_scheduler_t *scheduler, const char *ticket_id, GList *attr_delete, bool force) { xmlNode *ticket_state_xml = NULL; xmlNode *xml_top = NULL; int rc = pcmk_rc_ok; pcmk__assert((out != NULL) && (cib != NULL) && (scheduler != NULL)); if (ticket_id == NULL) { return EINVAL; } /* Nothing to do */ if (attr_delete == NULL) { return pcmk_rc_ok; } rc = build_ticket_modify_xml(cib, ticket_id, &ticket_state_xml, &xml_top); if (rc == pcmk_rc_duplicate_id) { out->info(out, "Multiple " PCMK__XE_TICKET_STATE "s match ticket=%s", ticket_id); } else if (rc != pcmk_rc_ok) { pcmk__xml_free(ticket_state_xml); return rc; } for (GList *list_iter = attr_delete; list_iter != NULL; list_iter = list_iter->next) { const char *key = list_iter->data; if (!force && pcmk__str_eq(key, PCMK__XA_GRANTED, pcmk__str_none)) { pcmk__xml_free(ticket_state_xml); return EACCES; } pcmk__xe_remove_attr(ticket_state_xml, key); } crm_log_xml_debug(xml_top, "Replace"); rc = cib->cmds->replace(cib, PCMK_XE_STATUS, ticket_state_xml, cib_sync_call); rc = pcmk_legacy2rc(rc); pcmk__xml_free(xml_top); return rc; } int pcmk_ticket_remove_attr(xmlNodePtr *xml, const char *ticket_id, GList *attr_delete, bool force) { pcmk_scheduler_t *scheduler = NULL; pcmk__output_t *out = NULL; int rc = pcmk_rc_ok; cib_t *cib = NULL; rc = pcmk__setup_output_cib_sched(&out, &cib, &scheduler, xml); if (rc != pcmk_rc_ok) { goto done; } rc = pcmk__ticket_remove_attr(out, cib, scheduler, ticket_id, attr_delete, force); done: if (cib != NULL) { cib__clean_up_connection(&cib); } pcmk__xml_output_finish(out, pcmk_rc2exitc(rc), xml); - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); return rc; } int pcmk__ticket_set_attr(pcmk__output_t *out, cib_t *cib, pcmk_scheduler_t *scheduler, const char *ticket_id, GHashTable *attr_set, bool force) { xmlNode *ticket_state_xml = NULL; xmlNode *xml_top = NULL; int rc = pcmk_rc_ok; pcmk__assert((out != NULL) && (cib != NULL) && (scheduler != NULL)); if (ticket_id == NULL) { return EINVAL; } /* Nothing to do */ if (attr_set == NULL || g_hash_table_size(attr_set) == 0) { return pcmk_rc_ok; } rc = build_ticket_modify_xml(cib, ticket_id, &ticket_state_xml, &xml_top); if (rc == pcmk_rc_duplicate_id) { out->info(out, "Multiple " PCMK__XE_TICKET_STATE "s match ticket=%s", ticket_id); } else if (rc != pcmk_rc_ok) { pcmk__xml_free(ticket_state_xml); return rc; } if (!force && g_hash_table_lookup(attr_set, PCMK__XA_GRANTED)) { pcmk__xml_free(ticket_state_xml); return EACCES; } add_attribute_xml(scheduler, ticket_id, attr_set, &ticket_state_xml); crm_log_xml_debug(xml_top, "Update"); rc = cib->cmds->modify(cib, PCMK_XE_STATUS, xml_top, cib_sync_call); rc = pcmk_legacy2rc(rc); pcmk__xml_free(xml_top); return rc; } int pcmk_ticket_set_attr(xmlNodePtr *xml, const char *ticket_id, GHashTable *attr_set, bool force) { pcmk_scheduler_t *scheduler = NULL; pcmk__output_t *out = NULL; int rc = pcmk_rc_ok; cib_t *cib = NULL; rc = pcmk__setup_output_cib_sched(&out, &cib, &scheduler, xml); if (rc != pcmk_rc_ok) { goto done; } rc = pcmk__ticket_set_attr(out, cib, scheduler, ticket_id, attr_set, force); done: if (cib != NULL) { cib__clean_up_connection(&cib); } pcmk__xml_output_finish(out, pcmk_rc2exitc(rc), xml); - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); return rc; } int pcmk__ticket_state(pcmk__output_t *out, cib_t *cib, const char *ticket_id) { xmlNode *state_xml = NULL; int rc = pcmk_rc_ok; pcmk__assert((out != NULL) && (cib != NULL)); rc = pcmk__get_ticket_state(cib, ticket_id, &state_xml); if (rc == pcmk_rc_duplicate_id) { out->info(out, "Multiple " PCMK__XE_TICKET_STATE "s match ticket=%s", ticket_id); } if (state_xml != NULL) { out->message(out, "ticket-state", state_xml); pcmk__xml_free(state_xml); } return rc; } int pcmk_ticket_state(xmlNodePtr *xml, const char *ticket_id) { pcmk__output_t *out = NULL; int rc = pcmk_rc_ok; cib_t *cib = NULL; rc = pcmk__setup_output_cib_sched(&out, &cib, NULL, xml); if (rc != pcmk_rc_ok) { goto done; } rc = pcmk__ticket_state(out, cib, ticket_id); done: if (cib != NULL) { cib__clean_up_connection(&cib); } pcmk__xml_output_finish(out, pcmk_rc2exitc(rc), xml); return rc; } diff --git a/lib/pacemaker/pcmk_verify.c b/lib/pacemaker/pcmk_verify.c index 893299c9ed..77dcc5a177 100644 --- a/lib/pacemaker/pcmk_verify.c +++ b/lib/pacemaker/pcmk_verify.c @@ -1,162 +1,162 @@ /* * Copyright 2023-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include #include #include "libpacemaker_private.h" int pcmk__parse_cib(pcmk__output_t *out, const char *cib_source, xmlNodePtr *cib_object) { // @COMPAT Take an enum for cib_source instead of trying to figure it out? const char *first = cib_source; if (cib_source == NULL) { return cib__signon_query(out, NULL, cib_object); } while (isspace(*first)) { first++; } if (*first == '<') { *cib_object = pcmk__xml_parse(cib_source); } else { *cib_object = pcmk__xml_read(cib_source); } return (*cib_object == NULL)? pcmk_rc_unpack_error : pcmk_rc_ok; } int pcmk__verify(pcmk_scheduler_t *scheduler, pcmk__output_t *out, xmlNode **cib_object) { int rc = pcmk_rc_ok; xmlNode *status = NULL; xmlNode *cib_object_copy = NULL; pcmk__assert(cib_object != NULL); /* Without the CIB element, we can't get a schema to validate against, so * report that separately from validation */ if (!pcmk__xe_is(*cib_object, PCMK_XE_CIB)) { out->err(out, "Input is not a CIB (outermost element is %s not " PCMK_XE_CIB ")", pcmk__s((const char *) (*cib_object)->name, "unrecognizable")); rc = pcmk_rc_schema_validation; goto verify_done; } status = pcmk_find_cib_element(*cib_object, PCMK_XE_STATUS); if (status == NULL) { pcmk__xe_create(*cib_object, PCMK_XE_STATUS); } if (!pcmk__validate_xml(*cib_object, NULL, (xmlRelaxNGValidityErrorFunc) out->err, out)) { pcmk__config_has_error = true; rc = pcmk_rc_schema_validation; goto verify_done; } rc = pcmk__update_configured_schema(cib_object, false); if (rc != pcmk_rc_ok) { pcmk__config_has_error = true; out->err(out, "The cluster will NOT be able to use this configuration.\n" "Please manually update the configuration to conform to the %s syntax.", pcmk__highest_schema_name()); goto verify_done; } /* Process the configuration to set pcmk__config_has_error and * pcmk__config_has_warning. * * @TODO Some parts of the configuration are unpacked only when needed (for * example, action configuration), so we aren't necessarily checking those. */ if (*cib_object != NULL) { unsigned long long flags = pcmk__sched_no_counts; if (status == NULL) { // No status available, so do minimal checks flags |= pcmk__sched_validate_only; } cib_object_copy = pcmk__xml_copy(NULL, *cib_object); /* The scheduler takes ownership of the XML object and potentially * frees it later. We want the caller of pcmk__verify to retain * ownership of the passed-in XML object, hence we pass in a copy * to the scheduler. */ pcmk__schedule_actions(cib_object_copy, flags, scheduler); } verify_done: if (pcmk__config_has_error) { rc = pcmk_rc_schema_validation; pcmk__config_err("CIB did not pass schema validation"); } else if (pcmk__config_has_warning) { rc = pcmk_rc_schema_validation; } return rc; } int pcmk_verify(xmlNodePtr *xml, const char *cib_source) { pcmk_scheduler_t *scheduler = NULL; pcmk__output_t *out = NULL; int rc = pcmk_rc_ok; xmlNode *cib_object = NULL; rc = pcmk__xml_output_new(&out, xml); if (rc != pcmk_rc_ok) { return rc; } pe__register_messages(out); pcmk__register_lib_messages(out); rc = pcmk__parse_cib(out, cib_source, &cib_object); if (rc != pcmk_rc_ok) { out->err(out, "Verification failed: %s", pcmk_rc_str(rc)); goto done; } - scheduler = pe_new_working_set(); + scheduler = pcmk_new_scheduler(); if (scheduler == NULL) { rc = errno; out->err(out, "Couldn't allocate scheduler data: %s", pcmk_rc_str(rc)); goto done; } scheduler->priv->out = out; rc = pcmk__verify(scheduler, out, &cib_object); done: - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); pcmk__xml_output_finish(out, pcmk_rc2exitc(rc), xml); pcmk__xml_free(cib_object); return rc; } diff --git a/lib/pengine/bundle.c b/lib/pengine/bundle.c index f19fbd11c0..8fe6ed9d47 100644 --- a/lib/pengine/bundle.c +++ b/lib/pengine/bundle.c @@ -1,2091 +1,2091 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include enum pe__bundle_mount_flags { pe__bundle_mount_none = 0x00, // mount instance-specific subdirectory rather than source directly pe__bundle_mount_subdir = 0x01 }; typedef struct { char *source; char *target; char *options; uint32_t flags; // bitmask of pe__bundle_mount_flags } pe__bundle_mount_t; typedef struct { char *source; char *target; } pe__bundle_port_t; enum pe__container_agent { PE__CONTAINER_AGENT_UNKNOWN, PE__CONTAINER_AGENT_DOCKER, PE__CONTAINER_AGENT_PODMAN, }; #define PE__CONTAINER_AGENT_UNKNOWN_S "unknown" #define PE__CONTAINER_AGENT_DOCKER_S "docker" #define PE__CONTAINER_AGENT_PODMAN_S "podman" typedef struct pe__bundle_variant_data_s { int promoted_max; int nreplicas; int nreplicas_per_host; char *prefix; char *image; const char *ip_last; char *host_network; char *host_netmask; char *control_port; char *container_network; char *ip_range_start; gboolean add_host; gchar *container_host_options; char *container_command; char *launcher_options; const char *attribute_target; pcmk_resource_t *child; GList *replicas; // pcmk__bundle_replica_t * GList *ports; // pe__bundle_port_t * GList *mounts; // pe__bundle_mount_t * enum pe__container_agent agent_type; } pe__bundle_variant_data_t; #define get_bundle_variant_data(data, rsc) do { \ pcmk__assert(pcmk__is_bundle(rsc)); \ data = rsc->priv->variant_opaque; \ } while (0) /*! * \internal * \brief Get maximum number of bundle replicas allowed to run * * \param[in] rsc Bundle or bundled resource to check * * \return Maximum replicas for bundle corresponding to \p rsc */ int pe__bundle_max(const pcmk_resource_t *rsc) { const pe__bundle_variant_data_t *bundle_data = NULL; get_bundle_variant_data(bundle_data, pe__const_top_resource(rsc, true)); return bundle_data->nreplicas; } /*! * \internal * \brief Get the resource inside a bundle * * \param[in] bundle Bundle to check * * \return Resource inside \p bundle if any, otherwise NULL */ pcmk_resource_t * pe__bundled_resource(const pcmk_resource_t *rsc) { const pe__bundle_variant_data_t *bundle_data = NULL; get_bundle_variant_data(bundle_data, pe__const_top_resource(rsc, true)); return bundle_data->child; } /*! * \internal * \brief Get containerized resource corresponding to a given bundle container * * \param[in] instance Collective instance that might be a bundle container * * \return Bundled resource instance inside \p instance if it is a bundle * container instance, otherwise NULL */ const pcmk_resource_t * pe__get_rsc_in_container(const pcmk_resource_t *instance) { const pe__bundle_variant_data_t *data = NULL; const pcmk_resource_t *top = pe__const_top_resource(instance, true); if (!pcmk__is_bundle(top)) { return NULL; } get_bundle_variant_data(data, top); for (const GList *iter = data->replicas; iter != NULL; iter = iter->next) { const pcmk__bundle_replica_t *replica = iter->data; if (instance == replica->container) { return replica->child; } } return NULL; } /*! * \internal * \brief Check whether a given node is created by a bundle * * \param[in] bundle Bundle resource to check * \param[in] node Node to check * * \return true if \p node is an instance of \p bundle, otherwise false */ bool pe__node_is_bundle_instance(const pcmk_resource_t *bundle, const pcmk_node_t *node) { pe__bundle_variant_data_t *bundle_data = NULL; get_bundle_variant_data(bundle_data, bundle); for (GList *iter = bundle_data->replicas; iter != NULL; iter = iter->next) { pcmk__bundle_replica_t *replica = iter->data; if (pcmk__same_node(node, replica->node)) { return true; } } return false; } /*! * \internal * \brief Get the container of a bundle's first replica * * \param[in] bundle Bundle resource to get container for * * \return Container resource from first replica of \p bundle if any, * otherwise NULL */ pcmk_resource_t * pe__first_container(const pcmk_resource_t *bundle) { const pe__bundle_variant_data_t *bundle_data = NULL; const pcmk__bundle_replica_t *replica = NULL; get_bundle_variant_data(bundle_data, bundle); if (bundle_data->replicas == NULL) { return NULL; } replica = bundle_data->replicas->data; return replica->container; } /*! * \internal * \brief Iterate over bundle replicas * * \param[in,out] bundle Bundle to iterate over * \param[in] fn Function to call for each replica (its return value * indicates whether to continue iterating) * \param[in,out] user_data Pointer to pass to \p fn */ void pe__foreach_bundle_replica(pcmk_resource_t *bundle, bool (*fn)(pcmk__bundle_replica_t *, void *), void *user_data) { const pe__bundle_variant_data_t *bundle_data = NULL; get_bundle_variant_data(bundle_data, bundle); for (GList *iter = bundle_data->replicas; iter != NULL; iter = iter->next) { if (!fn((pcmk__bundle_replica_t *) iter->data, user_data)) { break; } } } /*! * \internal * \brief Iterate over const bundle replicas * * \param[in] bundle Bundle to iterate over * \param[in] fn Function to call for each replica (its return value * indicates whether to continue iterating) * \param[in,out] user_data Pointer to pass to \p fn */ void pe__foreach_const_bundle_replica(const pcmk_resource_t *bundle, bool (*fn)(const pcmk__bundle_replica_t *, void *), void *user_data) { const pe__bundle_variant_data_t *bundle_data = NULL; get_bundle_variant_data(bundle_data, bundle); for (const GList *iter = bundle_data->replicas; iter != NULL; iter = iter->next) { if (!fn((const pcmk__bundle_replica_t *) iter->data, user_data)) { break; } } } static char * next_ip(const char *last_ip) { unsigned int oct1 = 0; unsigned int oct2 = 0; unsigned int oct3 = 0; unsigned int oct4 = 0; int rc = sscanf(last_ip, "%u.%u.%u.%u", &oct1, &oct2, &oct3, &oct4); if (rc != 4) { /*@ TODO check for IPv6 */ return NULL; } else if (oct3 > 253) { return NULL; } else if (oct4 > 253) { ++oct3; oct4 = 1; } else { ++oct4; } return crm_strdup_printf("%u.%u.%u.%u", oct1, oct2, oct3, oct4); } static void allocate_ip(pe__bundle_variant_data_t *data, pcmk__bundle_replica_t *replica, GString *buffer) { if(data->ip_range_start == NULL) { return; } else if(data->ip_last) { replica->ipaddr = next_ip(data->ip_last); } else { replica->ipaddr = strdup(data->ip_range_start); } data->ip_last = replica->ipaddr; switch (data->agent_type) { case PE__CONTAINER_AGENT_DOCKER: case PE__CONTAINER_AGENT_PODMAN: if (data->add_host) { g_string_append_printf(buffer, " --add-host=%s-%d:%s", data->prefix, replica->offset, replica->ipaddr); } else { g_string_append_printf(buffer, " --hosts-entry=%s=%s-%d", replica->ipaddr, data->prefix, replica->offset); } break; default: // PE__CONTAINER_AGENT_UNKNOWN break; } } static xmlNode * create_resource(const char *name, const char *provider, const char *kind) { xmlNode *rsc = pcmk__xe_create(NULL, PCMK_XE_PRIMITIVE); crm_xml_add(rsc, PCMK_XA_ID, name); crm_xml_add(rsc, PCMK_XA_CLASS, PCMK_RESOURCE_CLASS_OCF); crm_xml_add(rsc, PCMK_XA_PROVIDER, provider); crm_xml_add(rsc, PCMK_XA_TYPE, kind); return rsc; } /*! * \internal * \brief Check whether cluster can manage resource inside container * * \param[in,out] data Container variant data * * \return TRUE if networking configuration is acceptable, FALSE otherwise * * \note The resource is manageable if an IP range or control port has been * specified. If a control port is used without an IP range, replicas per * host must be 1. */ static bool valid_network(pe__bundle_variant_data_t *data) { if(data->ip_range_start) { return TRUE; } if(data->control_port) { if(data->nreplicas_per_host > 1) { pcmk__config_err("Specifying the '" PCMK_XA_CONTROL_PORT "' for %s " "requires '" PCMK_XA_REPLICAS_PER_HOST "=1'", data->prefix); data->nreplicas_per_host = 1; // @TODO to be sure: // pcmk__clear_rsc_flags(rsc, pcmk__rsc_unique); } return TRUE; } return FALSE; } static int create_ip_resource(pcmk_resource_t *parent, pe__bundle_variant_data_t *data, pcmk__bundle_replica_t *replica) { if(data->ip_range_start) { char *id = NULL; xmlNode *xml_ip = NULL; xmlNode *xml_obj = NULL; id = crm_strdup_printf("%s-ip-%s", data->prefix, replica->ipaddr); pcmk__xml_sanitize_id(id); xml_ip = create_resource(id, "heartbeat", "IPaddr2"); free(id); xml_obj = pcmk__xe_create(xml_ip, PCMK_XE_INSTANCE_ATTRIBUTES); pcmk__xe_set_id(xml_obj, "%s-attributes-%d", data->prefix, replica->offset); crm_create_nvpair_xml(xml_obj, NULL, "ip", replica->ipaddr); if(data->host_network) { crm_create_nvpair_xml(xml_obj, NULL, "nic", data->host_network); } if(data->host_netmask) { crm_create_nvpair_xml(xml_obj, NULL, "cidr_netmask", data->host_netmask); } else { crm_create_nvpair_xml(xml_obj, NULL, "cidr_netmask", "32"); } xml_obj = pcmk__xe_create(xml_ip, PCMK_XE_OPERATIONS); crm_create_op_xml(xml_obj, pcmk__xe_id(xml_ip), PCMK_ACTION_MONITOR, "60s", NULL); // TODO: Other ops? Timeouts and intervals from underlying resource? if (pe__unpack_resource(xml_ip, &replica->ip, parent, parent->priv->scheduler) != pcmk_rc_ok) { return pcmk_rc_unpack_error; } parent->priv->children = g_list_append(parent->priv->children, replica->ip); } return pcmk_rc_ok; } static const char* container_agent_str(enum pe__container_agent t) { switch (t) { case PE__CONTAINER_AGENT_DOCKER: return PE__CONTAINER_AGENT_DOCKER_S; case PE__CONTAINER_AGENT_PODMAN: return PE__CONTAINER_AGENT_PODMAN_S; default: // PE__CONTAINER_AGENT_UNKNOWN break; } return PE__CONTAINER_AGENT_UNKNOWN_S; } static int create_container_resource(pcmk_resource_t *parent, const pe__bundle_variant_data_t *data, pcmk__bundle_replica_t *replica) { char *id = NULL; xmlNode *xml_container = NULL; xmlNode *xml_obj = NULL; // Agent-specific const char *hostname_opt = NULL; const char *env_opt = NULL; const char *agent_str = NULL; GString *buffer = NULL; GString *dbuffer = NULL; // Where syntax differences are drop-in replacements, set them now switch (data->agent_type) { case PE__CONTAINER_AGENT_DOCKER: case PE__CONTAINER_AGENT_PODMAN: hostname_opt = "-h "; env_opt = "-e "; break; default: // PE__CONTAINER_AGENT_UNKNOWN return pcmk_rc_unpack_error; } agent_str = container_agent_str(data->agent_type); buffer = g_string_sized_new(4096); id = crm_strdup_printf("%s-%s-%d", data->prefix, agent_str, replica->offset); pcmk__xml_sanitize_id(id); xml_container = create_resource(id, "heartbeat", agent_str); free(id); xml_obj = pcmk__xe_create(xml_container, PCMK_XE_INSTANCE_ATTRIBUTES); pcmk__xe_set_id(xml_obj, "%s-attributes-%d", data->prefix, replica->offset); crm_create_nvpair_xml(xml_obj, NULL, "image", data->image); crm_create_nvpair_xml(xml_obj, NULL, "allow_pull", PCMK_VALUE_TRUE); crm_create_nvpair_xml(xml_obj, NULL, "force_kill", PCMK_VALUE_FALSE); crm_create_nvpair_xml(xml_obj, NULL, "reuse", PCMK_VALUE_FALSE); if (data->agent_type == PE__CONTAINER_AGENT_DOCKER) { g_string_append(buffer, " --restart=no"); } /* Set a container hostname only if we have an IP to map it to. The user can * set -h or --uts=host themselves if they want a nicer name for logs, but * this makes applications happy who need their hostname to match the IP * they bind to. */ if (data->ip_range_start != NULL) { g_string_append_printf(buffer, " %s%s-%d", hostname_opt, data->prefix, replica->offset); } pcmk__g_strcat(buffer, " ", env_opt, "PCMK_stderr=1", NULL); if (data->container_network != NULL) { pcmk__g_strcat(buffer, " --net=", data->container_network, NULL); } if (data->control_port != NULL) { pcmk__g_strcat(buffer, " ", env_opt, "PCMK_" PCMK__ENV_REMOTE_PORT "=", data->control_port, NULL); } else { g_string_append_printf(buffer, " %sPCMK_" PCMK__ENV_REMOTE_PORT "=%d", env_opt, DEFAULT_REMOTE_PORT); } for (GList *iter = data->mounts; iter != NULL; iter = iter->next) { pe__bundle_mount_t *mount = (pe__bundle_mount_t *) iter->data; char *source = NULL; if (pcmk_is_set(mount->flags, pe__bundle_mount_subdir)) { source = crm_strdup_printf("%s/%s-%d", mount->source, data->prefix, replica->offset); pcmk__add_separated_word(&dbuffer, 1024, source, ","); } switch (data->agent_type) { case PE__CONTAINER_AGENT_DOCKER: case PE__CONTAINER_AGENT_PODMAN: pcmk__g_strcat(buffer, " -v ", pcmk__s(source, mount->source), ":", mount->target, NULL); if (mount->options != NULL) { pcmk__g_strcat(buffer, ":", mount->options, NULL); } break; default: break; } free(source); } for (GList *iter = data->ports; iter != NULL; iter = iter->next) { pe__bundle_port_t *port = (pe__bundle_port_t *) iter->data; switch (data->agent_type) { case PE__CONTAINER_AGENT_DOCKER: case PE__CONTAINER_AGENT_PODMAN: if (replica->ipaddr != NULL) { pcmk__g_strcat(buffer, " -p ", replica->ipaddr, ":", port->source, ":", port->target, NULL); } else if (!pcmk__str_eq(data->container_network, PCMK_VALUE_HOST, pcmk__str_none)) { // No need to do port mapping if net == host pcmk__g_strcat(buffer, " -p ", port->source, ":", port->target, NULL); } break; default: break; } } /* @COMPAT: We should use pcmk__add_word() here, but we can't yet, because * it would cause restarts during rolling upgrades. * * In a previous version of the container resource creation logic, if * data->launcher_options is not NULL, we append * (" %s", data->launcher_options) even if data->launcher_options is an * empty string. Likewise for data->container_host_options. Using * * pcmk__add_word(buffer, 0, data->launcher_options) * * removes that extra trailing space, causing a resource definition change. */ if (data->launcher_options != NULL) { pcmk__g_strcat(buffer, " ", data->launcher_options, NULL); } if (data->container_host_options != NULL) { pcmk__g_strcat(buffer, " ", data->container_host_options, NULL); } crm_create_nvpair_xml(xml_obj, NULL, "run_opts", (const char *) buffer->str); g_string_free(buffer, TRUE); crm_create_nvpair_xml(xml_obj, NULL, "mount_points", (dbuffer != NULL)? (const char *) dbuffer->str : ""); if (dbuffer != NULL) { g_string_free(dbuffer, TRUE); } if (replica->child != NULL) { if (data->container_command != NULL) { crm_create_nvpair_xml(xml_obj, NULL, "run_cmd", data->container_command); } else { crm_create_nvpair_xml(xml_obj, NULL, "run_cmd", SBIN_DIR "/" PCMK__SERVER_REMOTED); } /* TODO: Allow users to specify their own? * * We just want to know if the container is alive; we'll monitor the * child independently. */ crm_create_nvpair_xml(xml_obj, NULL, "monitor_cmd", "/bin/true"); #if 0 /* @TODO Consider supporting the use case where we can start and stop * resources, but not proxy local commands (such as setting node * attributes), by running the local executor in stand-alone mode. * However, this would probably be better done via ACLs as with other * Pacemaker Remote nodes. */ } else if ((child != NULL) && data->untrusted) { crm_create_nvpair_xml(xml_obj, NULL, "run_cmd", CRM_DAEMON_DIR "/" PCMK__SERVER_EXECD); crm_create_nvpair_xml(xml_obj, NULL, "monitor_cmd", CRM_DAEMON_DIR "/pacemaker/cts-exec-helper -c poke"); #endif } else { if (data->container_command != NULL) { crm_create_nvpair_xml(xml_obj, NULL, "run_cmd", data->container_command); } /* TODO: Allow users to specify their own? * * We don't know what's in the container, so we just want to know if it * is alive. */ crm_create_nvpair_xml(xml_obj, NULL, "monitor_cmd", "/bin/true"); } xml_obj = pcmk__xe_create(xml_container, PCMK_XE_OPERATIONS); crm_create_op_xml(xml_obj, pcmk__xe_id(xml_container), PCMK_ACTION_MONITOR, "60s", NULL); // TODO: Other ops? Timeouts and intervals from underlying resource? if (pe__unpack_resource(xml_container, &replica->container, parent, parent->priv->scheduler) != pcmk_rc_ok) { return pcmk_rc_unpack_error; } pcmk__set_rsc_flags(replica->container, pcmk__rsc_replica_container); parent->priv->children = g_list_append(parent->priv->children, replica->container); return pcmk_rc_ok; } /*! * \brief Ban a node from a resource's (and its children's) allowed nodes list * * \param[in,out] rsc Resource to modify * \param[in] uname Name of node to ban */ static void disallow_node(pcmk_resource_t *rsc, const char *uname) { gpointer match = g_hash_table_lookup(rsc->priv->allowed_nodes, uname); if (match) { ((pcmk_node_t *) match)->assign->score = -PCMK_SCORE_INFINITY; ((pcmk_node_t *) match)->assign->probe_mode = pcmk__probe_never; } g_list_foreach(rsc->priv->children, (GFunc) disallow_node, (gpointer) uname); } static int create_remote_resource(pcmk_resource_t *parent, pe__bundle_variant_data_t *data, pcmk__bundle_replica_t *replica) { if (replica->child && valid_network(data)) { GHashTableIter gIter; pcmk_node_t *node = NULL; xmlNode *xml_remote = NULL; char *id = crm_strdup_printf("%s-%d", data->prefix, replica->offset); char *port_s = NULL; const char *uname = NULL; const char *connect_name = NULL; pcmk_scheduler_t *scheduler = parent->priv->scheduler; if (pe_find_resource(scheduler->priv->resources, id) != NULL) { free(id); // The biggest hammer we have id = crm_strdup_printf("pcmk-internal-%s-remote-%d", replica->child->id, replica->offset); //@TODO return error instead of asserting? pcmk__assert(pe_find_resource(scheduler->priv->resources, id) == NULL); } /* REMOTE_CONTAINER_HACK: Using "#uname" as the server name when the * connection does not have its own IP is a magic string that we use to * support nested remotes (i.e. a bundle running on a remote node). */ connect_name = (replica->ipaddr? replica->ipaddr : "#uname"); if (data->control_port == NULL) { port_s = pcmk__itoa(DEFAULT_REMOTE_PORT); } /* This sets replica->container as replica->remote's container, which is * similar to what happens with guest nodes. This is how the scheduler * knows that the bundle node is fenced by recovering the container, and * that remote should be ordered relative to the container. */ xml_remote = pe_create_remote_xml(NULL, id, replica->container->id, NULL, NULL, NULL, connect_name, (data->control_port? data->control_port : port_s)); free(port_s); /* Abandon our created ID, and pull the copy from the XML, because we * need something that will get freed during scheduler data cleanup to * use as the node ID and uname. */ free(id); id = NULL; uname = pcmk__xe_id(xml_remote); /* Ensure a node has been created for the guest (it may have already * been, if it has a permanent node attribute), and ensure its weight is * -INFINITY so no other resources can run on it. */ node = pcmk_find_node(scheduler, uname); if (node == NULL) { node = pe_create_node(uname, uname, PCMK_VALUE_REMOTE, -PCMK_SCORE_INFINITY, scheduler); } else { node->assign->score = -PCMK_SCORE_INFINITY; } node->assign->probe_mode = pcmk__probe_never; /* unpack_remote_nodes() ensures that each remote node and guest node * has a pcmk_node_t entry. Ideally, it would do the same for bundle * nodes. Unfortunately, a bundle has to be mostly unpacked before it's * obvious what nodes will be needed, so we do it just above. * * Worse, that means that the node may have been utilized while * unpacking other resources, without our weight correction. The most * likely place for this to happen is when pe__unpack_resource() calls * resource_location() to set a default score in symmetric clusters. * This adds a node *copy* to each resource's allowed nodes, and these * copies will have the wrong weight. * * As a hacky workaround, fix those copies here. * * @TODO Possible alternative: ensure bundles are unpacked before other * resources, so the weight is correct before any copies are made. */ g_list_foreach(scheduler->priv->resources, (GFunc) disallow_node, (gpointer) uname); replica->node = pe__copy_node(node); replica->node->assign->score = 500; replica->node->assign->probe_mode = pcmk__probe_exclusive; /* Ensure the node shows up as allowed and with the correct discovery set */ if (replica->child->priv->allowed_nodes != NULL) { g_hash_table_destroy(replica->child->priv->allowed_nodes); } replica->child->priv->allowed_nodes = pcmk__strkey_table(NULL, pcmk__free_node_copy); g_hash_table_insert(replica->child->priv->allowed_nodes, (gpointer) replica->node->priv->id, pe__copy_node(replica->node)); { const pcmk_resource_t *parent = replica->child->priv->parent; pcmk_node_t *copy = pe__copy_node(replica->node); copy->assign->score = -PCMK_SCORE_INFINITY; g_hash_table_insert(parent->priv->allowed_nodes, (gpointer) replica->node->priv->id, copy); } if (pe__unpack_resource(xml_remote, &replica->remote, parent, scheduler) != pcmk_rc_ok) { return pcmk_rc_unpack_error; } g_hash_table_iter_init(&gIter, replica->remote->priv->allowed_nodes); while (g_hash_table_iter_next(&gIter, NULL, (void **)&node)) { if (pcmk__is_pacemaker_remote_node(node)) { /* Remote resources can only run on 'normal' cluster node */ node->assign->score = -PCMK_SCORE_INFINITY; } } replica->node->priv->remote = replica->remote; // Ensure pcmk__is_guest_or_bundle_node() functions correctly replica->remote->priv->launcher = replica->container; /* A bundle's #kind is closer to "container" (guest node) than the * "remote" set by pe_create_node(). */ pcmk__insert_dup(replica->node->priv->attrs, CRM_ATTR_KIND, "container"); /* One effect of this is that unpack_launcher() will add * replica->remote to replica->container's launched resources, which * will make pe__resource_contains_guest_node() true for * replica->container. * * replica->child does NOT get added to replica->container's launched * resources. The only noticeable effect if it did would be for its * fail count to be taken into account when checking * replica->container's migration threshold. */ parent->priv->children = g_list_append(parent->priv->children, replica->remote); } return pcmk_rc_ok; } static int create_replica_resources(pcmk_resource_t *parent, pe__bundle_variant_data_t *data, pcmk__bundle_replica_t *replica) { int rc = pcmk_rc_ok; rc = create_container_resource(parent, data, replica); if (rc != pcmk_rc_ok) { return rc; } rc = create_ip_resource(parent, data, replica); if (rc != pcmk_rc_ok) { return rc; } rc = create_remote_resource(parent, data, replica); if (rc != pcmk_rc_ok) { return rc; } if ((replica->child != NULL) && (replica->ipaddr != NULL)) { pcmk__insert_meta(replica->child->priv, "external-ip", replica->ipaddr); } if (replica->remote != NULL) { /* * Allow the remote connection resource to be allocated to a * different node than the one on which the container is active. * * This makes it possible to have Pacemaker Remote nodes running * containers with the remote executor inside in order to start * services inside those containers. */ pcmk__set_rsc_flags(replica->remote, pcmk__rsc_remote_nesting_allowed); } return rc; } static void mount_add(pe__bundle_variant_data_t *bundle_data, const char *source, const char *target, const char *options, uint32_t flags) { pe__bundle_mount_t *mount = pcmk__assert_alloc(1, sizeof(pe__bundle_mount_t)); mount->source = pcmk__str_copy(source); mount->target = pcmk__str_copy(target); mount->options = pcmk__str_copy(options); mount->flags = flags; bundle_data->mounts = g_list_append(bundle_data->mounts, mount); } static void mount_free(pe__bundle_mount_t *mount) { free(mount->source); free(mount->target); free(mount->options); free(mount); } static void port_free(pe__bundle_port_t *port) { free(port->source); free(port->target); free(port); } static pcmk__bundle_replica_t * replica_for_remote(pcmk_resource_t *remote) { pcmk_resource_t *top = remote; pe__bundle_variant_data_t *bundle_data = NULL; if (top == NULL) { return NULL; } while (top->priv->parent != NULL) { top = top->priv->parent; } get_bundle_variant_data(bundle_data, top); for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pcmk__bundle_replica_t *replica = gIter->data; if (replica->remote == remote) { return replica; } } CRM_LOG_ASSERT(FALSE); return NULL; } bool pe__bundle_needs_remote_name(pcmk_resource_t *rsc) { const char *value; GHashTable *params = NULL; if (rsc == NULL) { return false; } // Use NULL node since pcmk__bundle_expand() uses that to set value params = pe_rsc_params(rsc, NULL, rsc->priv->scheduler); value = g_hash_table_lookup(params, PCMK_REMOTE_RA_ADDR); return pcmk__str_eq(value, "#uname", pcmk__str_casei) && xml_contains_remote_node(rsc->priv->xml); } const char * pe__add_bundle_remote_name(pcmk_resource_t *rsc, xmlNode *xml, const char *field) { // REMOTE_CONTAINER_HACK: Allow remote nodes that start containers with pacemaker remote inside pcmk_node_t *node = NULL; pcmk__bundle_replica_t *replica = NULL; if (!pe__bundle_needs_remote_name(rsc)) { return NULL; } replica = replica_for_remote(rsc); if (replica == NULL) { return NULL; } node = replica->container->priv->assigned_node; if (node == NULL) { /* If it won't be running anywhere after the * transition, go with where it's running now. */ node = pcmk__current_node(replica->container); } if(node == NULL) { crm_trace("Cannot determine address for bundle connection %s", rsc->id); return NULL; } crm_trace("Setting address for bundle connection %s to bundle host %s", rsc->id, pcmk__node_name(node)); if(xml != NULL && field != NULL) { crm_xml_add(xml, field, node->priv->name); } return node->priv->name; } #define pe__set_bundle_mount_flags(mount_xml, flags, flags_to_set) do { \ flags = pcmk__set_flags_as(__func__, __LINE__, LOG_TRACE, \ "Bundle mount", pcmk__xe_id(mount_xml), \ flags, (flags_to_set), #flags_to_set); \ } while (0) gboolean pe__unpack_bundle(pcmk_resource_t *rsc, pcmk_scheduler_t *scheduler) { const char *value = NULL; xmlNode *xml_obj = NULL; const xmlNode *xml_child = NULL; xmlNode *xml_resource = NULL; pe__bundle_variant_data_t *bundle_data = NULL; bool need_log_mount = TRUE; pcmk__assert(rsc != NULL); pcmk__rsc_trace(rsc, "Processing resource %s...", rsc->id); bundle_data = pcmk__assert_alloc(1, sizeof(pe__bundle_variant_data_t)); rsc->priv->variant_opaque = bundle_data; bundle_data->prefix = strdup(rsc->id); xml_obj = pcmk__xe_first_child(rsc->priv->xml, PCMK_XE_DOCKER, NULL, NULL); if (xml_obj != NULL) { bundle_data->agent_type = PE__CONTAINER_AGENT_DOCKER; } if (xml_obj == NULL) { xml_obj = pcmk__xe_first_child(rsc->priv->xml, PCMK_XE_PODMAN, NULL, NULL); if (xml_obj != NULL) { bundle_data->agent_type = PE__CONTAINER_AGENT_PODMAN; } } if (xml_obj == NULL) { return FALSE; } // Use 0 for default, minimum, and invalid PCMK_XA_PROMOTED_MAX value = crm_element_value(xml_obj, PCMK_XA_PROMOTED_MAX); pcmk__scan_min_int(value, &bundle_data->promoted_max, 0); /* Default replicas to PCMK_XA_PROMOTED_MAX if it was specified and 1 * otherwise */ value = crm_element_value(xml_obj, PCMK_XA_REPLICAS); if ((value == NULL) && (bundle_data->promoted_max > 0)) { bundle_data->nreplicas = bundle_data->promoted_max; } else { pcmk__scan_min_int(value, &bundle_data->nreplicas, 1); } /* * Communication between containers on the same host via the * floating IPs only works if the container is started with: * --userland-proxy=false --ip-masq=false */ value = crm_element_value(xml_obj, PCMK_XA_REPLICAS_PER_HOST); pcmk__scan_min_int(value, &bundle_data->nreplicas_per_host, 1); if (bundle_data->nreplicas_per_host == 1) { pcmk__clear_rsc_flags(rsc, pcmk__rsc_unique); } bundle_data->container_command = crm_element_value_copy(xml_obj, PCMK_XA_RUN_COMMAND); bundle_data->launcher_options = crm_element_value_copy(xml_obj, PCMK_XA_OPTIONS); bundle_data->image = crm_element_value_copy(xml_obj, PCMK_XA_IMAGE); bundle_data->container_network = crm_element_value_copy(xml_obj, PCMK_XA_NETWORK); xml_obj = pcmk__xe_first_child(rsc->priv->xml, PCMK_XE_NETWORK, NULL, NULL); if(xml_obj) { bundle_data->ip_range_start = crm_element_value_copy(xml_obj, PCMK_XA_IP_RANGE_START); bundle_data->host_netmask = crm_element_value_copy(xml_obj, PCMK_XA_HOST_NETMASK); bundle_data->host_network = crm_element_value_copy(xml_obj, PCMK_XA_HOST_INTERFACE); bundle_data->control_port = crm_element_value_copy(xml_obj, PCMK_XA_CONTROL_PORT); value = crm_element_value(xml_obj, PCMK_XA_ADD_HOST); if (crm_str_to_boolean(value, &bundle_data->add_host) != 1) { bundle_data->add_host = TRUE; } for (xml_child = pcmk__xe_first_child(xml_obj, PCMK_XE_PORT_MAPPING, NULL, NULL); xml_child != NULL; xml_child = pcmk__xe_next(xml_child, PCMK_XE_PORT_MAPPING)) { pe__bundle_port_t *port = pcmk__assert_alloc(1, sizeof(pe__bundle_port_t)); port->source = crm_element_value_copy(xml_child, PCMK_XA_PORT); if(port->source == NULL) { port->source = crm_element_value_copy(xml_child, PCMK_XA_RANGE); } else { port->target = crm_element_value_copy(xml_child, PCMK_XA_INTERNAL_PORT); } if(port->source != NULL && strlen(port->source) > 0) { if(port->target == NULL) { port->target = strdup(port->source); } bundle_data->ports = g_list_append(bundle_data->ports, port); } else { pcmk__config_err("Invalid " PCMK_XA_PORT " directive %s", pcmk__xe_id(xml_child)); port_free(port); } } } xml_obj = pcmk__xe_first_child(rsc->priv->xml, PCMK_XE_STORAGE, NULL, NULL); for (xml_child = pcmk__xe_first_child(xml_obj, PCMK_XE_STORAGE_MAPPING, NULL, NULL); xml_child != NULL; xml_child = pcmk__xe_next(xml_child, PCMK_XE_STORAGE_MAPPING)) { const char *source = crm_element_value(xml_child, PCMK_XA_SOURCE_DIR); const char *target = crm_element_value(xml_child, PCMK_XA_TARGET_DIR); const char *options = crm_element_value(xml_child, PCMK_XA_OPTIONS); int flags = pe__bundle_mount_none; if (source == NULL) { source = crm_element_value(xml_child, PCMK_XA_SOURCE_DIR_ROOT); pe__set_bundle_mount_flags(xml_child, flags, pe__bundle_mount_subdir); } if (source && target) { mount_add(bundle_data, source, target, options, flags); if (strcmp(target, "/var/log") == 0) { need_log_mount = FALSE; } } else { pcmk__config_err("Invalid mount directive %s", pcmk__xe_id(xml_child)); } } xml_obj = pcmk__xe_first_child(rsc->priv->xml, PCMK_XE_PRIMITIVE, NULL, NULL); if (xml_obj && valid_network(bundle_data)) { const char *suffix = NULL; char *value = NULL; xmlNode *xml_set = NULL; xml_resource = pcmk__xe_create(NULL, PCMK_XE_CLONE); /* @COMPAT We no longer use the tag, but we need to keep it as * part of the resource name, so that bundles don't restart in a rolling * upgrade. (It also avoids needing to change regression tests.) */ suffix = (const char *) xml_resource->name; if (bundle_data->promoted_max > 0) { suffix = "master"; } pcmk__xe_set_id(xml_resource, "%s-%s", bundle_data->prefix, suffix); xml_set = pcmk__xe_create(xml_resource, PCMK_XE_META_ATTRIBUTES); pcmk__xe_set_id(xml_set, "%s-%s-meta", bundle_data->prefix, xml_resource->name); crm_create_nvpair_xml(xml_set, NULL, PCMK_META_ORDERED, PCMK_VALUE_TRUE); value = pcmk__itoa(bundle_data->nreplicas); crm_create_nvpair_xml(xml_set, NULL, PCMK_META_CLONE_MAX, value); free(value); value = pcmk__itoa(bundle_data->nreplicas_per_host); crm_create_nvpair_xml(xml_set, NULL, PCMK_META_CLONE_NODE_MAX, value); free(value); crm_create_nvpair_xml(xml_set, NULL, PCMK_META_GLOBALLY_UNIQUE, pcmk__btoa(bundle_data->nreplicas_per_host > 1)); if (bundle_data->promoted_max) { crm_create_nvpair_xml(xml_set, NULL, PCMK_META_PROMOTABLE, PCMK_VALUE_TRUE); value = pcmk__itoa(bundle_data->promoted_max); crm_create_nvpair_xml(xml_set, NULL, PCMK_META_PROMOTED_MAX, value); free(value); } //crm_xml_add(xml_obj, PCMK_XA_ID, bundle_data->prefix); pcmk__xml_copy(xml_resource, xml_obj); } else if(xml_obj) { pcmk__config_err("Cannot control %s inside %s without either " PCMK_XA_IP_RANGE_START " or " PCMK_XA_CONTROL_PORT, rsc->id, pcmk__xe_id(xml_obj)); return FALSE; } if(xml_resource) { int lpc = 0; GList *childIter = NULL; pe__bundle_port_t *port = NULL; GString *buffer = NULL; if (pe__unpack_resource(xml_resource, &(bundle_data->child), rsc, scheduler) != pcmk_rc_ok) { return FALSE; } /* Currently, we always map the default authentication key location * into the same location inside the container. * * Ideally, we would respect the host's PCMK_authkey_location, but: * - it may be different on different nodes; * - the actual connection will do extra checking to make sure the key * file exists and is readable, that we can't do here on the DC * - tools such as crm_resource and crm_simulate may not have the same * environment variables as the cluster, causing operation digests to * differ * * Always using the default location inside the container is fine, * because we control the pacemaker_remote environment, and it avoids * having to pass another environment variable to the container. * * @TODO A better solution may be to have only pacemaker_remote use the * environment variable, and have the cluster nodes use a new * cluster option for key location. This would introduce the limitation * of the location being the same on all cluster nodes, but that's * reasonable. */ mount_add(bundle_data, DEFAULT_REMOTE_KEY_LOCATION, DEFAULT_REMOTE_KEY_LOCATION, NULL, pe__bundle_mount_none); if (need_log_mount) { mount_add(bundle_data, CRM_BUNDLE_DIR, "/var/log", NULL, pe__bundle_mount_subdir); } port = pcmk__assert_alloc(1, sizeof(pe__bundle_port_t)); if(bundle_data->control_port) { port->source = strdup(bundle_data->control_port); } else { /* If we wanted to respect PCMK_remote_port, we could use * crm_default_remote_port() here and elsewhere in this file instead * of DEFAULT_REMOTE_PORT. * * However, it gains nothing, since we control both the container * environment and the connection resource parameters, and the user * can use a different port if desired by setting * PCMK_XA_CONTROL_PORT. */ port->source = pcmk__itoa(DEFAULT_REMOTE_PORT); } port->target = strdup(port->source); bundle_data->ports = g_list_append(bundle_data->ports, port); buffer = g_string_sized_new(1024); for (childIter = bundle_data->child->priv->children; childIter != NULL; childIter = childIter->next) { pcmk__bundle_replica_t *replica = NULL; replica = pcmk__assert_alloc(1, sizeof(pcmk__bundle_replica_t)); replica->child = childIter->data; pcmk__set_rsc_flags(replica->child, pcmk__rsc_exclusive_probes); replica->offset = lpc++; // Ensure the child's notify gets set based on the underlying primitive's value if (pcmk_is_set(replica->child->flags, pcmk__rsc_notify)) { pcmk__set_rsc_flags(bundle_data->child, pcmk__rsc_notify); } allocate_ip(bundle_data, replica, buffer); bundle_data->replicas = g_list_append(bundle_data->replicas, replica); bundle_data->attribute_target = g_hash_table_lookup(replica->child->priv->meta, PCMK_META_CONTAINER_ATTRIBUTE_TARGET); } bundle_data->container_host_options = g_string_free(buffer, FALSE); if (bundle_data->attribute_target) { pcmk__insert_dup(rsc->priv->meta, PCMK_META_CONTAINER_ATTRIBUTE_TARGET, bundle_data->attribute_target); pcmk__insert_dup(bundle_data->child->priv->meta, PCMK_META_CONTAINER_ATTRIBUTE_TARGET, bundle_data->attribute_target); } } else { // Just a naked container, no pacemaker-remote GString *buffer = g_string_sized_new(1024); for (int lpc = 0; lpc < bundle_data->nreplicas; lpc++) { pcmk__bundle_replica_t *replica = NULL; replica = pcmk__assert_alloc(1, sizeof(pcmk__bundle_replica_t)); replica->offset = lpc; allocate_ip(bundle_data, replica, buffer); bundle_data->replicas = g_list_append(bundle_data->replicas, replica); } bundle_data->container_host_options = g_string_free(buffer, FALSE); } for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pcmk__bundle_replica_t *replica = gIter->data; if (create_replica_resources(rsc, bundle_data, replica) != pcmk_rc_ok) { pcmk__config_err("Failed unpacking resource %s", rsc->id); - rsc->priv->fns->free(rsc); + pcmk__free_resource(rsc); return FALSE; } /* Utilization needs special handling for bundles. It makes no sense for * the inner primitive to have utilization, because it is tied * one-to-one to the guest node created by the container resource -- and * there's no way to set capacities for that guest node anyway. * * What the user really wants is to configure utilization for the * container. However, the schema only allows utilization for * primitives, and the container resource is implicit anyway, so the * user can *only* configure utilization for the inner primitive. If * they do, move the primitive's utilization values to the container. * * @TODO This means that bundles without an inner primitive can't have * utilization. An alternative might be to allow utilization values in * the top-level bundle XML in the schema, and copy those to each * container. */ if (replica->child != NULL) { GHashTable *empty = replica->container->priv->utilization; replica->container->priv->utilization = replica->child->priv->utilization; replica->child->priv->utilization = empty; } } if (bundle_data->child) { rsc->priv->children = g_list_append(rsc->priv->children, bundle_data->child); } return TRUE; } static int replica_resource_active(pcmk_resource_t *rsc, gboolean all) { if (rsc) { gboolean child_active = rsc->priv->fns->active(rsc, all); if (child_active && !all) { return TRUE; } else if (!child_active && all) { return FALSE; } } return -1; } gboolean pe__bundle_active(pcmk_resource_t *rsc, gboolean all) { pe__bundle_variant_data_t *bundle_data = NULL; GList *iter = NULL; get_bundle_variant_data(bundle_data, rsc); for (iter = bundle_data->replicas; iter != NULL; iter = iter->next) { pcmk__bundle_replica_t *replica = iter->data; int rsc_active; rsc_active = replica_resource_active(replica->ip, all); if (rsc_active >= 0) { return (gboolean) rsc_active; } rsc_active = replica_resource_active(replica->child, all); if (rsc_active >= 0) { return (gboolean) rsc_active; } rsc_active = replica_resource_active(replica->container, all); if (rsc_active >= 0) { return (gboolean) rsc_active; } rsc_active = replica_resource_active(replica->remote, all); if (rsc_active >= 0) { return (gboolean) rsc_active; } } /* If "all" is TRUE, we've already checked that no resources were inactive, * so return TRUE; if "all" is FALSE, we didn't find any active resources, * so return FALSE. */ return all; } /*! * \internal * \brief Find the bundle replica corresponding to a given node * * \param[in] bundle Top-level bundle resource * \param[in] node Node to search for * * \return Bundle replica if found, NULL otherwise */ pcmk_resource_t * pe__find_bundle_replica(const pcmk_resource_t *bundle, const pcmk_node_t *node) { pe__bundle_variant_data_t *bundle_data = NULL; pcmk__assert((bundle != NULL) && (node != NULL)); get_bundle_variant_data(bundle_data, bundle); for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pcmk__bundle_replica_t *replica = gIter->data; pcmk__assert((replica != NULL) && (replica->node != NULL)); if (pcmk__same_node(replica->node, node)) { return replica->child; } } return NULL; } PCMK__OUTPUT_ARGS("bundle", "uint32_t", "pcmk_resource_t *", "GList *", "GList *") int pe__bundle_xml(pcmk__output_t *out, va_list args) { uint32_t show_opts = va_arg(args, uint32_t); pcmk_resource_t *rsc = va_arg(args, pcmk_resource_t *); GList *only_node = va_arg(args, GList *); GList *only_rsc = va_arg(args, GList *); pe__bundle_variant_data_t *bundle_data = NULL; int rc = pcmk_rc_no_output; gboolean printed_header = FALSE; gboolean print_everything = TRUE; const char *desc = NULL; pcmk__assert(rsc != NULL); get_bundle_variant_data(bundle_data, rsc); if (rsc->priv->fns->is_filtered(rsc, only_rsc, TRUE)) { return rc; } print_everything = pcmk__str_in_list(rsc->id, only_rsc, pcmk__str_star_matches); for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pcmk__bundle_replica_t *replica = gIter->data; pcmk_resource_t *ip = replica->ip; pcmk_resource_t *child = replica->child; pcmk_resource_t *container = replica->container; pcmk_resource_t *remote = replica->remote; char *id = NULL; gboolean print_ip, print_child, print_ctnr, print_remote; pcmk__assert(replica != NULL); if (pcmk__rsc_filtered_by_node(container, only_node)) { continue; } print_ip = (ip != NULL) && !ip->priv->fns->is_filtered(ip, only_rsc, print_everything); print_child = (child != NULL) && !child->priv->fns->is_filtered(child, only_rsc, print_everything); print_ctnr = !container->priv->fns->is_filtered(container, only_rsc, print_everything); print_remote = (remote != NULL) && !remote->priv->fns->is_filtered(remote, only_rsc, print_everything); if (!print_everything && !print_ip && !print_child && !print_ctnr && !print_remote) { continue; } if (!printed_header) { const char *type = container_agent_str(bundle_data->agent_type); const char *unique = pcmk__flag_text(rsc->flags, pcmk__rsc_unique); const char *maintenance = pcmk__flag_text(rsc->flags, pcmk__rsc_maintenance); const char *managed = pcmk__flag_text(rsc->flags, pcmk__rsc_managed); const char *failed = pcmk__flag_text(rsc->flags, pcmk__rsc_failed); printed_header = TRUE; desc = pe__resource_description(rsc, show_opts); rc = pe__name_and_nvpairs_xml(out, true, PCMK_XE_BUNDLE, PCMK_XA_ID, rsc->id, PCMK_XA_TYPE, type, PCMK_XA_IMAGE, bundle_data->image, PCMK_XA_UNIQUE, unique, PCMK_XA_MAINTENANCE, maintenance, PCMK_XA_MANAGED, managed, PCMK_XA_FAILED, failed, PCMK_XA_DESCRIPTION, desc, NULL); pcmk__assert(rc == pcmk_rc_ok); } id = pcmk__itoa(replica->offset); rc = pe__name_and_nvpairs_xml(out, true, PCMK_XE_REPLICA, PCMK_XA_ID, id, NULL); free(id); pcmk__assert(rc == pcmk_rc_ok); if (print_ip) { out->message(out, (const char *) ip->priv->xml->name, show_opts, ip, only_node, only_rsc); } if (print_child) { out->message(out, (const char *) child->priv->xml->name, show_opts, child, only_node, only_rsc); } if (print_ctnr) { out->message(out, (const char *) container->priv->xml->name, show_opts, container, only_node, only_rsc); } if (print_remote) { out->message(out, (const char *) remote->priv->xml->name, show_opts, remote, only_node, only_rsc); } pcmk__output_xml_pop_parent(out); // replica } if (printed_header) { pcmk__output_xml_pop_parent(out); // bundle } return rc; } static void pe__bundle_replica_output_html(pcmk__output_t *out, pcmk__bundle_replica_t *replica, pcmk_node_t *node, uint32_t show_opts) { pcmk_resource_t *rsc = replica->child; int offset = 0; char buffer[LINE_MAX]; if(rsc == NULL) { rsc = replica->container; } if (replica->remote) { offset += snprintf(buffer + offset, LINE_MAX - offset, "%s", rsc_printable_id(replica->remote)); } else { offset += snprintf(buffer + offset, LINE_MAX - offset, "%s", rsc_printable_id(replica->container)); } if (replica->ipaddr) { offset += snprintf(buffer + offset, LINE_MAX - offset, " (%s)", replica->ipaddr); } pe__common_output_html(out, rsc, buffer, node, show_opts); } /*! * \internal * \brief Get a string describing a resource's unmanaged state or lack thereof * * \param[in] rsc Resource to describe * * \return A string indicating that a resource is in maintenance mode or * otherwise unmanaged, or an empty string otherwise */ static const char * get_unmanaged_str(const pcmk_resource_t *rsc) { if (pcmk_is_set(rsc->flags, pcmk__rsc_maintenance)) { return " (maintenance)"; } if (!pcmk_is_set(rsc->flags, pcmk__rsc_managed)) { return " (unmanaged)"; } return ""; } PCMK__OUTPUT_ARGS("bundle", "uint32_t", "pcmk_resource_t *", "GList *", "GList *") int pe__bundle_html(pcmk__output_t *out, va_list args) { uint32_t show_opts = va_arg(args, uint32_t); pcmk_resource_t *rsc = va_arg(args, pcmk_resource_t *); GList *only_node = va_arg(args, GList *); GList *only_rsc = va_arg(args, GList *); const char *desc = NULL; pe__bundle_variant_data_t *bundle_data = NULL; int rc = pcmk_rc_no_output; gboolean print_everything = TRUE; pcmk__assert(rsc != NULL); get_bundle_variant_data(bundle_data, rsc); desc = pe__resource_description(rsc, show_opts); if (rsc->priv->fns->is_filtered(rsc, only_rsc, TRUE)) { return rc; } print_everything = pcmk__str_in_list(rsc->id, only_rsc, pcmk__str_star_matches); for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pcmk__bundle_replica_t *replica = gIter->data; pcmk_resource_t *ip = replica->ip; pcmk_resource_t *child = replica->child; pcmk_resource_t *container = replica->container; pcmk_resource_t *remote = replica->remote; gboolean print_ip, print_child, print_ctnr, print_remote; pcmk__assert(replica != NULL); if (pcmk__rsc_filtered_by_node(container, only_node)) { continue; } print_ip = (ip != NULL) && !ip->priv->fns->is_filtered(ip, only_rsc, print_everything); print_child = (child != NULL) && !child->priv->fns->is_filtered(child, only_rsc, print_everything); print_ctnr = !container->priv->fns->is_filtered(container, only_rsc, print_everything); print_remote = (remote != NULL) && !remote->priv->fns->is_filtered(remote, only_rsc, print_everything); if (pcmk_is_set(show_opts, pcmk_show_implicit_rscs) || (print_everything == FALSE && (print_ip || print_child || print_ctnr || print_remote))) { /* The text output messages used below require pe_print_implicit to * be set to do anything. */ uint32_t new_show_opts = show_opts | pcmk_show_implicit_rscs; PCMK__OUTPUT_LIST_HEADER(out, FALSE, rc, "Container bundle%s: %s [%s]%s%s%s%s%s", (bundle_data->nreplicas > 1)? " set" : "", rsc->id, bundle_data->image, pcmk_is_set(rsc->flags, pcmk__rsc_unique)? " (unique)" : "", desc ? " (" : "", desc ? desc : "", desc ? ")" : "", get_unmanaged_str(rsc)); if (pcmk__list_of_multiple(bundle_data->replicas)) { out->begin_list(out, NULL, NULL, "Replica[%d]", replica->offset); } if (print_ip) { out->message(out, (const char *) ip->priv->xml->name, new_show_opts, ip, only_node, only_rsc); } if (print_child) { out->message(out, (const char *) child->priv->xml->name, new_show_opts, child, only_node, only_rsc); } if (print_ctnr) { out->message(out, (const char *) container->priv->xml->name, new_show_opts, container, only_node, only_rsc); } if (print_remote) { out->message(out, (const char *) remote->priv->xml->name, new_show_opts, remote, only_node, only_rsc); } if (pcmk__list_of_multiple(bundle_data->replicas)) { out->end_list(out); } } else if (print_everything == FALSE && !(print_ip || print_child || print_ctnr || print_remote)) { continue; } else { PCMK__OUTPUT_LIST_HEADER(out, FALSE, rc, "Container bundle%s: %s [%s]%s%s%s%s%s", (bundle_data->nreplicas > 1)? " set" : "", rsc->id, bundle_data->image, pcmk_is_set(rsc->flags, pcmk__rsc_unique)? " (unique)" : "", desc ? " (" : "", desc ? desc : "", desc ? ")" : "", get_unmanaged_str(rsc)); pe__bundle_replica_output_html(out, replica, pcmk__current_node(container), show_opts); } } PCMK__OUTPUT_LIST_FOOTER(out, rc); return rc; } static void pe__bundle_replica_output_text(pcmk__output_t *out, pcmk__bundle_replica_t *replica, pcmk_node_t *node, uint32_t show_opts) { const pcmk_resource_t *rsc = replica->child; int offset = 0; char buffer[LINE_MAX]; if(rsc == NULL) { rsc = replica->container; } if (replica->remote) { offset += snprintf(buffer + offset, LINE_MAX - offset, "%s", rsc_printable_id(replica->remote)); } else { offset += snprintf(buffer + offset, LINE_MAX - offset, "%s", rsc_printable_id(replica->container)); } if (replica->ipaddr) { offset += snprintf(buffer + offset, LINE_MAX - offset, " (%s)", replica->ipaddr); } pe__common_output_text(out, rsc, buffer, node, show_opts); } PCMK__OUTPUT_ARGS("bundle", "uint32_t", "pcmk_resource_t *", "GList *", "GList *") int pe__bundle_text(pcmk__output_t *out, va_list args) { uint32_t show_opts = va_arg(args, uint32_t); pcmk_resource_t *rsc = va_arg(args, pcmk_resource_t *); GList *only_node = va_arg(args, GList *); GList *only_rsc = va_arg(args, GList *); const char *desc = NULL; pe__bundle_variant_data_t *bundle_data = NULL; int rc = pcmk_rc_no_output; gboolean print_everything = TRUE; desc = pe__resource_description(rsc, show_opts); pcmk__assert(rsc != NULL); get_bundle_variant_data(bundle_data, rsc); if (rsc->priv->fns->is_filtered(rsc, only_rsc, TRUE)) { return rc; } print_everything = pcmk__str_in_list(rsc->id, only_rsc, pcmk__str_star_matches); for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pcmk__bundle_replica_t *replica = gIter->data; pcmk_resource_t *ip = replica->ip; pcmk_resource_t *child = replica->child; pcmk_resource_t *container = replica->container; pcmk_resource_t *remote = replica->remote; gboolean print_ip, print_child, print_ctnr, print_remote; pcmk__assert(replica != NULL); if (pcmk__rsc_filtered_by_node(container, only_node)) { continue; } print_ip = (ip != NULL) && !ip->priv->fns->is_filtered(ip, only_rsc, print_everything); print_child = (child != NULL) && !child->priv->fns->is_filtered(child, only_rsc, print_everything); print_ctnr = !container->priv->fns->is_filtered(container, only_rsc, print_everything); print_remote = (remote != NULL) && !remote->priv->fns->is_filtered(remote, only_rsc, print_everything); if (pcmk_is_set(show_opts, pcmk_show_implicit_rscs) || (print_everything == FALSE && (print_ip || print_child || print_ctnr || print_remote))) { /* The text output messages used below require pe_print_implicit to * be set to do anything. */ uint32_t new_show_opts = show_opts | pcmk_show_implicit_rscs; PCMK__OUTPUT_LIST_HEADER(out, FALSE, rc, "Container bundle%s: %s [%s]%s%s%s%s%s", (bundle_data->nreplicas > 1)? " set" : "", rsc->id, bundle_data->image, pcmk_is_set(rsc->flags, pcmk__rsc_unique)? " (unique)" : "", desc ? " (" : "", desc ? desc : "", desc ? ")" : "", get_unmanaged_str(rsc)); if (pcmk__list_of_multiple(bundle_data->replicas)) { out->list_item(out, NULL, "Replica[%d]", replica->offset); } out->begin_list(out, NULL, NULL, NULL); if (print_ip) { out->message(out, (const char *) ip->priv->xml->name, new_show_opts, ip, only_node, only_rsc); } if (print_child) { out->message(out, (const char *) child->priv->xml->name, new_show_opts, child, only_node, only_rsc); } if (print_ctnr) { out->message(out, (const char *) container->priv->xml->name, new_show_opts, container, only_node, only_rsc); } if (print_remote) { out->message(out, (const char *) remote->priv->xml->name, new_show_opts, remote, only_node, only_rsc); } out->end_list(out); } else if (print_everything == FALSE && !(print_ip || print_child || print_ctnr || print_remote)) { continue; } else { PCMK__OUTPUT_LIST_HEADER(out, FALSE, rc, "Container bundle%s: %s [%s]%s%s%s%s%s", (bundle_data->nreplicas > 1)? " set" : "", rsc->id, bundle_data->image, pcmk_is_set(rsc->flags, pcmk__rsc_unique)? " (unique)" : "", desc ? " (" : "", desc ? desc : "", desc ? ")" : "", get_unmanaged_str(rsc)); pe__bundle_replica_output_text(out, replica, pcmk__current_node(container), show_opts); } } PCMK__OUTPUT_LIST_FOOTER(out, rc); return rc; } static void free_bundle_replica(pcmk__bundle_replica_t *replica) { if (replica == NULL) { return; } pcmk__free_node_copy(replica->node); replica->node = NULL; if (replica->ip) { pcmk__xml_free(replica->ip->priv->xml); replica->ip->priv->xml = NULL; - replica->ip->priv->fns->free(replica->ip); + pcmk__free_resource(replica->ip); } if (replica->container) { pcmk__xml_free(replica->container->priv->xml); replica->container->priv->xml = NULL; - replica->container->priv->fns->free(replica->container); + pcmk__free_resource(replica->container); } if (replica->remote) { pcmk__xml_free(replica->remote->priv->xml); replica->remote->priv->xml = NULL; - replica->remote->priv->fns->free(replica->remote); + pcmk__free_resource(replica->remote); } free(replica->ipaddr); free(replica); } void pe__free_bundle(pcmk_resource_t *rsc) { pe__bundle_variant_data_t *bundle_data = NULL; CRM_CHECK(rsc != NULL, return); get_bundle_variant_data(bundle_data, rsc); pcmk__rsc_trace(rsc, "Freeing %s", rsc->id); free(bundle_data->prefix); free(bundle_data->image); free(bundle_data->control_port); free(bundle_data->host_network); free(bundle_data->host_netmask); free(bundle_data->ip_range_start); free(bundle_data->container_network); free(bundle_data->launcher_options); free(bundle_data->container_command); g_free(bundle_data->container_host_options); g_list_free_full(bundle_data->replicas, (GDestroyNotify) free_bundle_replica); g_list_free_full(bundle_data->mounts, (GDestroyNotify)mount_free); g_list_free_full(bundle_data->ports, (GDestroyNotify)port_free); g_list_free(rsc->priv->children); if(bundle_data->child) { pcmk__xml_free(bundle_data->child->priv->xml); bundle_data->child->priv->xml = NULL; - bundle_data->child->priv->fns->free(bundle_data->child); + pcmk__free_resource(bundle_data->child); } common_free(rsc); } enum rsc_role_e pe__bundle_resource_state(const pcmk_resource_t *rsc, gboolean current) { enum rsc_role_e container_role = pcmk_role_unknown; return container_role; } /*! * \brief Get the number of configured replicas in a bundle * * \param[in] rsc Bundle resource * * \return Number of configured replicas, or 0 on error */ int pe_bundle_replicas(const pcmk_resource_t *rsc) { if (pcmk__is_bundle(rsc)) { pe__bundle_variant_data_t *bundle_data = NULL; get_bundle_variant_data(bundle_data, rsc); return bundle_data->nreplicas; } return 0; } void pe__count_bundle(pcmk_resource_t *rsc) { pe__bundle_variant_data_t *bundle_data = NULL; get_bundle_variant_data(bundle_data, rsc); for (GList *item = bundle_data->replicas; item != NULL; item = item->next) { pcmk__bundle_replica_t *replica = item->data; if (replica->ip) { replica->ip->priv->fns->count(replica->ip); } if (replica->child) { replica->child->priv->fns->count(replica->child); } if (replica->container) { replica->container->priv->fns->count(replica->container); } if (replica->remote) { replica->remote->priv->fns->count(replica->remote); } } } gboolean pe__bundle_is_filtered(const pcmk_resource_t *rsc, GList *only_rsc, gboolean check_parent) { gboolean passes = FALSE; pe__bundle_variant_data_t *bundle_data = NULL; if (pcmk__str_in_list(rsc_printable_id(rsc), only_rsc, pcmk__str_star_matches)) { passes = TRUE; } else { get_bundle_variant_data(bundle_data, rsc); for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pcmk__bundle_replica_t *replica = gIter->data; pcmk_resource_t *ip = replica->ip; pcmk_resource_t *child = replica->child; pcmk_resource_t *container = replica->container; pcmk_resource_t *remote = replica->remote; if ((ip != NULL) && !ip->priv->fns->is_filtered(ip, only_rsc, FALSE)) { passes = TRUE; break; } if ((child != NULL) && !child->priv->fns->is_filtered(child, only_rsc, FALSE)) { passes = TRUE; break; } if (!container->priv->fns->is_filtered(container, only_rsc, FALSE)) { passes = TRUE; break; } if ((remote != NULL) && !remote->priv->fns->is_filtered(remote, only_rsc, FALSE)) { passes = TRUE; break; } } } return !passes; } /*! * \internal * \brief Get a list of a bundle's containers * * \param[in] bundle Bundle resource * * \return Newly created list of \p bundle's containers * \note It is the caller's responsibility to free the result with * g_list_free(). */ GList * pe__bundle_containers(const pcmk_resource_t *bundle) { GList *containers = NULL; const pe__bundle_variant_data_t *data = NULL; get_bundle_variant_data(data, bundle); for (GList *iter = data->replicas; iter != NULL; iter = iter->next) { pcmk__bundle_replica_t *replica = iter->data; containers = g_list_append(containers, replica->container); } return containers; } // Bundle implementation of pcmk__rsc_methods_t:active_node() pcmk_node_t * pe__bundle_active_node(const pcmk_resource_t *rsc, unsigned int *count_all, unsigned int *count_clean) { pcmk_node_t *active = NULL; pcmk_node_t *node = NULL; pcmk_resource_t *container = NULL; GList *containers = NULL; GList *iter = NULL; GHashTable *nodes = NULL; const pe__bundle_variant_data_t *data = NULL; if (count_all != NULL) { *count_all = 0; } if (count_clean != NULL) { *count_clean = 0; } if (rsc == NULL) { return NULL; } /* For the purposes of this method, we only care about where the bundle's * containers are active, so build a list of active containers. */ get_bundle_variant_data(data, rsc); for (iter = data->replicas; iter != NULL; iter = iter->next) { pcmk__bundle_replica_t *replica = iter->data; if (replica->container->priv->active_nodes != NULL) { containers = g_list_append(containers, replica->container); } } if (containers == NULL) { return NULL; } /* If the bundle has only a single active container, just use that * container's method. If live migration is ever supported for bundle * containers, this will allow us to prefer the migration source when there * is only one container and it is migrating. For now, this just lets us * avoid creating the nodes table. */ if (pcmk__list_of_1(containers)) { container = containers->data; node = container->priv->fns->active_node(container, count_all, count_clean); g_list_free(containers); return node; } // Add all containers' active nodes to a hash table (for uniqueness) nodes = g_hash_table_new(NULL, NULL); for (iter = containers; iter != NULL; iter = iter->next) { container = iter->data; for (GList *node_iter = container->priv->active_nodes; node_iter != NULL; node_iter = node_iter->next) { node = node_iter->data; // If insert returns true, we haven't counted this node yet if (g_hash_table_insert(nodes, (gpointer) node->details, (gpointer) node) && !pe__count_active_node(rsc, node, &active, count_all, count_clean)) { goto done; } } } done: g_list_free(containers); g_hash_table_destroy(nodes); return active; } /*! * \internal * \brief Get maximum bundle resource instances per node * * \param[in] rsc Bundle resource to check * * \return Maximum number of \p rsc instances that can be active on one node */ unsigned int pe__bundle_max_per_node(const pcmk_resource_t *rsc) { pe__bundle_variant_data_t *bundle_data = NULL; get_bundle_variant_data(bundle_data, rsc); pcmk__assert(bundle_data->nreplicas_per_host >= 0); return (unsigned int) bundle_data->nreplicas_per_host; } diff --git a/lib/pengine/clone.c b/lib/pengine/clone.c index 24953cd131..2764af26fc 100644 --- a/lib/pengine/clone.c +++ b/lib/pengine/clone.c @@ -1,1256 +1,1256 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include typedef struct clone_variant_data_s { int clone_max; int clone_node_max; int promoted_max; int promoted_node_max; int total_clones; uint32_t flags; // Group of enum pcmk__clone_flags notify_data_t *stop_notify; notify_data_t *start_notify; notify_data_t *demote_notify; notify_data_t *promote_notify; xmlNode *xml_obj_child; } clone_variant_data_t; #define get_clone_variant_data(data, rsc) do { \ pcmk__assert(pcmk__is_clone(rsc)); \ data = rsc->priv->variant_opaque; \ } while (0) /*! * \internal * \brief Return the maximum number of clone instances allowed to be run * * \param[in] clone Clone or clone instance to check * * \return Maximum instances for \p clone */ int pe__clone_max(const pcmk_resource_t *clone) { const clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, pe__const_top_resource(clone, false)); return clone_data->clone_max; } /*! * \internal * \brief Return the maximum number of clone instances allowed per node * * \param[in] clone Promotable clone or clone instance to check * * \return Maximum allowed instances per node for \p clone */ int pe__clone_node_max(const pcmk_resource_t *clone) { const clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, pe__const_top_resource(clone, false)); return clone_data->clone_node_max; } /*! * \internal * \brief Return the maximum number of clone instances allowed to be promoted * * \param[in] clone Promotable clone or clone instance to check * * \return Maximum promoted instances for \p clone */ int pe__clone_promoted_max(const pcmk_resource_t *clone) { clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, pe__const_top_resource(clone, false)); return clone_data->promoted_max; } /*! * \internal * \brief Return the maximum number of clone instances allowed to be promoted * * \param[in] clone Promotable clone or clone instance to check * * \return Maximum promoted instances for \p clone */ int pe__clone_promoted_node_max(const pcmk_resource_t *clone) { clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, pe__const_top_resource(clone, false)); return clone_data->promoted_node_max; } static GList * sorted_hash_table_values(GHashTable *table) { GList *retval = NULL; GHashTableIter iter; gpointer key, value; g_hash_table_iter_init(&iter, table); while (g_hash_table_iter_next(&iter, &key, &value)) { if (!g_list_find_custom(retval, value, (GCompareFunc) strcmp)) { retval = g_list_prepend(retval, (char *) value); } } retval = g_list_sort(retval, (GCompareFunc) strcmp); return retval; } static GList * nodes_with_status(GHashTable *table, const char *status) { GList *retval = NULL; GHashTableIter iter; gpointer key, value; g_hash_table_iter_init(&iter, table); while (g_hash_table_iter_next(&iter, &key, &value)) { if (!strcmp((char *) value, status)) { retval = g_list_prepend(retval, key); } } retval = g_list_sort(retval, (GCompareFunc) pcmk__numeric_strcasecmp); return retval; } static GString * node_list_to_str(const GList *list) { GString *retval = NULL; for (const GList *iter = list; iter != NULL; iter = iter->next) { pcmk__add_word(&retval, 1024, (const char *) iter->data); } return retval; } static void clone_header(pcmk__output_t *out, int *rc, const pcmk_resource_t *rsc, clone_variant_data_t *clone_data, const char *desc) { GString *attrs = NULL; if (pcmk_is_set(rsc->flags, pcmk__rsc_promotable)) { pcmk__add_separated_word(&attrs, 64, "promotable", ", "); } if (pcmk_is_set(rsc->flags, pcmk__rsc_unique)) { pcmk__add_separated_word(&attrs, 64, "unique", ", "); } if (pe__resource_is_disabled(rsc)) { pcmk__add_separated_word(&attrs, 64, "disabled", ", "); } if (pcmk_is_set(rsc->flags, pcmk__rsc_maintenance)) { pcmk__add_separated_word(&attrs, 64, "maintenance", ", "); } else if (!pcmk_is_set(rsc->flags, pcmk__rsc_managed)) { pcmk__add_separated_word(&attrs, 64, "unmanaged", ", "); } if (attrs != NULL) { PCMK__OUTPUT_LIST_HEADER(out, FALSE, *rc, "Clone Set: %s [%s] (%s)%s%s%s", rsc->id, pcmk__xe_id(clone_data->xml_obj_child), (const char *) attrs->str, desc ? " (" : "", desc ? desc : "", desc ? ")" : ""); g_string_free(attrs, TRUE); } else { PCMK__OUTPUT_LIST_HEADER(out, FALSE, *rc, "Clone Set: %s [%s]%s%s%s", rsc->id, pcmk__xe_id(clone_data->xml_obj_child), desc ? " (" : "", desc ? desc : "", desc ? ")" : ""); } } void pe__force_anon(const char *standard, pcmk_resource_t *rsc, const char *rid, pcmk_scheduler_t *scheduler) { if (pcmk__is_clone(rsc)) { clone_variant_data_t *clone_data = rsc->priv->variant_opaque; pcmk__config_warn("Ignoring " PCMK_META_GLOBALLY_UNIQUE " for %s " "because %s resources such as %s can be used only as " "anonymous clones", rsc->id, standard, rid); clone_data->clone_node_max = 1; clone_data->clone_max = QB_MIN(clone_data->clone_max, g_list_length(scheduler->nodes)); } } pcmk_resource_t * pe__create_clone_child(pcmk_resource_t *rsc, pcmk_scheduler_t *scheduler) { gboolean as_orphan = FALSE; char *inc_num = NULL; char *inc_max = NULL; pcmk_resource_t *child_rsc = NULL; xmlNode *child_copy = NULL; clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, rsc); CRM_CHECK(clone_data->xml_obj_child != NULL, return FALSE); if (clone_data->total_clones >= clone_data->clone_max) { // If we've already used all available instances, this is an orphan as_orphan = TRUE; } // Allocate instance numbers in numerical order (starting at 0) inc_num = pcmk__itoa(clone_data->total_clones); inc_max = pcmk__itoa(clone_data->clone_max); child_copy = pcmk__xml_copy(NULL, clone_data->xml_obj_child); crm_xml_add(child_copy, PCMK__META_CLONE, inc_num); if (pe__unpack_resource(child_copy, &child_rsc, rsc, scheduler) != pcmk_rc_ok) { goto bail; } /* child_rsc->globally_unique = rsc->globally_unique; */ pcmk__assert(child_rsc != NULL); clone_data->total_clones += 1; pcmk__rsc_trace(child_rsc, "Setting clone attributes for: %s", child_rsc->id); rsc->priv->children = g_list_append(rsc->priv->children, child_rsc); if (as_orphan) { pe__set_resource_flags_recursive(child_rsc, pcmk__rsc_removed); } pcmk__insert_meta(child_rsc->priv, PCMK_META_CLONE_MAX, inc_max); pcmk__rsc_trace(rsc, "Added %s instance %s", rsc->id, child_rsc->id); bail: free(inc_num); free(inc_max); return child_rsc; } /*! * \internal * \brief Unpack a nonnegative integer value from a resource meta-attribute * * \param[in] rsc Resource with meta-attribute * \param[in] meta_name Name of meta-attribute to unpack * \param[in] deprecated_name If not NULL, try unpacking this * if \p meta_name is unset * \param[in] default_value Value to use if unset * * \return Integer parsed from resource's specified meta-attribute if a valid * nonnegative integer, \p default_value if unset, or 0 if invalid */ static int unpack_meta_int(const pcmk_resource_t *rsc, const char *meta_name, const char *deprecated_name, int default_value) { int integer = default_value; const char *value = g_hash_table_lookup(rsc->priv->meta, meta_name); if ((value == NULL) && (deprecated_name != NULL)) { value = g_hash_table_lookup(rsc->priv->meta, deprecated_name); if (value != NULL) { if (pcmk__str_eq(deprecated_name, PCMK__META_PROMOTED_MAX_LEGACY, pcmk__str_none)) { pcmk__warn_once(pcmk__wo_clone_master_max, "Support for the " PCMK__META_PROMOTED_MAX_LEGACY " meta-attribute (such as in %s) is deprecated " "and will be removed in a future release. Use the " PCMK_META_PROMOTED_MAX " meta-attribute instead.", rsc->id); } else if (pcmk__str_eq(deprecated_name, PCMK__META_PROMOTED_NODE_MAX_LEGACY, pcmk__str_none)) { pcmk__warn_once(pcmk__wo_clone_master_node_max, "Support for the " PCMK__META_PROMOTED_NODE_MAX_LEGACY " meta-attribute (such as in %s) is deprecated " "and will be removed in a future release. Use the " PCMK_META_PROMOTED_NODE_MAX " meta-attribute instead.", rsc->id); } } } if (value != NULL) { pcmk__scan_min_int(value, &integer, 0); } return integer; } gboolean clone_unpack(pcmk_resource_t *rsc, pcmk_scheduler_t *scheduler) { int lpc = 0; xmlNode *a_child = NULL; xmlNode *xml_obj = rsc->priv->xml; clone_variant_data_t *clone_data = NULL; pcmk__rsc_trace(rsc, "Processing resource %s...", rsc->id); clone_data = pcmk__assert_alloc(1, sizeof(clone_variant_data_t)); rsc->priv->variant_opaque = clone_data; if (pcmk_is_set(rsc->flags, pcmk__rsc_promotable)) { // Use 1 as default but 0 for minimum and invalid // @COMPAT PCMK__META_PROMOTED_MAX_LEGACY deprecated since 2.0.0 clone_data->promoted_max = unpack_meta_int(rsc, PCMK_META_PROMOTED_MAX, PCMK__META_PROMOTED_MAX_LEGACY, 1); // Use 1 as default but 0 for minimum and invalid // @COMPAT PCMK__META_PROMOTED_NODE_MAX_LEGACY deprecated since 2.0.0 clone_data->promoted_node_max = unpack_meta_int(rsc, PCMK_META_PROMOTED_NODE_MAX, PCMK__META_PROMOTED_NODE_MAX_LEGACY, 1); } // Use 1 as default but 0 for minimum and invalid clone_data->clone_node_max = unpack_meta_int(rsc, PCMK_META_CLONE_NODE_MAX, NULL, 1); /* Use number of nodes (but always at least 1, which is handy for crm_verify * for a CIB without nodes) as default, but 0 for minimum and invalid */ clone_data->clone_max = unpack_meta_int(rsc, PCMK_META_CLONE_MAX, NULL, QB_MAX(1, g_list_length(scheduler->nodes))); if (crm_is_true(g_hash_table_lookup(rsc->priv->meta, PCMK_META_ORDERED))) { clone_data->flags = pcmk__set_flags_as(__func__, __LINE__, LOG_TRACE, "Clone", rsc->id, clone_data->flags, pcmk__clone_ordered, "pcmk__clone_ordered"); } if (!pcmk_is_set(rsc->flags, pcmk__rsc_unique) && (clone_data->clone_node_max > 1)) { pcmk__config_err("Ignoring " PCMK_META_CLONE_NODE_MAX " of %d for %s " "because anonymous clones support only one instance " "per node", clone_data->clone_node_max, rsc->id); clone_data->clone_node_max = 1; } pcmk__rsc_trace(rsc, "Options for %s", rsc->id); pcmk__rsc_trace(rsc, "\tClone max: %d", clone_data->clone_max); pcmk__rsc_trace(rsc, "\tClone node max: %d", clone_data->clone_node_max); pcmk__rsc_trace(rsc, "\tClone is unique: %s", pcmk__flag_text(rsc->flags, pcmk__rsc_unique)); pcmk__rsc_trace(rsc, "\tClone is promotable: %s", pcmk__flag_text(rsc->flags, pcmk__rsc_promotable)); // Clones may contain a single group or primitive for (a_child = pcmk__xe_first_child(xml_obj, NULL, NULL, NULL); a_child != NULL; a_child = pcmk__xe_next(a_child, NULL)) { if (pcmk__str_any_of((const char *) a_child->name, PCMK_XE_PRIMITIVE, PCMK_XE_GROUP, NULL)) { clone_data->xml_obj_child = a_child; break; } } if (clone_data->xml_obj_child == NULL) { pcmk__config_err("%s has nothing to clone", rsc->id); return FALSE; } /* * Make clones ever so slightly sticky by default * * This helps ensure clone instances are not shuffled around the cluster * for no benefit in situations when pre-allocation is not appropriate */ if (g_hash_table_lookup(rsc->priv->meta, PCMK_META_RESOURCE_STICKINESS) == NULL) { pcmk__insert_meta(rsc->priv, PCMK_META_RESOURCE_STICKINESS, "1"); } /* This ensures that the PCMK_META_GLOBALLY_UNIQUE value always exists for * children to inherit when being unpacked, as well as in resource agents' * environment. */ pcmk__insert_meta(rsc->priv, PCMK_META_GLOBALLY_UNIQUE, pcmk__flag_text(rsc->flags, pcmk__rsc_unique)); if (clone_data->clone_max <= 0) { /* Create one child instance so that unpack_find_resource() will hook up * any orphans up to the parent correctly. */ if (pe__create_clone_child(rsc, scheduler) == NULL) { return FALSE; } } else { // Create a child instance for each available instance number for (lpc = 0; lpc < clone_data->clone_max; lpc++) { if (pe__create_clone_child(rsc, scheduler) == NULL) { return FALSE; } } } pcmk__rsc_trace(rsc, "Added %d children to resource %s...", clone_data->clone_max, rsc->id); return TRUE; } gboolean clone_active(pcmk_resource_t * rsc, gboolean all) { for (GList *gIter = rsc->priv->children; gIter != NULL; gIter = gIter->next) { pcmk_resource_t *child_rsc = (pcmk_resource_t *) gIter->data; gboolean child_active = child_rsc->priv->fns->active(child_rsc, all); if (all == FALSE && child_active) { return TRUE; } else if (all && child_active == FALSE) { return FALSE; } } if (all) { return TRUE; } else { return FALSE; } } static const char * configured_role_str(pcmk_resource_t * rsc) { const char *target_role = g_hash_table_lookup(rsc->priv->meta, PCMK_META_TARGET_ROLE); if ((target_role == NULL) && (rsc->priv->children != NULL)) { // Any instance will do pcmk_resource_t *instance = rsc->priv->children->data; target_role = g_hash_table_lookup(instance->priv->meta, PCMK_META_TARGET_ROLE); } return target_role; } static enum rsc_role_e configured_role(pcmk_resource_t *rsc) { enum rsc_role_e role = pcmk_role_unknown; const char *target_role = configured_role_str(rsc); if (target_role != NULL) { role = pcmk_parse_role(target_role); if (role == pcmk_role_unknown) { pcmk__config_err("Invalid " PCMK_META_TARGET_ROLE " for resource %s", rsc->id); } } return role; } bool is_set_recursive(const pcmk_resource_t *rsc, long long flag, bool any) { bool all = !any; if (pcmk_is_set(rsc->flags, flag)) { if(any) { return TRUE; } } else if(all) { return FALSE; } for (GList *gIter = rsc->priv->children; gIter != NULL; gIter = gIter->next) { if(is_set_recursive(gIter->data, flag, any)) { if(any) { return TRUE; } } else if(all) { return FALSE; } } if(all) { return TRUE; } return FALSE; } PCMK__OUTPUT_ARGS("clone", "uint32_t", "pcmk_resource_t *", "GList *", "GList *") int pe__clone_xml(pcmk__output_t *out, va_list args) { uint32_t show_opts = va_arg(args, uint32_t); pcmk_resource_t *rsc = va_arg(args, pcmk_resource_t *); GList *only_node = va_arg(args, GList *); GList *only_rsc = va_arg(args, GList *); GList *all = NULL; int rc = pcmk_rc_no_output; gboolean printed_header = FALSE; gboolean print_everything = TRUE; if (rsc->priv->fns->is_filtered(rsc, only_rsc, TRUE)) { return rc; } print_everything = pcmk__str_in_list(rsc_printable_id(rsc), only_rsc, pcmk__str_star_matches) || (strstr(rsc->id, ":") != NULL && pcmk__str_in_list(rsc->id, only_rsc, pcmk__str_star_matches)); all = g_list_prepend(all, (gpointer) "*"); for (GList *gIter = rsc->priv->children; gIter != NULL; gIter = gIter->next) { pcmk_resource_t *child_rsc = (pcmk_resource_t *) gIter->data; if (pcmk__rsc_filtered_by_node(child_rsc, only_node)) { continue; } if (child_rsc->priv->fns->is_filtered(child_rsc, only_rsc, print_everything)) { continue; } if (!printed_header) { const char *multi_state = pcmk__flag_text(rsc->flags, pcmk__rsc_promotable); const char *unique = pcmk__flag_text(rsc->flags, pcmk__rsc_unique); const char *maintenance = pcmk__flag_text(rsc->flags, pcmk__rsc_maintenance); const char *managed = pcmk__flag_text(rsc->flags, pcmk__rsc_managed); const char *disabled = pcmk__btoa(pe__resource_is_disabled(rsc)); const char *failed = pcmk__flag_text(rsc->flags, pcmk__rsc_failed); const char *ignored = pcmk__flag_text(rsc->flags, pcmk__rsc_ignore_failure); const char *target_role = configured_role_str(rsc); const char *desc = pe__resource_description(rsc, show_opts); printed_header = TRUE; rc = pe__name_and_nvpairs_xml(out, true, PCMK_XE_CLONE, PCMK_XA_ID, rsc->id, PCMK_XA_MULTI_STATE, multi_state, PCMK_XA_UNIQUE, unique, PCMK_XA_MAINTENANCE, maintenance, PCMK_XA_MANAGED, managed, PCMK_XA_DISABLED, disabled, PCMK_XA_FAILED, failed, PCMK_XA_FAILURE_IGNORED, ignored, PCMK_XA_TARGET_ROLE, target_role, PCMK_XA_DESCRIPTION, desc, NULL); pcmk__assert(rc == pcmk_rc_ok); } out->message(out, (const char *) child_rsc->priv->xml->name, show_opts, child_rsc, only_node, all); } if (printed_header) { pcmk__output_xml_pop_parent(out); } g_list_free(all); return rc; } PCMK__OUTPUT_ARGS("clone", "uint32_t", "pcmk_resource_t *", "GList *", "GList *") int pe__clone_default(pcmk__output_t *out, va_list args) { uint32_t show_opts = va_arg(args, uint32_t); pcmk_resource_t *rsc = va_arg(args, pcmk_resource_t *); GList *only_node = va_arg(args, GList *); GList *only_rsc = va_arg(args, GList *); GHashTable *stopped = NULL; GString *list_text = NULL; GList *promoted_list = NULL; GList *started_list = NULL; GList *gIter = NULL; const char *desc = NULL; clone_variant_data_t *clone_data = NULL; int active_instances = 0; int rc = pcmk_rc_no_output; gboolean print_everything = TRUE; desc = pe__resource_description(rsc, show_opts); get_clone_variant_data(clone_data, rsc); if (rsc->priv->fns->is_filtered(rsc, only_rsc, TRUE)) { return rc; } print_everything = pcmk__str_in_list(rsc_printable_id(rsc), only_rsc, pcmk__str_star_matches) || (strstr(rsc->id, ":") != NULL && pcmk__str_in_list(rsc->id, only_rsc, pcmk__str_star_matches)); for (gIter = rsc->priv->children; gIter != NULL; gIter = gIter->next) { gboolean print_full = FALSE; pcmk_resource_t *child_rsc = (pcmk_resource_t *) gIter->data; gboolean partially_active = child_rsc->priv->fns->active(child_rsc, FALSE); if (pcmk__rsc_filtered_by_node(child_rsc, only_node)) { continue; } if (child_rsc->priv->fns->is_filtered(child_rsc, only_rsc, print_everything)) { continue; } if (pcmk_is_set(show_opts, pcmk_show_clone_detail)) { print_full = TRUE; } if (pcmk_is_set(rsc->flags, pcmk__rsc_unique)) { // Print individual instance when unique (except stopped orphans) if (partially_active || !pcmk_is_set(rsc->flags, pcmk__rsc_removed)) { print_full = TRUE; } // Everything else in this block is for anonymous clones } else if (pcmk_is_set(show_opts, pcmk_show_pending) && (child_rsc->priv->pending_action != NULL) && (strcmp(child_rsc->priv->pending_action, "probe") != 0)) { // Print individual instance when non-probe action is pending print_full = TRUE; } else if (partially_active == FALSE) { // List stopped instances when requested (except orphans) if (!pcmk_is_set(child_rsc->flags, pcmk__rsc_removed) && !pcmk_is_set(show_opts, pcmk_show_clone_detail) && pcmk_is_set(show_opts, pcmk_show_inactive_rscs)) { if (stopped == NULL) { stopped = pcmk__strkey_table(free, free); } pcmk__insert_dup(stopped, child_rsc->id, "Stopped"); } } else if (is_set_recursive(child_rsc, pcmk__rsc_removed, TRUE) || !is_set_recursive(child_rsc, pcmk__rsc_managed, FALSE) || is_set_recursive(child_rsc, pcmk__rsc_failed, TRUE)) { // Print individual instance when active orphaned/unmanaged/failed print_full = TRUE; } else if (child_rsc->priv->fns->active(child_rsc, TRUE)) { // Instance of fully active anonymous clone pcmk_node_t *location = NULL; location = child_rsc->priv->fns->location(child_rsc, NULL, pcmk__rsc_node_current); if (location) { // Instance is active on a single node enum rsc_role_e a_role; a_role = child_rsc->priv->fns->state(child_rsc, TRUE); if (location->details->online == FALSE && location->details->unclean) { print_full = TRUE; } else if (a_role > pcmk_role_unpromoted) { promoted_list = g_list_append(promoted_list, location); } else { started_list = g_list_append(started_list, location); } } else { /* uncolocated group - bleh */ print_full = TRUE; } } else { // Instance of partially active anonymous clone print_full = TRUE; } if (print_full) { GList *all = NULL; clone_header(out, &rc, rsc, clone_data, desc); /* Print every resource that's a child of this clone. */ all = g_list_prepend(all, (gpointer) "*"); out->message(out, (const char *) child_rsc->priv->xml->name, show_opts, child_rsc, only_node, all); g_list_free(all); } } if (pcmk_is_set(show_opts, pcmk_show_clone_detail)) { PCMK__OUTPUT_LIST_FOOTER(out, rc); return pcmk_rc_ok; } /* Promoted */ promoted_list = g_list_sort(promoted_list, pe__cmp_node_name); for (gIter = promoted_list; gIter; gIter = gIter->next) { pcmk_node_t *host = gIter->data; if (!pcmk__str_in_list(host->priv->name, only_node, pcmk__str_star_matches|pcmk__str_casei)) { continue; } pcmk__add_word(&list_text, 1024, host->priv->name); active_instances++; } g_list_free(promoted_list); if ((list_text != NULL) && (list_text->len > 0)) { clone_header(out, &rc, rsc, clone_data, desc); out->list_item(out, NULL, PCMK_ROLE_PROMOTED ": [ %s ]", (const char *) list_text->str); g_string_truncate(list_text, 0); } /* Started/Unpromoted */ started_list = g_list_sort(started_list, pe__cmp_node_name); for (gIter = started_list; gIter; gIter = gIter->next) { pcmk_node_t *host = gIter->data; if (!pcmk__str_in_list(host->priv->name, only_node, pcmk__str_star_matches|pcmk__str_casei)) { continue; } pcmk__add_word(&list_text, 1024, host->priv->name); active_instances++; } g_list_free(started_list); if ((list_text != NULL) && (list_text->len > 0)) { clone_header(out, &rc, rsc, clone_data, desc); if (pcmk_is_set(rsc->flags, pcmk__rsc_promotable)) { enum rsc_role_e role = configured_role(rsc); if (role == pcmk_role_unpromoted) { out->list_item(out, NULL, PCMK_ROLE_UNPROMOTED " (" PCMK_META_TARGET_ROLE "): [ %s ]", (const char *) list_text->str); } else { out->list_item(out, NULL, PCMK_ROLE_UNPROMOTED ": [ %s ]", (const char *) list_text->str); } } else { out->list_item(out, NULL, "Started: [ %s ]", (const char *) list_text->str); } } if (list_text != NULL) { g_string_free(list_text, TRUE); } if (pcmk_is_set(show_opts, pcmk_show_inactive_rscs)) { if (!pcmk_is_set(rsc->flags, pcmk__rsc_unique) && (clone_data->clone_max > active_instances)) { GList *nIter; GList *list = g_hash_table_get_values(rsc->priv->allowed_nodes); /* Custom stopped table for non-unique clones */ if (stopped != NULL) { g_hash_table_destroy(stopped); stopped = NULL; } if (list == NULL) { /* Clusters with PCMK_OPT_SYMMETRIC_CLUSTER=false haven't * calculated allowed nodes yet. If we've not probed for them * yet, the Stopped list will be empty. */ list = g_hash_table_get_values(rsc->priv->probed_nodes); } list = g_list_sort(list, pe__cmp_node_name); for (nIter = list; nIter != NULL; nIter = nIter->next) { pcmk_node_t *node = (pcmk_node_t *) nIter->data; if ((pcmk__find_node_in_list(rsc->priv->active_nodes, node->priv->name) == NULL) && pcmk__str_in_list(node->priv->name, only_node, pcmk__str_star_matches|pcmk__str_casei)) { xmlNode *probe_op = NULL; const char *state = "Stopped"; if (configured_role(rsc) == pcmk_role_stopped) { state = "Stopped (disabled)"; } if (stopped == NULL) { stopped = pcmk__strkey_table(free, free); } probe_op = pe__failed_probe_for_rsc(rsc, node->priv->name); if (probe_op != NULL) { int rc; pcmk__scan_min_int(crm_element_value(probe_op, PCMK__XA_RC_CODE), &rc, 0); g_hash_table_insert(stopped, strdup(node->priv->name), crm_strdup_printf("Stopped (%s)", crm_exit_str(rc))); } else { pcmk__insert_dup(stopped, node->priv->name, state); } } } g_list_free(list); } if (stopped != NULL) { GList *list = sorted_hash_table_values(stopped); clone_header(out, &rc, rsc, clone_data, desc); for (GList *status_iter = list; status_iter != NULL; status_iter = status_iter->next) { const char *status = status_iter->data; GList *nodes = nodes_with_status(stopped, status); GString *nodes_str = node_list_to_str(nodes); if (nodes_str != NULL) { if (nodes_str->len > 0) { out->list_item(out, NULL, "%s: [ %s ]", status, (const char *) nodes_str->str); } g_string_free(nodes_str, TRUE); } g_list_free(nodes); } g_list_free(list); g_hash_table_destroy(stopped); /* If there are no instances of this clone (perhaps because there are no * nodes configured), simply output the clone header by itself. This can * come up in PCS testing. */ } else if (active_instances == 0) { clone_header(out, &rc, rsc, clone_data, desc); PCMK__OUTPUT_LIST_FOOTER(out, rc); return rc; } } PCMK__OUTPUT_LIST_FOOTER(out, rc); return rc; } void clone_free(pcmk_resource_t * rsc) { clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, rsc); pcmk__rsc_trace(rsc, "Freeing %s", rsc->id); for (GList *gIter = rsc->priv->children; gIter != NULL; gIter = gIter->next) { pcmk_resource_t *child_rsc = (pcmk_resource_t *) gIter->data; pcmk__assert(child_rsc != NULL); pcmk__rsc_trace(child_rsc, "Freeing child %s", child_rsc->id); pcmk__xml_free(child_rsc->priv->xml); child_rsc->priv->xml = NULL; /* There could be a saved unexpanded xml */ pcmk__xml_free(child_rsc->priv->orig_xml); child_rsc->priv->orig_xml = NULL; - child_rsc->priv->fns->free(child_rsc); + pcmk__free_resource(child_rsc); } g_list_free(rsc->priv->children); if (clone_data) { pcmk__assert((clone_data->demote_notify == NULL) && (clone_data->stop_notify == NULL) && (clone_data->start_notify == NULL) && (clone_data->promote_notify == NULL)); } common_free(rsc); } enum rsc_role_e clone_resource_state(const pcmk_resource_t * rsc, gboolean current) { enum rsc_role_e clone_role = pcmk_role_unknown; for (GList *gIter = rsc->priv->children; gIter != NULL; gIter = gIter->next) { pcmk_resource_t *child_rsc = (pcmk_resource_t *) gIter->data; enum rsc_role_e a_role = child_rsc->priv->fns->state(child_rsc, current); if (a_role > clone_role) { clone_role = a_role; } } pcmk__rsc_trace(rsc, "%s role: %s", rsc->id, pcmk_role_text(clone_role)); return clone_role; } /*! * \internal * \brief Check whether a clone has an instance for every node * * \param[in] rsc Clone to check * \param[in] scheduler Scheduler data */ bool pe__is_universal_clone(const pcmk_resource_t *rsc, const pcmk_scheduler_t *scheduler) { if (pcmk__is_clone(rsc)) { clone_variant_data_t *clone_data = rsc->priv->variant_opaque; if (clone_data->clone_max == g_list_length(scheduler->nodes)) { return TRUE; } } return FALSE; } gboolean pe__clone_is_filtered(const pcmk_resource_t *rsc, GList *only_rsc, gboolean check_parent) { gboolean passes = FALSE; clone_variant_data_t *clone_data = NULL; if (pcmk__str_in_list(rsc_printable_id(rsc), only_rsc, pcmk__str_star_matches)) { passes = TRUE; } else { get_clone_variant_data(clone_data, rsc); passes = pcmk__str_in_list(pcmk__xe_id(clone_data->xml_obj_child), only_rsc, pcmk__str_star_matches); if (!passes) { for (const GList *iter = rsc->priv->children; iter != NULL; iter = iter->next) { const pcmk_resource_t *child_rsc = NULL; child_rsc = (const pcmk_resource_t *) iter->data; if (!child_rsc->priv->fns->is_filtered(child_rsc, only_rsc, FALSE)) { passes = TRUE; break; } } } } return !passes; } const char * pe__clone_child_id(const pcmk_resource_t *rsc) { clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, rsc); return pcmk__xe_id(clone_data->xml_obj_child); } /*! * \internal * \brief Check whether a clone is ordered * * \param[in] clone Clone resource to check * * \return true if clone is ordered, otherwise false */ bool pe__clone_is_ordered(const pcmk_resource_t *clone) { clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, clone); return pcmk_is_set(clone_data->flags, pcmk__clone_ordered); } /*! * \internal * \brief Set a clone flag * * \param[in,out] clone Clone resource to set flag for * \param[in] flag Clone flag to set * * \return Standard Pacemaker return code (either pcmk_rc_ok if flag was not * already set or pcmk_rc_already if it was) */ int pe__set_clone_flag(pcmk_resource_t *clone, enum pcmk__clone_flags flag) { clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, clone); if (pcmk_is_set(clone_data->flags, flag)) { return pcmk_rc_already; } clone_data->flags = pcmk__set_flags_as(__func__, __LINE__, LOG_TRACE, "Clone", clone->id, clone_data->flags, flag, "flag"); return pcmk_rc_ok; } /*! * \internal * \brief Check whether a clone flag is set * * \param[in] group Clone resource to check * \param[in] flags Flag or flags to check * * \return \c true if all \p flags are set for \p clone, otherwise \c false */ bool pe__clone_flag_is_set(const pcmk_resource_t *clone, uint32_t flags) { clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, clone); pcmk__assert(clone_data != NULL); return pcmk_all_flags_set(clone_data->flags, flags); } /*! * \internal * \brief Create pseudo-actions needed for promotable clones * * \param[in,out] clone Promotable clone to create actions for * \param[in] any_promoting Whether any instances will be promoted * \param[in] any_demoting Whether any instance will be demoted */ void pe__create_promotable_pseudo_ops(pcmk_resource_t *clone, bool any_promoting, bool any_demoting) { pcmk_action_t *action = NULL; pcmk_action_t *action_complete = NULL; clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, clone); // Create a "promote" action for the clone itself action = pe__new_rsc_pseudo_action(clone, PCMK_ACTION_PROMOTE, !any_promoting, true); // Create a "promoted" action for when all promotions are done action_complete = pe__new_rsc_pseudo_action(clone, PCMK_ACTION_PROMOTED, !any_promoting, true); action_complete->priority = PCMK_SCORE_INFINITY; // Create notification pseudo-actions for promotion if (clone_data->promote_notify == NULL) { clone_data->promote_notify = pe__action_notif_pseudo_ops(clone, PCMK_ACTION_PROMOTE, action, action_complete); } // Create a "demote" action for the clone itself action = pe__new_rsc_pseudo_action(clone, PCMK_ACTION_DEMOTE, !any_demoting, true); // Create a "demoted" action for when all demotions are done action_complete = pe__new_rsc_pseudo_action(clone, PCMK_ACTION_DEMOTED, !any_demoting, true); action_complete->priority = PCMK_SCORE_INFINITY; // Create notification pseudo-actions for demotion if (clone_data->demote_notify == NULL) { clone_data->demote_notify = pe__action_notif_pseudo_ops(clone, PCMK_ACTION_DEMOTE, action, action_complete); if (clone_data->promote_notify != NULL) { order_actions(clone_data->stop_notify->post_done, clone_data->promote_notify->pre, pcmk__ar_ordered); order_actions(clone_data->start_notify->post_done, clone_data->promote_notify->pre, pcmk__ar_ordered); order_actions(clone_data->demote_notify->post_done, clone_data->promote_notify->pre, pcmk__ar_ordered); order_actions(clone_data->demote_notify->post_done, clone_data->start_notify->pre, pcmk__ar_ordered); order_actions(clone_data->demote_notify->post_done, clone_data->stop_notify->pre, pcmk__ar_ordered); } } } /*! * \internal * \brief Create all notification data and actions for a clone * * \param[in,out] clone Clone to create notifications for */ void pe__create_clone_notifications(pcmk_resource_t *clone) { clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, clone); pe__create_action_notifications(clone, clone_data->start_notify); pe__create_action_notifications(clone, clone_data->stop_notify); pe__create_action_notifications(clone, clone_data->promote_notify); pe__create_action_notifications(clone, clone_data->demote_notify); } /*! * \internal * \brief Free all notification data for a clone * * \param[in,out] clone Clone to free notification data for */ void pe__free_clone_notification_data(pcmk_resource_t *clone) { clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, clone); pe__free_action_notification_data(clone_data->demote_notify); clone_data->demote_notify = NULL; pe__free_action_notification_data(clone_data->stop_notify); clone_data->stop_notify = NULL; pe__free_action_notification_data(clone_data->start_notify); clone_data->start_notify = NULL; pe__free_action_notification_data(clone_data->promote_notify); clone_data->promote_notify = NULL; } /*! * \internal * \brief Create pseudo-actions for clone start/stop notifications * * \param[in,out] clone Clone to create pseudo-actions for * \param[in,out] start Start action for \p clone * \param[in,out] stop Stop action for \p clone * \param[in,out] started Started action for \p clone * \param[in,out] stopped Stopped action for \p clone */ void pe__create_clone_notif_pseudo_ops(pcmk_resource_t *clone, pcmk_action_t *start, pcmk_action_t *started, pcmk_action_t *stop, pcmk_action_t *stopped) { clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, clone); if (clone_data->start_notify == NULL) { clone_data->start_notify = pe__action_notif_pseudo_ops(clone, PCMK_ACTION_START, start, started); } if (clone_data->stop_notify == NULL) { clone_data->stop_notify = pe__action_notif_pseudo_ops(clone, PCMK_ACTION_STOP, stop, stopped); if ((clone_data->start_notify != NULL) && (clone_data->stop_notify != NULL)) { order_actions(clone_data->stop_notify->post_done, clone_data->start_notify->pre, pcmk__ar_ordered); } } } /*! * \internal * \brief Get maximum clone resource instances per node * * \param[in] rsc Clone resource to check * * \return Maximum number of \p rsc instances that can be active on one node */ unsigned int pe__clone_max_per_node(const pcmk_resource_t *rsc) { const clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, rsc); return clone_data->clone_node_max; } diff --git a/lib/pengine/complex.c b/lib/pengine/complex.c index 93ee265989..84b00e32c4 100644 --- a/lib/pengine/complex.c +++ b/lib/pengine/complex.c @@ -1,1258 +1,1258 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include "pe_status_private.h" void populate_hash(xmlNode * nvpair_list, GHashTable * hash, const char **attrs, int attrs_length); static pcmk_node_t *active_node(const pcmk_resource_t *rsc, unsigned int *count_all, unsigned int *count_clean); static pcmk__rsc_methods_t resource_class_functions[] = { { native_unpack, native_find_rsc, native_parameter, native_active, native_resource_state, native_location, native_free, pe__count_common, pe__native_is_filtered, active_node, pe__primitive_max_per_node, }, { group_unpack, native_find_rsc, native_parameter, group_active, group_resource_state, native_location, group_free, pe__count_common, pe__group_is_filtered, active_node, pe__group_max_per_node, }, { clone_unpack, native_find_rsc, native_parameter, clone_active, clone_resource_state, native_location, clone_free, pe__count_common, pe__clone_is_filtered, active_node, pe__clone_max_per_node, }, { pe__unpack_bundle, native_find_rsc, native_parameter, pe__bundle_active, pe__bundle_resource_state, native_location, pe__free_bundle, pe__count_bundle, pe__bundle_is_filtered, pe__bundle_active_node, pe__bundle_max_per_node, } }; static enum pcmk__rsc_variant get_resource_type(const char *name) { if (pcmk__str_eq(name, PCMK_XE_PRIMITIVE, pcmk__str_casei)) { return pcmk__rsc_variant_primitive; } else if (pcmk__str_eq(name, PCMK_XE_GROUP, pcmk__str_casei)) { return pcmk__rsc_variant_group; } else if (pcmk__str_eq(name, PCMK_XE_CLONE, pcmk__str_casei)) { return pcmk__rsc_variant_clone; } else if (pcmk__str_eq(name, PCMK_XE_BUNDLE, pcmk__str_casei)) { return pcmk__rsc_variant_bundle; } return pcmk__rsc_variant_unknown; } /*! * \internal * \brief Insert a meta-attribute if not already present * * \param[in] key Meta-attribute name * \param[in] value Meta-attribute value to add if not already present * \param[in,out] table Meta-attribute hash table to insert into * * \note This is like pcmk__insert_meta() except it won't overwrite existing * values. */ static void dup_attr(gpointer key, gpointer value, gpointer user_data) { GHashTable *table = user_data; CRM_CHECK((key != NULL) && (table != NULL), return); if (pcmk__str_eq((const char *) value, "#default", pcmk__str_casei)) { // @COMPAT Deprecated since 2.1.8 pcmk__config_warn("Support for setting meta-attributes (such as %s) to " "the explicit value '#default' is deprecated and " "will be removed in a future release", (const char *) key); } else if ((value != NULL) && (g_hash_table_lookup(table, key) == NULL)) { pcmk__insert_dup(table, (const char *) key, (const char *) value); } } static void expand_parents_fixed_nvpairs(pcmk_resource_t *rsc, const pcmk_rule_input_t *rule_input, GHashTable *meta_hash, pcmk_scheduler_t *scheduler) { GHashTable *parent_orig_meta = pcmk__strkey_table(free, free); pcmk_resource_t *p = rsc->priv->parent; if (p == NULL) { return ; } /* Search all parent resources, get the fixed value of * PCMK_XE_META_ATTRIBUTES set only in the original xml, and stack it in the * hash table. The fixed value of the lower parent resource takes precedence * and is not overwritten. */ while(p != NULL) { /* A hash table for comparison is generated, including the id-ref. */ pe__unpack_dataset_nvpairs(p->priv->xml, PCMK_XE_META_ATTRIBUTES, rule_input, parent_orig_meta, NULL, scheduler); p = p->priv->parent; } if (parent_orig_meta != NULL) { // This will not overwrite any values already existing for child g_hash_table_foreach(parent_orig_meta, dup_attr, meta_hash); } if (parent_orig_meta != NULL) { g_hash_table_destroy(parent_orig_meta); } return ; } /* * \brief Get fully evaluated resource meta-attributes * * \param[in,out] meta_hash Where to store evaluated meta-attributes * \param[in] rsc Resource to get meta-attributes for * \param[in] node Ignored * \param[in,out] scheduler Scheduler data */ void get_meta_attributes(GHashTable * meta_hash, pcmk_resource_t * rsc, pcmk_node_t *node, pcmk_scheduler_t *scheduler) { const pcmk_rule_input_t rule_input = { .now = scheduler->priv->now, .rsc_standard = crm_element_value(rsc->priv->xml, PCMK_XA_CLASS), .rsc_provider = crm_element_value(rsc->priv->xml, PCMK_XA_PROVIDER), .rsc_agent = crm_element_value(rsc->priv->xml, PCMK_XA_TYPE) }; for (xmlAttrPtr a = pcmk__xe_first_attr(rsc->priv->xml); a != NULL; a = a->next) { if (a->children != NULL) { dup_attr((gpointer) a->name, (gpointer) a->children->content, meta_hash); } } pe__unpack_dataset_nvpairs(rsc->priv->xml, PCMK_XE_META_ATTRIBUTES, &rule_input, meta_hash, NULL, scheduler); /* Set the PCMK_XE_META_ATTRIBUTES explicitly set in the parent resource to * the hash table of the child resource. If it is already explicitly set as * a child, it will not be overwritten. */ if (rsc->priv->parent != NULL) { expand_parents_fixed_nvpairs(rsc, &rule_input, meta_hash, scheduler); } /* check the defaults */ pe__unpack_dataset_nvpairs(scheduler->priv->rsc_defaults, PCMK_XE_META_ATTRIBUTES, &rule_input, meta_hash, NULL, scheduler); /* If there is PCMK_XE_META_ATTRIBUTES that the parent resource has not * explicitly set, set a value that is not set from PCMK_XE_RSC_DEFAULTS * either. The values already set up to this point will not be overwritten. */ if (rsc->priv->parent != NULL) { g_hash_table_foreach(rsc->priv->parent->priv->meta, dup_attr, meta_hash); } } /*! * \brief Get final values of a resource's instance attributes * * \param[in,out] instance_attrs Where to store the instance attributes * \param[in] rsc Resource to get instance attributes for * \param[in] node If not NULL, evaluate rules for this node * \param[in,out] scheduler Scheduler data */ void get_rsc_attributes(GHashTable *instance_attrs, const pcmk_resource_t *rsc, const pcmk_node_t *node, pcmk_scheduler_t *scheduler) { pcmk_rule_input_t rule_input = { .now = NULL, }; CRM_CHECK((instance_attrs != NULL) && (rsc != NULL) && (scheduler != NULL), return); rule_input.now = scheduler->priv->now; if (node != NULL) { rule_input.node_attrs = node->priv->attrs; } // Evaluate resource's own values, then its ancestors' values pe__unpack_dataset_nvpairs(rsc->priv->xml, PCMK_XE_INSTANCE_ATTRIBUTES, &rule_input, instance_attrs, NULL, scheduler); if (rsc->priv->parent != NULL) { get_rsc_attributes(instance_attrs, rsc->priv->parent, node, scheduler); } } static char * template_op_key(xmlNode * op) { const char *name = crm_element_value(op, PCMK_XA_NAME); const char *role = crm_element_value(op, PCMK_XA_ROLE); char *key = NULL; if ((role == NULL) || pcmk__strcase_any_of(role, PCMK_ROLE_STARTED, PCMK_ROLE_UNPROMOTED, PCMK__ROLE_UNPROMOTED_LEGACY, NULL)) { role = PCMK__ROLE_UNKNOWN; } key = crm_strdup_printf("%s-%s", name, role); return key; } static gboolean unpack_template(xmlNode *xml_obj, xmlNode **expanded_xml, pcmk_scheduler_t *scheduler) { xmlNode *cib_resources = NULL; xmlNode *template = NULL; xmlNode *new_xml = NULL; xmlNode *child_xml = NULL; xmlNode *rsc_ops = NULL; xmlNode *template_ops = NULL; const char *template_ref = NULL; const char *id = NULL; if (xml_obj == NULL) { pcmk__config_err("No resource object for template unpacking"); return FALSE; } template_ref = crm_element_value(xml_obj, PCMK_XA_TEMPLATE); if (template_ref == NULL) { return TRUE; } id = pcmk__xe_id(xml_obj); if (id == NULL) { pcmk__config_err("'%s' object must have a id", xml_obj->name); return FALSE; } if (pcmk__str_eq(template_ref, id, pcmk__str_none)) { pcmk__config_err("The resource object '%s' should not reference itself", id); return FALSE; } cib_resources = get_xpath_object("//" PCMK_XE_RESOURCES, scheduler->input, LOG_TRACE); if (cib_resources == NULL) { pcmk__config_err("No resources configured"); return FALSE; } template = pcmk__xe_first_child(cib_resources, PCMK_XE_TEMPLATE, PCMK_XA_ID, template_ref); if (template == NULL) { pcmk__config_err("No template named '%s'", template_ref); return FALSE; } new_xml = pcmk__xml_copy(NULL, template); xmlNodeSetName(new_xml, xml_obj->name); crm_xml_add(new_xml, PCMK_XA_ID, id); crm_xml_add(new_xml, PCMK__META_CLONE, crm_element_value(xml_obj, PCMK__META_CLONE)); template_ops = pcmk__xe_first_child(new_xml, PCMK_XE_OPERATIONS, NULL, NULL); for (child_xml = pcmk__xe_first_child(xml_obj, NULL, NULL, NULL); child_xml != NULL; child_xml = pcmk__xe_next(child_xml, NULL)) { xmlNode *new_child = pcmk__xml_copy(new_xml, child_xml); if (pcmk__xe_is(new_child, PCMK_XE_OPERATIONS)) { rsc_ops = new_child; } } if (template_ops && rsc_ops) { xmlNode *op = NULL; GHashTable *rsc_ops_hash = pcmk__strkey_table(free, NULL); for (op = pcmk__xe_first_child(rsc_ops, NULL, NULL, NULL); op != NULL; op = pcmk__xe_next(op, NULL)) { char *key = template_op_key(op); g_hash_table_insert(rsc_ops_hash, key, op); } for (op = pcmk__xe_first_child(template_ops, NULL, NULL, NULL); op != NULL; op = pcmk__xe_next(op, NULL)) { char *key = template_op_key(op); if (g_hash_table_lookup(rsc_ops_hash, key) == NULL) { pcmk__xml_copy(rsc_ops, op); } free(key); } if (rsc_ops_hash) { g_hash_table_destroy(rsc_ops_hash); } pcmk__xml_free(template_ops); } /*pcmk__xml_free(*expanded_xml); */ *expanded_xml = new_xml; #if 0 /* Disable multi-level templates for now */ if (!unpack_template(new_xml, expanded_xml, scheduler)) { pcmk__xml_free(*expanded_xml); *expanded_xml = NULL; return FALSE; } #endif return TRUE; } static gboolean add_template_rsc(xmlNode *xml_obj, pcmk_scheduler_t *scheduler) { const char *template_ref = NULL; const char *id = NULL; if (xml_obj == NULL) { pcmk__config_err("No resource object for processing resource list " "of template"); return FALSE; } template_ref = crm_element_value(xml_obj, PCMK_XA_TEMPLATE); if (template_ref == NULL) { return TRUE; } id = pcmk__xe_id(xml_obj); if (id == NULL) { pcmk__config_err("'%s' object must have a id", xml_obj->name); return FALSE; } if (pcmk__str_eq(template_ref, id, pcmk__str_none)) { pcmk__config_err("The resource object '%s' should not reference itself", id); return FALSE; } pcmk__add_idref(scheduler->priv->templates, template_ref, id); return TRUE; } /*! * \internal * \brief Check whether a clone or instance being unpacked is globally unique * * \param[in] rsc Clone or clone instance to check * * \return \c true if \p rsc is globally unique according to its * meta-attributes, otherwise \c false */ static bool detect_unique(const pcmk_resource_t *rsc) { const char *value = g_hash_table_lookup(rsc->priv->meta, PCMK_META_GLOBALLY_UNIQUE); if (value == NULL) { // Default to true if clone-node-max > 1 value = g_hash_table_lookup(rsc->priv->meta, PCMK_META_CLONE_NODE_MAX); if (value != NULL) { int node_max = 1; if ((pcmk__scan_min_int(value, &node_max, 0) == pcmk_rc_ok) && (node_max > 1)) { return true; } } return false; } return crm_is_true(value); } static void free_params_table(gpointer data) { g_hash_table_destroy((GHashTable *) data); } /*! * \brief Get a table of resource parameters * * \param[in,out] rsc Resource to query * \param[in] node Node for evaluating rules (NULL for defaults) * \param[in,out] scheduler Scheduler data * * \return Hash table containing resource parameter names and values * (or NULL if \p rsc or \p scheduler is NULL) * \note The returned table will be destroyed when the resource is freed, so * callers should not destroy it. */ GHashTable * pe_rsc_params(pcmk_resource_t *rsc, const pcmk_node_t *node, pcmk_scheduler_t *scheduler) { GHashTable *params_on_node = NULL; /* A NULL node is used to request the resource's default parameters * (not evaluated for node), but we always want something non-NULL * as a hash table key. */ const char *node_name = ""; // Sanity check if ((rsc == NULL) || (scheduler == NULL)) { return NULL; } if ((node != NULL) && (node->priv->name != NULL)) { node_name = node->priv->name; } // Find the parameter table for given node if (rsc->priv->parameter_cache == NULL) { rsc->priv->parameter_cache = pcmk__strikey_table(free, free_params_table); } else { params_on_node = g_hash_table_lookup(rsc->priv->parameter_cache, node_name); } // If none exists yet, create one with parameters evaluated for node if (params_on_node == NULL) { params_on_node = pcmk__strkey_table(free, free); get_rsc_attributes(params_on_node, rsc, node, scheduler); g_hash_table_insert(rsc->priv->parameter_cache, strdup(node_name), params_on_node); } return params_on_node; } /*! * \internal * \brief Unpack a resource's \c PCMK_META_REQUIRES meta-attribute * * \param[in,out] rsc Resource being unpacked * \param[in] value Value of \c PCMK_META_REQUIRES meta-attribute * \param[in] is_default Whether \p value was selected by default */ static void unpack_requires(pcmk_resource_t *rsc, const char *value, bool is_default) { const pcmk_scheduler_t *scheduler = rsc->priv->scheduler; if (pcmk__str_eq(value, PCMK_VALUE_NOTHING, pcmk__str_casei)) { } else if (pcmk__str_eq(value, PCMK_VALUE_QUORUM, pcmk__str_casei)) { pcmk__set_rsc_flags(rsc, pcmk__rsc_needs_quorum); } else if (pcmk__str_eq(value, PCMK_VALUE_FENCING, pcmk__str_casei)) { pcmk__set_rsc_flags(rsc, pcmk__rsc_needs_fencing); if (!pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled)) { pcmk__config_warn("%s requires fencing but fencing is disabled", rsc->id); } } else if (pcmk__str_eq(value, PCMK_VALUE_UNFENCING, pcmk__str_casei)) { if (pcmk_is_set(rsc->flags, pcmk__rsc_fence_device)) { pcmk__config_warn("Resetting \"" PCMK_META_REQUIRES "\" for %s " "to \"" PCMK_VALUE_QUORUM "\" because fencing " "devices cannot require unfencing", rsc->id); unpack_requires(rsc, PCMK_VALUE_QUORUM, true); return; } else if (!pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled)) { pcmk__config_warn("Resetting \"" PCMK_META_REQUIRES "\" for %s " "to \"" PCMK_VALUE_QUORUM "\" because fencing is " "disabled", rsc->id); unpack_requires(rsc, PCMK_VALUE_QUORUM, true); return; } else { pcmk__set_rsc_flags(rsc, pcmk__rsc_needs_fencing |pcmk__rsc_needs_unfencing); } } else { const char *orig_value = value; if (pcmk_is_set(rsc->flags, pcmk__rsc_fence_device)) { value = PCMK_VALUE_QUORUM; } else if (pcmk__is_primitive(rsc) && xml_contains_remote_node(rsc->priv->xml)) { value = PCMK_VALUE_QUORUM; } else if (pcmk_is_set(scheduler->flags, pcmk__sched_enable_unfencing)) { value = PCMK_VALUE_UNFENCING; } else if (pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled)) { value = PCMK_VALUE_FENCING; } else if (scheduler->no_quorum_policy == pcmk_no_quorum_ignore) { value = PCMK_VALUE_NOTHING; } else { value = PCMK_VALUE_QUORUM; } if (orig_value != NULL) { pcmk__config_err("Resetting '" PCMK_META_REQUIRES "' for %s " "to '%s' because '%s' is not valid", rsc->id, value, orig_value); } unpack_requires(rsc, value, true); return; } pcmk__rsc_trace(rsc, "\tRequired to start: %s%s", value, (is_default? " (default)" : "")); } /*! * \internal * \brief Parse resource priority from meta-attribute * * \param[in,out] rsc Resource being unpacked */ static void unpack_priority(pcmk_resource_t *rsc) { const char *value = g_hash_table_lookup(rsc->priv->meta, PCMK_META_PRIORITY); int rc = pcmk_parse_score(value, &(rsc->priv->priority), 0); if (rc != pcmk_rc_ok) { pcmk__config_warn("Using default (0) for resource %s " PCMK_META_PRIORITY " because '%s' is not a valid value: %s", rsc->id, value, pcmk_rc_str(rc)); } } /*! * \internal * \brief Parse resource stickiness from meta-attribute * * \param[in,out] rsc Resource being unpacked */ static void unpack_stickiness(pcmk_resource_t *rsc) { const char *value = g_hash_table_lookup(rsc->priv->meta, PCMK_META_RESOURCE_STICKINESS); if (pcmk__str_eq(value, PCMK_VALUE_DEFAULT, pcmk__str_casei)) { // @COMPAT Deprecated since 2.1.8 pcmk__config_warn("Support for setting " PCMK_META_RESOURCE_STICKINESS " to the explicit value '" PCMK_VALUE_DEFAULT "' is deprecated and will be removed in a " "future release (just leave it unset)"); } else { int rc = pcmk_parse_score(value, &(rsc->priv->stickiness), 0); if (rc != pcmk_rc_ok) { pcmk__config_warn("Using default (0) for resource %s " PCMK_META_RESOURCE_STICKINESS " because '%s' is not a valid value: %s", rsc->id, value, pcmk_rc_str(rc)); } } } /*! * \internal * \brief Parse resource migration threshold from meta-attribute * * \param[in,out] rsc Resource being unpacked */ static void unpack_migration_threshold(pcmk_resource_t *rsc) { const char *value = g_hash_table_lookup(rsc->priv->meta, PCMK_META_MIGRATION_THRESHOLD); if (pcmk__str_eq(value, PCMK_VALUE_DEFAULT, pcmk__str_casei)) { // @COMPAT Deprecated since 2.1.8 pcmk__config_warn("Support for setting " PCMK_META_MIGRATION_THRESHOLD " to the explicit value '" PCMK_VALUE_DEFAULT "' is deprecated and will be removed in a " "future release (just leave it unset)"); rsc->priv->ban_after_failures = PCMK_SCORE_INFINITY; } else { int rc = pcmk_parse_score(value, &(rsc->priv->ban_after_failures), PCMK_SCORE_INFINITY); if ((rc != pcmk_rc_ok) || (rsc->priv->ban_after_failures < 0)) { pcmk__config_warn("Using default (" PCMK_VALUE_INFINITY ") for resource %s meta-attribute " PCMK_META_MIGRATION_THRESHOLD " because '%s' is not a valid value: %s", rsc->id, value, pcmk_rc_str(rc)); rsc->priv->ban_after_failures = PCMK_SCORE_INFINITY; } } } /*! * \internal * \brief Unpack configuration XML for a given resource * * Unpack the XML object containing a resource's configuration into a new * \c pcmk_resource_t object. * * \param[in] xml_obj XML node containing the resource's configuration * \param[out] rsc Where to store the unpacked resource information * \param[in] parent Resource's parent, if any * \param[in,out] scheduler Scheduler data * * \return Standard Pacemaker return code * \note If pcmk_rc_ok is returned, \p *rsc is guaranteed to be non-NULL, and * the caller is responsible for freeing it using its variant-specific * free() method. Otherwise, \p *rsc is guaranteed to be NULL. */ int pe__unpack_resource(xmlNode *xml_obj, pcmk_resource_t **rsc, pcmk_resource_t *parent, pcmk_scheduler_t *scheduler) { xmlNode *expanded_xml = NULL; xmlNode *ops = NULL; const char *value = NULL; const char *id = NULL; bool guest_node = false; bool remote_node = false; pcmk__resource_private_t *rsc_private = NULL; pcmk_rule_input_t rule_input = { .now = NULL, }; CRM_CHECK(rsc != NULL, return EINVAL); CRM_CHECK((xml_obj != NULL) && (scheduler != NULL), *rsc = NULL; return EINVAL); rule_input.now = scheduler->priv->now; crm_log_xml_trace(xml_obj, "[raw XML]"); id = crm_element_value(xml_obj, PCMK_XA_ID); if (id == NULL) { pcmk__config_err("Ignoring <%s> configuration without " PCMK_XA_ID, xml_obj->name); return pcmk_rc_unpack_error; } if (unpack_template(xml_obj, &expanded_xml, scheduler) == FALSE) { return pcmk_rc_unpack_error; } *rsc = calloc(1, sizeof(pcmk_resource_t)); if (*rsc == NULL) { pcmk__sched_err(scheduler, "Unable to allocate memory for resource '%s'", id); return ENOMEM; } (*rsc)->priv = calloc(1, sizeof(pcmk__resource_private_t)); if ((*rsc)->priv == NULL) { pcmk__sched_err(scheduler, "Unable to allocate memory for resource '%s'", id); free(*rsc); return ENOMEM; } rsc_private = (*rsc)->priv; rsc_private->scheduler = scheduler; if (expanded_xml) { crm_log_xml_trace(expanded_xml, "[expanded XML]"); rsc_private->xml = expanded_xml; rsc_private->orig_xml = xml_obj; } else { rsc_private->xml = xml_obj; rsc_private->orig_xml = NULL; } /* Do not use xml_obj from here on, use (*rsc)->xml in case templates are involved */ rsc_private->parent = parent; ops = pcmk__xe_first_child(rsc_private->xml, PCMK_XE_OPERATIONS, NULL, NULL); rsc_private->ops_xml = pcmk__xe_resolve_idref(ops, scheduler->input); rsc_private->variant = get_resource_type((const char *) rsc_private->xml->name); if (rsc_private->variant == pcmk__rsc_variant_unknown) { pcmk__config_err("Ignoring resource '%s' of unknown type '%s'", id, rsc_private->xml->name); common_free(*rsc); *rsc = NULL; return pcmk_rc_unpack_error; } rsc_private->meta = pcmk__strkey_table(free, free); rsc_private->utilization = pcmk__strkey_table(free, free); rsc_private->probed_nodes = pcmk__strkey_table(NULL, pcmk__free_node_copy); rsc_private->allowed_nodes = pcmk__strkey_table(NULL, pcmk__free_node_copy); value = crm_element_value(rsc_private->xml, PCMK__META_CLONE); if (value) { (*rsc)->id = crm_strdup_printf("%s:%s", id, value); pcmk__insert_meta(rsc_private, PCMK__META_CLONE, value); } else { (*rsc)->id = strdup(id); } rsc_private->fns = &resource_class_functions[rsc_private->variant]; get_meta_attributes(rsc_private->meta, *rsc, NULL, scheduler); (*rsc)->flags = 0; pcmk__set_rsc_flags(*rsc, pcmk__rsc_unassigned); if (!pcmk_is_set(scheduler->flags, pcmk__sched_in_maintenance)) { pcmk__set_rsc_flags(*rsc, pcmk__rsc_managed); } rsc_private->orig_role = pcmk_role_stopped; rsc_private->next_role = pcmk_role_unknown; unpack_priority(*rsc); value = g_hash_table_lookup(rsc_private->meta, PCMK_META_CRITICAL); if ((value == NULL) || crm_is_true(value)) { pcmk__set_rsc_flags(*rsc, pcmk__rsc_critical); } value = g_hash_table_lookup(rsc_private->meta, PCMK_META_NOTIFY); if (crm_is_true(value)) { pcmk__set_rsc_flags(*rsc, pcmk__rsc_notify); } if (xml_contains_remote_node(rsc_private->xml)) { pcmk__set_rsc_flags(*rsc, pcmk__rsc_is_remote_connection); if (g_hash_table_lookup(rsc_private->meta, PCMK__META_CONTAINER)) { guest_node = true; } else { remote_node = true; } } value = g_hash_table_lookup(rsc_private->meta, PCMK_META_ALLOW_MIGRATE); if (crm_is_true(value)) { pcmk__set_rsc_flags(*rsc, pcmk__rsc_migratable); } else if ((value == NULL) && remote_node) { /* By default, we want remote nodes to be able * to float around the cluster without having to stop all the * resources within the remote-node before moving. Allowing * migration support enables this feature. If this ever causes * problems, migration support can be explicitly turned off with * PCMK_META_ALLOW_MIGRATE=false. */ pcmk__set_rsc_flags(*rsc, pcmk__rsc_migratable); } value = g_hash_table_lookup(rsc_private->meta, PCMK_META_IS_MANAGED); if (value != NULL) { if (pcmk__str_eq(PCMK_VALUE_DEFAULT, value, pcmk__str_casei)) { // @COMPAT Deprecated since 2.1.8 pcmk__config_warn("Support for setting " PCMK_META_IS_MANAGED " to the explicit value '" PCMK_VALUE_DEFAULT "' is deprecated and will be removed in a " "future release (just leave it unset)"); } else if (crm_is_true(value)) { pcmk__set_rsc_flags(*rsc, pcmk__rsc_managed); } else { pcmk__clear_rsc_flags(*rsc, pcmk__rsc_managed); } } value = g_hash_table_lookup(rsc_private->meta, PCMK_META_MAINTENANCE); if (crm_is_true(value)) { pcmk__clear_rsc_flags(*rsc, pcmk__rsc_managed); pcmk__set_rsc_flags(*rsc, pcmk__rsc_maintenance); } if (pcmk_is_set(scheduler->flags, pcmk__sched_in_maintenance)) { pcmk__clear_rsc_flags(*rsc, pcmk__rsc_managed); pcmk__set_rsc_flags(*rsc, pcmk__rsc_maintenance); } if (pcmk__is_clone(pe__const_top_resource(*rsc, false))) { if (detect_unique(*rsc)) { pcmk__set_rsc_flags(*rsc, pcmk__rsc_unique); } if (crm_is_true(g_hash_table_lookup((*rsc)->priv->meta, PCMK_META_PROMOTABLE))) { pcmk__set_rsc_flags(*rsc, pcmk__rsc_promotable); } } else { pcmk__set_rsc_flags(*rsc, pcmk__rsc_unique); } value = g_hash_table_lookup(rsc_private->meta, PCMK_META_MULTIPLE_ACTIVE); if (pcmk__str_eq(value, PCMK_VALUE_STOP_ONLY, pcmk__str_casei)) { rsc_private->multiply_active_policy = pcmk__multiply_active_stop; pcmk__rsc_trace(*rsc, "%s multiple running resource recovery: stop only", (*rsc)->id); } else if (pcmk__str_eq(value, PCMK_VALUE_BLOCK, pcmk__str_casei)) { rsc_private->multiply_active_policy = pcmk__multiply_active_block; pcmk__rsc_trace(*rsc, "%s multiple running resource recovery: block", (*rsc)->id); } else if (pcmk__str_eq(value, PCMK_VALUE_STOP_UNEXPECTED, pcmk__str_casei)) { rsc_private->multiply_active_policy = pcmk__multiply_active_unexpected; pcmk__rsc_trace(*rsc, "%s multiple running resource recovery: " "stop unexpected instances", (*rsc)->id); } else { // PCMK_VALUE_STOP_START if (!pcmk__str_eq(value, PCMK_VALUE_STOP_START, pcmk__str_casei|pcmk__str_null_matches)) { pcmk__config_warn("%s is not a valid value for " PCMK_META_MULTIPLE_ACTIVE ", using default of " "\"" PCMK_VALUE_STOP_START "\"", value); } rsc_private->multiply_active_policy = pcmk__multiply_active_restart; pcmk__rsc_trace(*rsc, "%s multiple running resource recovery: stop/start", (*rsc)->id); } unpack_stickiness(*rsc); unpack_migration_threshold(*rsc); if (pcmk__str_eq(crm_element_value(rsc_private->xml, PCMK_XA_CLASS), PCMK_RESOURCE_CLASS_STONITH, pcmk__str_casei)) { pcmk__set_scheduler_flags(scheduler, pcmk__sched_have_fencing); pcmk__set_rsc_flags(*rsc, pcmk__rsc_fence_device); } value = g_hash_table_lookup(rsc_private->meta, PCMK_META_REQUIRES); unpack_requires(*rsc, value, false); value = g_hash_table_lookup(rsc_private->meta, PCMK_META_FAILURE_TIMEOUT); if (value != NULL) { pcmk_parse_interval_spec(value, &(rsc_private->failure_expiration_ms)); } if (remote_node) { GHashTable *params = pe_rsc_params(*rsc, NULL, scheduler); /* Grabbing the value now means that any rules based on node attributes * will evaluate to false, so such rules should not be used with * PCMK_REMOTE_RA_RECONNECT_INTERVAL. * * @TODO Evaluate per node before using */ value = g_hash_table_lookup(params, PCMK_REMOTE_RA_RECONNECT_INTERVAL); if (value) { /* reconnect delay works by setting failure_timeout and preventing the * connection from starting until the failure is cleared. */ pcmk_parse_interval_spec(value, &(rsc_private->remote_reconnect_ms)); /* We want to override any default failure_timeout in use when remote * PCMK_REMOTE_RA_RECONNECT_INTERVAL is in use. */ rsc_private->failure_expiration_ms = rsc_private->remote_reconnect_ms; } } get_target_role(*rsc, &(rsc_private->next_role)); pcmk__rsc_trace(*rsc, "%s desired next state: %s", (*rsc)->id, (rsc_private->next_role == pcmk_role_unknown)? "default" : pcmk_role_text(rsc_private->next_role)); if (rsc_private->fns->unpack(*rsc, scheduler) == FALSE) { - rsc_private->fns->free(*rsc); + pcmk__free_resource(*rsc); *rsc = NULL; return pcmk_rc_unpack_error; } if (pcmk_is_set(scheduler->flags, pcmk__sched_symmetric_cluster)) { // This tag must stay exactly the same because it is tested elsewhere resource_location(*rsc, NULL, 0, "symmetric_default", scheduler); } else if (guest_node) { /* remote resources tied to a container resource must always be allowed * to opt-in to the cluster. Whether the connection resource is actually * allowed to be placed on a node is dependent on the container resource */ resource_location(*rsc, NULL, 0, "remote_connection_default", scheduler); } pcmk__rsc_trace(*rsc, "%s action notification: %s", (*rsc)->id, pcmk_is_set((*rsc)->flags, pcmk__rsc_notify)? "required" : "not required"); pe__unpack_dataset_nvpairs(rsc_private->xml, PCMK_XE_UTILIZATION, &rule_input, rsc_private->utilization, NULL, scheduler); if (expanded_xml) { if (add_template_rsc(xml_obj, scheduler) == FALSE) { - rsc_private->fns->free(*rsc); + pcmk__free_resource(*rsc); *rsc = NULL; return pcmk_rc_unpack_error; } } return pcmk_rc_ok; } gboolean is_parent(pcmk_resource_t *child, pcmk_resource_t *rsc) { pcmk_resource_t *parent = child; if (parent == NULL || rsc == NULL) { return FALSE; } while (parent->priv->parent != NULL) { if (parent->priv->parent == rsc) { return TRUE; } parent = parent->priv->parent; } return FALSE; } pcmk_resource_t * uber_parent(pcmk_resource_t *rsc) { pcmk_resource_t *parent = rsc; if (parent == NULL) { return NULL; } while ((parent->priv->parent != NULL) && !pcmk__is_bundle(parent->priv->parent)) { parent = parent->priv->parent; } return parent; } /*! * \internal * \brief Get the topmost parent of a resource as a const pointer * * \param[in] rsc Resource to check * \param[in] include_bundle If true, go all the way to bundle * * \return \p NULL if \p rsc is NULL, \p rsc if \p rsc has no parent, * the bundle if \p rsc is bundled and \p include_bundle is true, * otherwise the topmost parent of \p rsc up to a clone */ const pcmk_resource_t * pe__const_top_resource(const pcmk_resource_t *rsc, bool include_bundle) { const pcmk_resource_t *parent = rsc; if (parent == NULL) { return NULL; } while (parent->priv->parent != NULL) { if (!include_bundle && pcmk__is_bundle(parent->priv->parent)) { break; } parent = parent->priv->parent; } return parent; } void common_free(pcmk_resource_t * rsc) { if (rsc == NULL) { return; } pcmk__rsc_trace(rsc, "Freeing %s", rsc->id); if (rsc->priv->parameter_cache != NULL) { g_hash_table_destroy(rsc->priv->parameter_cache); } if ((rsc->priv->parent == NULL) && pcmk_is_set(rsc->flags, pcmk__rsc_removed)) { pcmk__xml_free(rsc->priv->xml); rsc->priv->xml = NULL; pcmk__xml_free(rsc->priv->orig_xml); rsc->priv->orig_xml = NULL; } else if (rsc->priv->orig_xml != NULL) { // rsc->private->xml was expanded from a template pcmk__xml_free(rsc->priv->xml); rsc->priv->xml = NULL; } free(rsc->id); free(rsc->priv->variant_opaque); free(rsc->priv->history_id); free(rsc->priv->pending_action); pcmk__free_node_copy(rsc->priv->assigned_node); g_list_free(rsc->priv->actions); g_list_free(rsc->priv->active_nodes); g_list_free(rsc->priv->launched); g_list_free(rsc->priv->dangling_migration_sources); g_list_free(rsc->priv->with_this_colocations); g_list_free(rsc->priv->this_with_colocations); g_list_free(rsc->priv->location_constraints); g_list_free(rsc->priv->ticket_constraints); if (rsc->priv->meta != NULL) { g_hash_table_destroy(rsc->priv->meta); } if (rsc->priv->utilization != NULL) { g_hash_table_destroy(rsc->priv->utilization); } if (rsc->priv->probed_nodes != NULL) { g_hash_table_destroy(rsc->priv->probed_nodes); } if (rsc->priv->allowed_nodes != NULL) { g_hash_table_destroy(rsc->priv->allowed_nodes); } free(rsc->priv); free(rsc); } /*! * \internal * \brief Count a node and update most preferred to it as appropriate * * \param[in] rsc An active resource * \param[in] node A node that \p rsc is active on * \param[in,out] active This will be set to \p node if \p node is more * preferred than the current value * \param[in,out] count_all If not NULL, this will be incremented * \param[in,out] count_clean If not NULL, this will be incremented if \p node * is online and clean * * \return true if the count should continue, or false if sufficiently known */ bool pe__count_active_node(const pcmk_resource_t *rsc, pcmk_node_t *node, pcmk_node_t **active, unsigned int *count_all, unsigned int *count_clean) { bool keep_looking = false; bool is_happy = false; CRM_CHECK((rsc != NULL) && (node != NULL) && (active != NULL), return false); is_happy = node->details->online && !node->details->unclean; if (count_all != NULL) { ++*count_all; } if ((count_clean != NULL) && is_happy) { ++*count_clean; } if ((count_all != NULL) || (count_clean != NULL)) { keep_looking = true; // We're counting, so go through entire list } if (rsc->priv->partial_migration_source != NULL) { if (pcmk__same_node(node, rsc->priv->partial_migration_source)) { *active = node; // This is the migration source } else { keep_looking = true; } } else if (!pcmk_is_set(rsc->flags, pcmk__rsc_needs_fencing)) { if (is_happy && ((*active == NULL) || !(*active)->details->online || (*active)->details->unclean)) { *active = node; // This is the first clean node } else { keep_looking = true; } } if (*active == NULL) { *active = node; // This is the first node checked } return keep_looking; } // Shared implementation of pcmk__rsc_methods_t:active_node() static pcmk_node_t * active_node(const pcmk_resource_t *rsc, unsigned int *count_all, unsigned int *count_clean) { pcmk_node_t *active = NULL; if (count_all != NULL) { *count_all = 0; } if (count_clean != NULL) { *count_clean = 0; } if (rsc == NULL) { return NULL; } for (GList *iter = rsc->priv->active_nodes; iter != NULL; iter = iter->next) { if (!pe__count_active_node(rsc, (pcmk_node_t *) iter->data, &active, count_all, count_clean)) { break; // Don't waste time iterating if we don't have to } } return active; } /*! * \brief * \internal Find and count active nodes according to \c PCMK_META_REQUIRES * * \param[in] rsc Resource to check * \param[out] count If not NULL, will be set to count of active nodes * * \return An active node (or NULL if resource is not active anywhere) * * \note This is a convenience wrapper for active_node() where the count of all * active nodes or only clean active nodes is desired according to the * \c PCMK_META_REQUIRES meta-attribute. */ pcmk_node_t * pe__find_active_requires(const pcmk_resource_t *rsc, unsigned int *count) { if (rsc == NULL) { if (count != NULL) { *count = 0; } return NULL; } if (pcmk_is_set(rsc->flags, pcmk__rsc_needs_fencing)) { return rsc->priv->fns->active_node(rsc, count, NULL); } else { return rsc->priv->fns->active_node(rsc, NULL, count); } } void pe__count_common(pcmk_resource_t *rsc) { if (rsc->priv->children != NULL) { for (GList *item = rsc->priv->children; item != NULL; item = item->next) { pcmk_resource_t *child = item->data; child->priv->fns->count(item->data); } } else if (!pcmk_is_set(rsc->flags, pcmk__rsc_removed) || (rsc->priv->orig_role > pcmk_role_stopped)) { rsc->priv->scheduler->priv->ninstances++; if (pe__resource_is_disabled(rsc)) { rsc->priv->scheduler->priv->disabled_resources++; } if (pcmk_is_set(rsc->flags, pcmk__rsc_blocked)) { rsc->priv->scheduler->priv->blocked_resources++; } } } /*! * \internal * \brief Update a resource's next role * * \param[in,out] rsc Resource to be updated * \param[in] role Resource's new next role * \param[in] why Human-friendly reason why role is changing (for logs) */ void pe__set_next_role(pcmk_resource_t *rsc, enum rsc_role_e role, const char *why) { pcmk__assert((rsc != NULL) && (why != NULL)); if (rsc->priv->next_role != role) { pcmk__rsc_trace(rsc, "Resetting next role for %s from %s to %s (%s)", rsc->id, pcmk_role_text(rsc->priv->next_role), pcmk_role_text(role), why); rsc->priv->next_role = role; } } diff --git a/lib/pengine/failcounts.c b/lib/pengine/failcounts.c index 1288166c2e..7d3b0de6a6 100644 --- a/lib/pengine/failcounts.c +++ b/lib/pengine/failcounts.c @@ -1,479 +1,479 @@ /* * Copyright 2008-2024 the Pacemaker project contributors * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include static gboolean is_matched_failure(const char *rsc_id, const xmlNode *conf_op_xml, const xmlNode *lrm_op_xml) { gboolean matched = FALSE; const char *conf_op_name = NULL; const char *lrm_op_task = NULL; const char *conf_op_interval_spec = NULL; guint conf_op_interval_ms = 0; guint lrm_op_interval_ms = 0; const char *lrm_op_id = NULL; char *last_failure_key = NULL; if (rsc_id == NULL || conf_op_xml == NULL || lrm_op_xml == NULL) { return FALSE; } // Get name and interval from configured op conf_op_name = crm_element_value(conf_op_xml, PCMK_XA_NAME); conf_op_interval_spec = crm_element_value(conf_op_xml, PCMK_META_INTERVAL); pcmk_parse_interval_spec(conf_op_interval_spec, &conf_op_interval_ms); // Get name and interval from op history entry lrm_op_task = crm_element_value(lrm_op_xml, PCMK_XA_OPERATION); crm_element_value_ms(lrm_op_xml, PCMK_META_INTERVAL, &lrm_op_interval_ms); if ((conf_op_interval_ms != lrm_op_interval_ms) || !pcmk__str_eq(conf_op_name, lrm_op_task, pcmk__str_casei)) { return FALSE; } lrm_op_id = pcmk__xe_id(lrm_op_xml); last_failure_key = pcmk__op_key(rsc_id, "last_failure", 0); if (pcmk__str_eq(last_failure_key, lrm_op_id, pcmk__str_casei)) { matched = TRUE; } else { char *expected_op_key = pcmk__op_key(rsc_id, conf_op_name, conf_op_interval_ms); if (pcmk__str_eq(expected_op_key, lrm_op_id, pcmk__str_casei)) { int rc = 0; int target_rc = pe__target_rc_from_xml(lrm_op_xml); crm_element_value_int(lrm_op_xml, PCMK__XA_RC_CODE, &rc); if (rc != target_rc) { matched = TRUE; } } free(expected_op_key); } free(last_failure_key); return matched; } static gboolean block_failure(const pcmk_node_t *node, pcmk_resource_t *rsc, const xmlNode *xml_op) { char *xml_name = clone_strip(rsc->id); /* @TODO This xpath search occurs after template expansion, but it is unable * to properly detect on-fail in id-ref, operation meta-attributes, or * op_defaults, or evaluate rules. * * Also, PCMK_META_ON_FAIL defaults to PCMK_VALUE_BLOCK (in * unpack_operation()) for stop actions when stonith is disabled. * * Ideally, we'd unpack the operation before this point, and pass in a * meta-attributes table that takes all that into consideration. */ char *xpath = crm_strdup_printf("//" PCMK_XE_PRIMITIVE "[@" PCMK_XA_ID "='%s']" "//" PCMK_XE_OP "[@" PCMK_META_ON_FAIL "='" PCMK_VALUE_BLOCK "']", xml_name); xmlXPathObject *xpathObj = xpath_search(rsc->priv->xml, xpath); gboolean should_block = FALSE; free(xpath); if (xpathObj) { int max = numXpathResults(xpathObj); int lpc = 0; for (lpc = 0; lpc < max; lpc++) { xmlNode *pref = getXpathResult(xpathObj, lpc); if (xml_op) { should_block = is_matched_failure(xml_name, pref, xml_op); if (should_block) { break; } } else { const char *conf_op_name = NULL; const char *conf_op_interval_spec = NULL; guint conf_op_interval_ms = 0; char *lrm_op_xpath = NULL; xmlXPathObject *lrm_op_xpathObj = NULL; // Get name and interval from configured op conf_op_name = crm_element_value(pref, PCMK_XA_NAME); conf_op_interval_spec = crm_element_value(pref, PCMK_META_INTERVAL); pcmk_parse_interval_spec(conf_op_interval_spec, &conf_op_interval_ms); #define XPATH_FMT "//" PCMK__XE_NODE_STATE "[@" PCMK_XA_UNAME "='%s']" \ "//" PCMK__XE_LRM_RESOURCE "[@" PCMK_XA_ID "='%s']" \ "/" PCMK__XE_LRM_RSC_OP "[@" PCMK_XA_OPERATION "='%s']" \ "[@" PCMK_META_INTERVAL "='%u']" lrm_op_xpath = crm_strdup_printf(XPATH_FMT, node->priv->name, xml_name, conf_op_name, conf_op_interval_ms); lrm_op_xpathObj = xpath_search(rsc->priv->scheduler->input, lrm_op_xpath); free(lrm_op_xpath); if (lrm_op_xpathObj) { int max2 = numXpathResults(lrm_op_xpathObj); int lpc2 = 0; for (lpc2 = 0; lpc2 < max2; lpc2++) { xmlNode *lrm_op_xml = getXpathResult(lrm_op_xpathObj, lpc2); should_block = is_matched_failure(xml_name, pref, lrm_op_xml); if (should_block) { break; } } } freeXpathObject(lrm_op_xpathObj); if (should_block) { break; } } } } free(xml_name); freeXpathObject(xpathObj); return should_block; } /*! * \internal * \brief Get resource name as used in failure-related node attributes * * \param[in] rsc Resource to check * * \return Newly allocated string containing resource's fail name * \note The caller is responsible for freeing the result. */ static inline char * rsc_fail_name(const pcmk_resource_t *rsc) { const char *name = pcmk__s(rsc->priv->history_id, rsc->id); return pcmk_is_set(rsc->flags, pcmk__rsc_unique)? strdup(name) : clone_strip(name); } /*! * \internal * \brief Compile regular expression to match a failure-related node attribute * * \param[in] prefix Attribute prefix to match * \param[in] rsc_name Resource name to match as used in failure attributes * \param[in] is_unique Whether the resource is a globally unique clone * \param[out] re Where to store resulting regular expression * * \return Standard Pacemaker return code * \note Fail attributes are named like PREFIX-RESOURCE#OP_INTERVAL. * The caller is responsible for freeing re with regfree(). */ static int generate_fail_regex(const char *prefix, const char *rsc_name, bool is_unique, regex_t *re) { char *pattern = NULL; const char *op_pattern = "#.+_[0-9]+"; /* Ignore instance numbers for anything other than globally unique clones. * Anonymous clone fail counts could contain an instance number if the * clone was initially unique, failed, then was converted to anonymous. */ const char *instance_pattern = (is_unique? "" : "(:[0-9]+)?"); pattern = crm_strdup_printf("^%s-%s%s%s$", prefix, rsc_name, instance_pattern, op_pattern); if (regcomp(re, pattern, REG_EXTENDED|REG_NOSUB) != 0) { free(pattern); return EINVAL; } free(pattern); return pcmk_rc_ok; } /*! * \internal * \brief Compile regular expressions to match failure-related node attributes * * \param[in] rsc Resource being checked for failures * \param[out] failcount_re Storage for regular expression for fail count * \param[out] lastfailure_re Storage for regular expression for last failure * * \return Standard Pacemaker return code * \note On success, the caller is responsible for freeing the expressions with * regfree(). */ static int generate_fail_regexes(const pcmk_resource_t *rsc, regex_t *failcount_re, regex_t *lastfailure_re) { int rc = pcmk_rc_ok; char *rsc_name = rsc_fail_name(rsc); if (generate_fail_regex(PCMK__FAIL_COUNT_PREFIX, rsc_name, pcmk_is_set(rsc->flags, pcmk__rsc_unique), failcount_re) != pcmk_rc_ok) { rc = EINVAL; } else if (generate_fail_regex(PCMK__LAST_FAILURE_PREFIX, rsc_name, pcmk_is_set(rsc->flags, pcmk__rsc_unique), lastfailure_re) != pcmk_rc_ok) { rc = EINVAL; regfree(failcount_re); } free(rsc_name); return rc; } // Data for fail-count-related iterators struct failcount_data { const pcmk_node_t *node;// Node to check for fail count pcmk_resource_t *rsc; // Resource to check for fail count uint32_t flags; // Fail count flags const xmlNode *xml_op; // History entry for expiration purposes (or NULL) regex_t failcount_re; // Fail count regular expression to match regex_t lastfailure_re; // Last failure regular expression to match int failcount; // Fail count so far time_t last_failure; // Time of most recent failure so far }; /*! * \internal * \brief Update fail count and last failure appropriately for a node attribute * * \param[in] key Node attribute name * \param[in] value Node attribute value * \param[in] user_data Fail count data to update */ static void update_failcount_for_attr(gpointer key, gpointer value, gpointer user_data) { struct failcount_data *fc_data = user_data; // If this is a matching fail count attribute, update fail count if (regexec(&(fc_data->failcount_re), (const char *) key, 0, NULL, 0) == 0) { int score = 0; int rc = pcmk_parse_score(value, &score, 0); if (rc != pcmk_rc_ok) { crm_warn("Ignoring %s for %s " "because '%s' is not a valid fail count: %s", (const char *) key, pcmk__node_name(fc_data->node), value, pcmk_rc_str(rc)); return; } fc_data->failcount = pcmk__add_scores(fc_data->failcount, score); pcmk__rsc_trace(fc_data->rsc, "Added %s (%s) to %s fail count (now %s)", (const char *) key, (const char *) value, fc_data->rsc->id, pcmk_readable_score(fc_data->failcount)); return; } // If this is a matching last failure attribute, update last failure if (regexec(&(fc_data->lastfailure_re), (const char *) key, 0, NULL, 0) == 0) { long long last_ll; int rc = pcmk__scan_ll(value, &last_ll, 0LL); if (rc != pcmk_rc_ok) { crm_info("Ignoring invalid value '%s' for %s: %s", (const char *) value, (const char *) key, pcmk_rc_str(rc)); return; } fc_data->last_failure = (time_t) QB_MAX(fc_data->last_failure, last_ll); } } /*! * \internal * \brief Update fail count and last failure appropriately for launched resource * * \param[in] data Launched resource * \param[in] user_data Fail count data to update */ static void update_launched_failcount(gpointer data, gpointer user_data) { pcmk_resource_t *launched = data; struct failcount_data *fc_data = user_data; time_t launched_last_failure = 0; fc_data->failcount += pe_get_failcount(fc_data->node, launched, &launched_last_failure, fc_data->flags, fc_data->xml_op); fc_data->last_failure = QB_MAX(fc_data->last_failure, launched_last_failure); } #define readable_expiration(rsc) \ pcmk__readable_interval((rsc)->priv->failure_expiration_ms) /*! * \internal * \brief Get a resource's fail count on a node * * \param[in] node Node to check * \param[in,out] rsc Resource to check * \param[out] last_failure If not NULL, where to set time of most recent * failure of \p rsc on \p node * \param[in] flags Group of enum pcmk__fc_flags * \param[in] xml_op If not NULL, consider only the action in this * history entry when determining whether on-fail * is configured as "blocked", otherwise consider * all actions configured for \p rsc * * \return Fail count for \p rsc on \p node according to \p flags */ int pe_get_failcount(const pcmk_node_t *node, pcmk_resource_t *rsc, time_t *last_failure, uint32_t flags, const xmlNode *xml_op) { struct failcount_data fc_data = { .node = node, .rsc = rsc, .flags = flags, .xml_op = xml_op, .failcount = 0, .last_failure = (time_t) 0, }; // Calculate resource failcount as sum of all matching operation failcounts CRM_CHECK(generate_fail_regexes(rsc, &fc_data.failcount_re, &fc_data.lastfailure_re) == pcmk_rc_ok, return 0); g_hash_table_foreach(node->priv->attrs, update_failcount_for_attr, &fc_data); regfree(&(fc_data.failcount_re)); regfree(&(fc_data.lastfailure_re)); // If failure blocks the resource, disregard any failure timeout if ((fc_data.failcount > 0) && (rsc->priv->failure_expiration_ms > 0) && block_failure(node, rsc, xml_op)) { pcmk__config_warn("Ignoring failure timeout (%s) for %s " "because it conflicts with " PCMK_META_ON_FAIL "=" PCMK_VALUE_BLOCK, readable_expiration(rsc), rsc->id); rsc->priv->failure_expiration_ms = 0; } // If all failures have expired, ignore fail count if (pcmk_is_set(flags, pcmk__fc_effective) && (fc_data.failcount > 0) && (fc_data.last_failure > 0) && (rsc->priv->failure_expiration_ms > 0)) { - time_t now = get_effective_time(rsc->priv->scheduler); + time_t now = pcmk__scheduler_epoch_time(rsc->priv->scheduler); const guint expiration = pcmk__timeout_ms2s(rsc->priv->failure_expiration_ms); if (now > (fc_data.last_failure + expiration)) { pcmk__rsc_debug(rsc, "Failcount for %s on %s expired after %s", rsc->id, pcmk__node_name(node), readable_expiration(rsc)); fc_data.failcount = 0; } } /* Add the fail count of any launched resources, except that we never want * the fail counts of a bundle container's launched resources to count * towards the container's fail count. * * Most importantly, a Pacemaker Remote connection to a bundle container * is launched by the container, but can reside on a different node than the * container itself. Counting its fail count on its node towards the * container's fail count on that node could lead to attempting to stop the * container on the wrong node. */ if (pcmk_is_set(flags, pcmk__fc_launched) && (rsc->priv->launched != NULL) && !pcmk__is_bundled(rsc)) { g_list_foreach(rsc->priv->launched, update_launched_failcount, &fc_data); if (fc_data.failcount > 0) { pcmk__rsc_info(rsc, "Container %s and the resources within it " "have failed %s time%s on %s", rsc->id, pcmk_readable_score(fc_data.failcount), pcmk__plural_s(fc_data.failcount), pcmk__node_name(node)); } } else if (fc_data.failcount > 0) { pcmk__rsc_info(rsc, "%s has failed %s time%s on %s", rsc->id, pcmk_readable_score(fc_data.failcount), pcmk__plural_s(fc_data.failcount), pcmk__node_name(node)); } if (last_failure != NULL) { if ((fc_data.failcount > 0) && (fc_data.last_failure > 0)) { *last_failure = fc_data.last_failure; } else { *last_failure = 0; } } return fc_data.failcount; } /*! * \brief Schedule a controller operation to clear a fail count * * \param[in,out] rsc Resource with failure * \param[in] node Node failure occurred on * \param[in] reason Readable description why needed (for logging) * \param[in,out] scheduler Scheduler data cluster * * \return Scheduled action */ pcmk_action_t * pe__clear_failcount(pcmk_resource_t *rsc, const pcmk_node_t *node, const char *reason, pcmk_scheduler_t *scheduler) { char *key = NULL; pcmk_action_t *clear = NULL; CRM_CHECK(rsc && node && reason && scheduler, return NULL); key = pcmk__op_key(rsc->id, PCMK_ACTION_CLEAR_FAILCOUNT, 0); clear = custom_action(rsc, key, PCMK_ACTION_CLEAR_FAILCOUNT, node, FALSE, scheduler); pcmk__insert_meta(clear, PCMK__META_OP_NO_WAIT, PCMK_VALUE_TRUE); crm_notice("Clearing failure of %s on %s because %s " QB_XS " %s", rsc->id, pcmk__node_name(node), reason, clear->uuid); return clear; } diff --git a/lib/pengine/group.c b/lib/pengine/group.c index 2f3793da6c..9d22ae1312 100644 --- a/lib/pengine/group.c +++ b/lib/pengine/group.c @@ -1,469 +1,469 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include typedef struct group_variant_data_s { pcmk_resource_t *last_child; // Last group member uint32_t flags; // Group of enum pcmk__group_flags } group_variant_data_t; /*! * \internal * \brief Get a group's last member * * \param[in] group Group resource to check * * \return Last member of \p group if any, otherwise NULL */ pcmk_resource_t * pe__last_group_member(const pcmk_resource_t *group) { if (group != NULL) { const group_variant_data_t *group_data = NULL; CRM_CHECK(pcmk__is_group(group), return NULL); group_data = group->priv->variant_opaque; return group_data->last_child; } return NULL; } /*! * \internal * \brief Check whether a group flag is set * * \param[in] group Group resource to check * \param[in] flags Flag or flags to check * * \return true if all \p flags are set for \p group, otherwise false */ bool pe__group_flag_is_set(const pcmk_resource_t *group, uint32_t flags) { const group_variant_data_t *group_data = NULL; CRM_CHECK(pcmk__is_group(group), return false); group_data = group->priv->variant_opaque; return pcmk_all_flags_set(group_data->flags, flags); } /*! * \internal * \brief Set a (deprecated) group flag * * \param[in,out] group Group resource to check * \param[in] option Name of boolean configuration option * \param[in] flag Flag to set if \p option is true (which is default) * \param[in] wo_bit "Warn once" flag to use for deprecation warning */ static void set_group_flag(pcmk_resource_t *group, const char *option, uint32_t flag, uint32_t wo_bit) { const char *value_s = NULL; int value = 0; value_s = g_hash_table_lookup(group->priv->meta, option); // We don't actually need the null check but it speeds up the common case if ((value_s == NULL) || (crm_str_to_boolean(value_s, &value) < 0) || (value != 0)) { group_variant_data_t *group_data = group->priv->variant_opaque; group_data->flags |= flag; } else { pcmk__warn_once(wo_bit, "Support for the '%s' group meta-attribute is " "deprecated and will be removed in a future release " "(use a resource set instead)", option); } } static int inactive_resources(pcmk_resource_t *rsc) { int retval = 0; for (GList *gIter = rsc->priv->children; gIter != NULL; gIter = gIter->next) { pcmk_resource_t *child_rsc = (pcmk_resource_t *) gIter->data; if (!child_rsc->priv->fns->active(child_rsc, TRUE)) { retval++; } } return retval; } static void group_header(pcmk__output_t *out, int *rc, const pcmk_resource_t *rsc, int n_inactive, bool show_inactive, const char *desc) { GString *attrs = NULL; if (n_inactive > 0 && !show_inactive) { attrs = g_string_sized_new(64); g_string_append_printf(attrs, "%d member%s inactive", n_inactive, pcmk__plural_s(n_inactive)); } if (pe__resource_is_disabled(rsc)) { pcmk__add_separated_word(&attrs, 64, "disabled", ", "); } if (pcmk_is_set(rsc->flags, pcmk__rsc_maintenance)) { pcmk__add_separated_word(&attrs, 64, "maintenance", ", "); } else if (!pcmk_is_set(rsc->flags, pcmk__rsc_managed)) { pcmk__add_separated_word(&attrs, 64, "unmanaged", ", "); } if (attrs != NULL) { PCMK__OUTPUT_LIST_HEADER(out, FALSE, *rc, "Resource Group: %s (%s)%s%s%s", rsc->id, (const char *) attrs->str, desc ? " (" : "", desc ? desc : "", desc ? ")" : ""); g_string_free(attrs, TRUE); } else { PCMK__OUTPUT_LIST_HEADER(out, FALSE, *rc, "Resource Group: %s%s%s%s", rsc->id, desc ? " (" : "", desc ? desc : "", desc ? ")" : ""); } } static bool skip_child_rsc(pcmk_resource_t *rsc, pcmk_resource_t *child, gboolean parent_passes, GList *only_rsc, uint32_t show_opts) { bool star_list = pcmk__list_of_1(only_rsc) && pcmk__str_eq("*", g_list_first(only_rsc)->data, pcmk__str_none); bool child_filtered = child->priv->fns->is_filtered(child, only_rsc, FALSE); bool child_active = child->priv->fns->active(child, FALSE); bool show_inactive = pcmk_is_set(show_opts, pcmk_show_inactive_rscs); /* If the resource is in only_rsc by name (so, ignoring "*") then allow * it regardless of if it's active or not. */ if (!star_list && !child_filtered) { return false; } else if (!child_filtered && (child_active || show_inactive)) { return false; } else if (parent_passes && (child_active || show_inactive)) { return false; } return true; } gboolean group_unpack(pcmk_resource_t *rsc, pcmk_scheduler_t *scheduler) { xmlNode *xml_obj = rsc->priv->xml; xmlNode *xml_native_rsc = NULL; group_variant_data_t *group_data = NULL; const char *clone_id = NULL; pcmk__rsc_trace(rsc, "Processing resource %s...", rsc->id); group_data = pcmk__assert_alloc(1, sizeof(group_variant_data_t)); group_data->last_child = NULL; rsc->priv->variant_opaque = group_data; // @COMPAT These are deprecated since 2.1.5 set_group_flag(rsc, PCMK_META_ORDERED, pcmk__group_ordered, pcmk__wo_group_order); set_group_flag(rsc, "collocated", pcmk__group_colocated, pcmk__wo_group_coloc); clone_id = crm_element_value(rsc->priv->xml, PCMK__META_CLONE); for (xml_native_rsc = pcmk__xe_first_child(xml_obj, PCMK_XE_PRIMITIVE, NULL, NULL); xml_native_rsc != NULL; xml_native_rsc = pcmk__xe_next(xml_native_rsc, PCMK_XE_PRIMITIVE)) { pcmk_resource_t *new_rsc = NULL; crm_xml_add(xml_native_rsc, PCMK__META_CLONE, clone_id); if (pe__unpack_resource(xml_native_rsc, &new_rsc, rsc, scheduler) != pcmk_rc_ok) { continue; } rsc->priv->children = g_list_append(rsc->priv->children, new_rsc); group_data->last_child = new_rsc; pcmk__rsc_trace(rsc, "Added %s member %s", rsc->id, new_rsc->id); } if (rsc->priv->children == NULL) { // Not possible with schema validation enabled free(group_data); rsc->priv->variant_opaque = NULL; pcmk__config_err("Group %s has no members", rsc->id); return FALSE; } return TRUE; } gboolean group_active(pcmk_resource_t *rsc, gboolean all) { gboolean c_all = TRUE; gboolean c_any = FALSE; for (GList *gIter = rsc->priv->children; gIter != NULL; gIter = gIter->next) { pcmk_resource_t *child_rsc = (pcmk_resource_t *) gIter->data; if (child_rsc->priv->fns->active(child_rsc, all)) { c_any = TRUE; } else { c_all = FALSE; } } if (c_any == FALSE) { return FALSE; } else if (all && c_all == FALSE) { return FALSE; } return TRUE; } PCMK__OUTPUT_ARGS("group", "uint32_t", "pcmk_resource_t *", "GList *", "GList *") int pe__group_xml(pcmk__output_t *out, va_list args) { uint32_t show_opts = va_arg(args, uint32_t); pcmk_resource_t *rsc = va_arg(args, pcmk_resource_t *); GList *only_node = va_arg(args, GList *); GList *only_rsc = va_arg(args, GList *); const char *desc = NULL; int rc = pcmk_rc_no_output; gboolean parent_passes = pcmk__str_in_list(rsc_printable_id(rsc), only_rsc, pcmk__str_star_matches) || (strstr(rsc->id, ":") != NULL && pcmk__str_in_list(rsc->id, only_rsc, pcmk__str_star_matches)); desc = pe__resource_description(rsc, show_opts); if (rsc->priv->fns->is_filtered(rsc, only_rsc, TRUE)) { return rc; } for (GList *gIter = rsc->priv->children; gIter != NULL; gIter = gIter->next) { pcmk_resource_t *child_rsc = (pcmk_resource_t *) gIter->data; if (skip_child_rsc(rsc, child_rsc, parent_passes, only_rsc, show_opts)) { continue; } if (rc == pcmk_rc_no_output) { char *count = pcmk__itoa(g_list_length(gIter)); const char *maintenance = pcmk__flag_text(rsc->flags, pcmk__rsc_maintenance); const char *managed = pcmk__flag_text(rsc->flags, pcmk__rsc_managed); const char *disabled = pcmk__btoa(pe__resource_is_disabled(rsc)); rc = pe__name_and_nvpairs_xml(out, true, PCMK_XE_GROUP, PCMK_XA_ID, rsc->id, PCMK_XA_NUMBER_RESOURCES, count, PCMK_XA_MAINTENANCE, maintenance, PCMK_XA_MANAGED, managed, PCMK_XA_DISABLED, disabled, PCMK_XA_DESCRIPTION, desc, NULL); free(count); pcmk__assert(rc == pcmk_rc_ok); } out->message(out, (const char *) child_rsc->priv->xml->name, show_opts, child_rsc, only_node, only_rsc); } if (rc == pcmk_rc_ok) { pcmk__output_xml_pop_parent(out); } return rc; } PCMK__OUTPUT_ARGS("group", "uint32_t", "pcmk_resource_t *", "GList *", "GList *") int pe__group_default(pcmk__output_t *out, va_list args) { uint32_t show_opts = va_arg(args, uint32_t); pcmk_resource_t *rsc = va_arg(args, pcmk_resource_t *); GList *only_node = va_arg(args, GList *); GList *only_rsc = va_arg(args, GList *); const char *desc = NULL; int rc = pcmk_rc_no_output; gboolean parent_passes = pcmk__str_in_list(rsc_printable_id(rsc), only_rsc, pcmk__str_star_matches) || (strstr(rsc->id, ":") != NULL && pcmk__str_in_list(rsc->id, only_rsc, pcmk__str_star_matches)); gboolean active = rsc->priv->fns->active(rsc, TRUE); gboolean partially_active = rsc->priv->fns->active(rsc, FALSE); desc = pe__resource_description(rsc, show_opts); if (rsc->priv->fns->is_filtered(rsc, only_rsc, TRUE)) { return rc; } if (pcmk_is_set(show_opts, pcmk_show_brief)) { GList *rscs = pe__filter_rsc_list(rsc->priv->children, only_rsc); if (rscs != NULL) { group_header(out, &rc, rsc, !active && partially_active ? inactive_resources(rsc) : 0, pcmk_is_set(show_opts, pcmk_show_inactive_rscs), desc); pe__rscs_brief_output(out, rscs, show_opts | pcmk_show_inactive_rscs); rc = pcmk_rc_ok; g_list_free(rscs); } } else { for (GList *gIter = rsc->priv->children; gIter != NULL; gIter = gIter->next) { pcmk_resource_t *child_rsc = (pcmk_resource_t *) gIter->data; if (skip_child_rsc(rsc, child_rsc, parent_passes, only_rsc, show_opts)) { continue; } group_header(out, &rc, rsc, !active && partially_active ? inactive_resources(rsc) : 0, pcmk_is_set(show_opts, pcmk_show_inactive_rscs), desc); out->message(out, (const char *) child_rsc->priv->xml->name, show_opts, child_rsc, only_node, only_rsc); } } PCMK__OUTPUT_LIST_FOOTER(out, rc); return rc; } void group_free(pcmk_resource_t * rsc) { CRM_CHECK(rsc != NULL, return); pcmk__rsc_trace(rsc, "Freeing %s", rsc->id); for (GList *gIter = rsc->priv->children; gIter != NULL; gIter = gIter->next) { pcmk_resource_t *child_rsc = (pcmk_resource_t *) gIter->data; pcmk__assert(child_rsc != NULL); pcmk__rsc_trace(child_rsc, "Freeing child %s", child_rsc->id); - child_rsc->priv->fns->free(child_rsc); + pcmk__free_resource(child_rsc); } pcmk__rsc_trace(rsc, "Freeing child list"); g_list_free(rsc->priv->children); common_free(rsc); } enum rsc_role_e group_resource_state(const pcmk_resource_t * rsc, gboolean current) { enum rsc_role_e group_role = pcmk_role_unknown; for (GList *gIter = rsc->priv->children; gIter != NULL; gIter = gIter->next) { pcmk_resource_t *child_rsc = (pcmk_resource_t *) gIter->data; enum rsc_role_e role = child_rsc->priv->fns->state(child_rsc, current); if (role > group_role) { group_role = role; } } pcmk__rsc_trace(rsc, "%s role: %s", rsc->id, pcmk_role_text(group_role)); return group_role; } gboolean pe__group_is_filtered(const pcmk_resource_t *rsc, GList *only_rsc, gboolean check_parent) { gboolean passes = FALSE; if (check_parent && pcmk__str_in_list(rsc_printable_id(pe__const_top_resource(rsc, false)), only_rsc, pcmk__str_star_matches)) { passes = TRUE; } else if (pcmk__str_in_list(rsc_printable_id(rsc), only_rsc, pcmk__str_star_matches)) { passes = TRUE; } else if (strstr(rsc->id, ":") != NULL && pcmk__str_in_list(rsc->id, only_rsc, pcmk__str_star_matches)) { passes = TRUE; } else { for (const GList *iter = rsc->priv->children; iter != NULL; iter = iter->next) { const pcmk_resource_t *child_rsc = iter->data; if (!child_rsc->priv->fns->is_filtered(child_rsc, only_rsc, FALSE)) { passes = TRUE; break; } } } return !passes; } /*! * \internal * \brief Get maximum group resource instances per node * * \param[in] rsc Group resource to check * * \return Maximum number of \p rsc instances that can be active on one node */ unsigned int pe__group_max_per_node(const pcmk_resource_t *rsc) { pcmk__assert(pcmk__is_group(rsc)); return 1U; } diff --git a/lib/pengine/pe_actions.c b/lib/pengine/pe_actions.c index 7bc65dcad1..2cec664f7f 100644 --- a/lib/pengine/pe_actions.c +++ b/lib/pengine/pe_actions.c @@ -1,1770 +1,1748 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include "pe_status_private.h" static void unpack_operation(pcmk_action_t *action, const xmlNode *xml_obj, guint interval_ms); static void add_singleton(pcmk_scheduler_t *scheduler, pcmk_action_t *action) { if (scheduler->priv->singletons == NULL) { scheduler->priv->singletons = pcmk__strkey_table(NULL, NULL); } g_hash_table_insert(scheduler->priv->singletons, action->uuid, action); } static pcmk_action_t * lookup_singleton(pcmk_scheduler_t *scheduler, const char *action_uuid) { /* @TODO This is the only use of the pcmk_scheduler_t:singletons hash table. * Compare the performance of this approach to keeping the * pcmk_scheduler_t:actions list sorted by action key and just searching * that instead. */ if (scheduler->priv->singletons == NULL) { return NULL; } return g_hash_table_lookup(scheduler->priv->singletons, action_uuid); } /*! * \internal * \brief Find an existing action that matches arguments * * \param[in] key Action key to match * \param[in] rsc Resource to match (if any) * \param[in] node Node to match (if any) * \param[in] scheduler Scheduler data * * \return Existing action that matches arguments (or NULL if none) */ static pcmk_action_t * find_existing_action(const char *key, const pcmk_resource_t *rsc, const pcmk_node_t *node, const pcmk_scheduler_t *scheduler) { /* When rsc is NULL, it would be quicker to check * scheduler->priv->singletons, but checking all scheduler->priv->actions * takes the node into account. */ GList *actions = (rsc == NULL)? scheduler->priv->actions : rsc->priv->actions; GList *matches = find_actions(actions, key, node); pcmk_action_t *action = NULL; if (matches == NULL) { return NULL; } CRM_LOG_ASSERT(!pcmk__list_of_multiple(matches)); action = matches->data; g_list_free(matches); return action; } /*! * \internal * \brief Find the XML configuration corresponding to a specific action key * * \param[in] rsc Resource to find action configuration for * \param[in] key "RSC_ACTION_INTERVAL" of action to find * \param[in] include_disabled If false, do not return disabled actions * * \return XML configuration of desired action if any, otherwise NULL */ static xmlNode * find_exact_action_config(const pcmk_resource_t *rsc, const char *action_name, guint interval_ms, bool include_disabled) { for (xmlNode *operation = pcmk__xe_first_child(rsc->priv->ops_xml, PCMK_XE_OP, NULL, NULL); operation != NULL; operation = pcmk__xe_next(operation, PCMK_XE_OP)) { bool enabled = false; const char *config_name = NULL; const char *interval_spec = NULL; guint tmp_ms = 0U; // @TODO This does not consider meta-attributes, rules, defaults, etc. if (!include_disabled && (pcmk__xe_get_bool_attr(operation, PCMK_META_ENABLED, &enabled) == pcmk_rc_ok) && !enabled) { continue; } interval_spec = crm_element_value(operation, PCMK_META_INTERVAL); pcmk_parse_interval_spec(interval_spec, &tmp_ms); if (tmp_ms != interval_ms) { continue; } config_name = crm_element_value(operation, PCMK_XA_NAME); if (pcmk__str_eq(action_name, config_name, pcmk__str_none)) { return operation; } } return NULL; } /*! * \internal * \brief Find the XML configuration of a resource action * * \param[in] rsc Resource to find action configuration for * \param[in] action_name Action name to search for * \param[in] interval_ms Action interval (in milliseconds) to search for * \param[in] include_disabled If false, do not return disabled actions * * \return XML configuration of desired action if any, otherwise NULL */ xmlNode * pcmk__find_action_config(const pcmk_resource_t *rsc, const char *action_name, guint interval_ms, bool include_disabled) { xmlNode *action_config = NULL; // Try requested action first action_config = find_exact_action_config(rsc, action_name, interval_ms, include_disabled); // For migrate_to and migrate_from actions, retry with "migrate" // @TODO This should be either documented or deprecated if ((action_config == NULL) && pcmk__str_any_of(action_name, PCMK_ACTION_MIGRATE_TO, PCMK_ACTION_MIGRATE_FROM, NULL)) { action_config = find_exact_action_config(rsc, "migrate", 0, include_disabled); } return action_config; } /*! * \internal * \brief Create a new action object * * \param[in] key Action key * \param[in] task Action name * \param[in,out] rsc Resource that action is for (if any) * \param[in] node Node that action is on (if any) * \param[in] optional Whether action should be considered optional * \param[in,out] scheduler Scheduler data * * \return Newly allocated action * \note This function takes ownership of \p key. It is the caller's - * responsibility to free the return value with pe_free_action(). + * responsibility to free the return value using pcmk__free_action(). */ static pcmk_action_t * new_action(char *key, const char *task, pcmk_resource_t *rsc, const pcmk_node_t *node, bool optional, pcmk_scheduler_t *scheduler) { pcmk_action_t *action = pcmk__assert_alloc(1, sizeof(pcmk_action_t)); action->rsc = rsc; action->task = pcmk__str_copy(task); action->uuid = key; action->scheduler = scheduler; if (node) { action->node = pe__copy_node(node); } if (pcmk__str_eq(task, PCMK_ACTION_LRM_DELETE, pcmk__str_casei)) { // Resource history deletion for a node can be done on the DC pcmk__set_action_flags(action, pcmk__action_on_dc); } pcmk__set_action_flags(action, pcmk__action_runnable); if (optional) { pcmk__set_action_flags(action, pcmk__action_optional); } else { pcmk__clear_action_flags(action, pcmk__action_optional); } if (rsc == NULL) { action->meta = pcmk__strkey_table(free, free); } else { guint interval_ms = 0; parse_op_key(key, NULL, NULL, &interval_ms); action->op_entry = pcmk__find_action_config(rsc, task, interval_ms, true); /* If the given key is for one of the many notification pseudo-actions * (pre_notify_promote, etc.), the actual action name is "notify" */ if ((action->op_entry == NULL) && (strstr(key, "_notify_") != NULL)) { action->op_entry = find_exact_action_config(rsc, PCMK_ACTION_NOTIFY, 0, true); } unpack_operation(action, action->op_entry, interval_ms); } pcmk__rsc_trace(rsc, "Created %s action %d (%s): %s for %s on %s", (optional? "optional" : "required"), scheduler->priv->next_action_id, key, task, ((rsc == NULL)? "no resource" : rsc->id), pcmk__node_name(node)); action->id = scheduler->priv->next_action_id++; scheduler->priv->actions = g_list_prepend(scheduler->priv->actions, action); if (rsc == NULL) { add_singleton(scheduler, action); } else { rsc->priv->actions = g_list_prepend(rsc->priv->actions, action); } return action; } /*! * \internal * \brief Unpack a resource's action-specific instance parameters * * \param[in] action_xml XML of action's configuration in CIB (if any) * \param[in,out] node_attrs Table of node attributes (for rule evaluation) - * \param[in,out] scheduler Cluster working set (for rule evaluation) + * \param[in,out] scheduler Scheduler data (for rule evaluation) * * \return Newly allocated hash table of action-specific instance parameters */ GHashTable * pcmk__unpack_action_rsc_params(const xmlNode *action_xml, GHashTable *node_attrs, pcmk_scheduler_t *scheduler) { GHashTable *params = pcmk__strkey_table(free, free); const pcmk_rule_input_t rule_input = { .now = scheduler->priv->now, .node_attrs = node_attrs, }; pe__unpack_dataset_nvpairs(action_xml, PCMK_XE_INSTANCE_ATTRIBUTES, &rule_input, params, NULL, scheduler); return params; } /*! * \internal * \brief Update an action's optional flag * * \param[in,out] action Action to update * \param[in] optional Requested optional status */ static void update_action_optional(pcmk_action_t *action, gboolean optional) { // Force a non-recurring action to be optional if its resource is unmanaged if ((action->rsc != NULL) && (action->node != NULL) && !pcmk_is_set(action->flags, pcmk__action_pseudo) && !pcmk_is_set(action->rsc->flags, pcmk__rsc_managed) && (g_hash_table_lookup(action->meta, PCMK_META_INTERVAL) == NULL)) { pcmk__rsc_debug(action->rsc, "%s on %s is optional (%s is unmanaged)", action->uuid, pcmk__node_name(action->node), action->rsc->id); pcmk__set_action_flags(action, pcmk__action_optional); // We shouldn't clear runnable here because ... something // Otherwise require the action if requested } else if (!optional) { pcmk__clear_action_flags(action, pcmk__action_optional); } } static enum pe_quorum_policy effective_quorum_policy(pcmk_resource_t *rsc, pcmk_scheduler_t *scheduler) { enum pe_quorum_policy policy = scheduler->no_quorum_policy; if (pcmk_is_set(scheduler->flags, pcmk__sched_quorate)) { policy = pcmk_no_quorum_ignore; } else if (scheduler->no_quorum_policy == pcmk_no_quorum_demote) { switch (rsc->priv->orig_role) { case pcmk_role_promoted: case pcmk_role_unpromoted: if (rsc->priv->next_role > pcmk_role_unpromoted) { pe__set_next_role(rsc, pcmk_role_unpromoted, PCMK_OPT_NO_QUORUM_POLICY "=demote"); } policy = pcmk_no_quorum_ignore; break; default: policy = pcmk_no_quorum_stop; break; } } return policy; } /*! * \internal * \brief Update a resource action's runnable flag * * \param[in,out] action Action to update * \param[in,out] scheduler Scheduler data * * \note This may also schedule fencing if a stop is unrunnable. */ static void update_resource_action_runnable(pcmk_action_t *action, pcmk_scheduler_t *scheduler) { pcmk_resource_t *rsc = action->rsc; if (pcmk_is_set(action->flags, pcmk__action_pseudo)) { return; } if (action->node == NULL) { pcmk__rsc_trace(rsc, "%s is unrunnable (unallocated)", action->uuid); pcmk__clear_action_flags(action, pcmk__action_runnable); } else if (!pcmk_is_set(action->flags, pcmk__action_on_dc) && !(action->node->details->online) && (!pcmk__is_guest_or_bundle_node(action->node) || pcmk_is_set(action->node->priv->flags, pcmk__node_remote_reset))) { pcmk__clear_action_flags(action, pcmk__action_runnable); do_crm_log(LOG_WARNING, "%s on %s is unrunnable (node is offline)", action->uuid, pcmk__node_name(action->node)); if (pcmk_is_set(rsc->flags, pcmk__rsc_managed) && pcmk__str_eq(action->task, PCMK_ACTION_STOP, pcmk__str_casei) && !(action->node->details->unclean)) { pe_fence_node(scheduler, action->node, "stop is unrunnable", false); } } else if (!pcmk_is_set(action->flags, pcmk__action_on_dc) && action->node->details->pending) { pcmk__clear_action_flags(action, pcmk__action_runnable); do_crm_log(LOG_WARNING, "Action %s on %s is unrunnable (node is pending)", action->uuid, pcmk__node_name(action->node)); } else if (action->needs == pcmk__requires_nothing) { pe_action_set_reason(action, NULL, TRUE); if (pcmk__is_guest_or_bundle_node(action->node) && !pe_can_fence(scheduler, action->node)) { /* An action that requires nothing usually does not require any * fencing in order to be runnable. However, there is an exception: * such an action cannot be completed if it is on a guest node whose * host is unclean and cannot be fenced. */ pcmk__rsc_debug(rsc, "%s on %s is unrunnable " "(node's host cannot be fenced)", action->uuid, pcmk__node_name(action->node)); pcmk__clear_action_flags(action, pcmk__action_runnable); } else { pcmk__rsc_trace(rsc, "%s on %s does not require fencing or quorum", action->uuid, pcmk__node_name(action->node)); pcmk__set_action_flags(action, pcmk__action_runnable); } } else { switch (effective_quorum_policy(rsc, scheduler)) { case pcmk_no_quorum_stop: pcmk__rsc_debug(rsc, "%s on %s is unrunnable (no quorum)", action->uuid, pcmk__node_name(action->node)); pcmk__clear_action_flags(action, pcmk__action_runnable); pe_action_set_reason(action, "no quorum", true); break; case pcmk_no_quorum_freeze: if (!rsc->priv->fns->active(rsc, TRUE) || (rsc->priv->next_role > rsc->priv->orig_role)) { pcmk__rsc_debug(rsc, "%s on %s is unrunnable (no quorum)", action->uuid, pcmk__node_name(action->node)); pcmk__clear_action_flags(action, pcmk__action_runnable); pe_action_set_reason(action, "quorum freeze", true); } break; default: //pe_action_set_reason(action, NULL, TRUE); pcmk__set_action_flags(action, pcmk__action_runnable); break; } } } static bool valid_stop_on_fail(const char *value) { return !pcmk__strcase_any_of(value, PCMK_VALUE_STANDBY, PCMK_VALUE_DEMOTE, PCMK_VALUE_STOP, NULL); } /*! * \internal * \brief Validate (and possibly reset) resource action's on_fail meta-attribute * * \param[in] rsc Resource that action is for * \param[in] action_name Action name * \param[in] action_config Action configuration XML from CIB (if any) * \param[in,out] meta Table of action meta-attributes */ static void validate_on_fail(const pcmk_resource_t *rsc, const char *action_name, const xmlNode *action_config, GHashTable *meta) { const char *name = NULL; const char *role = NULL; const char *interval_spec = NULL; const char *value = g_hash_table_lookup(meta, PCMK_META_ON_FAIL); guint interval_ms = 0U; // Stop actions can only use certain on-fail values if (pcmk__str_eq(action_name, PCMK_ACTION_STOP, pcmk__str_none) && !valid_stop_on_fail(value)) { pcmk__config_err("Resetting '" PCMK_META_ON_FAIL "' for %s stop " "action to default value because '%s' is not " "allowed for stop", rsc->id, value); g_hash_table_remove(meta, PCMK_META_ON_FAIL); return; } /* Demote actions default on-fail to the on-fail value for the first * recurring monitor for the promoted role (if any). */ if (pcmk__str_eq(action_name, PCMK_ACTION_DEMOTE, pcmk__str_none) && (value == NULL)) { /* @TODO This does not consider promote options set in a meta-attribute * block (which may have rules that need to be evaluated) rather than * XML properties. */ for (xmlNode *operation = pcmk__xe_first_child(rsc->priv->ops_xml, PCMK_XE_OP, NULL, NULL); operation != NULL; operation = pcmk__xe_next(operation, PCMK_XE_OP)) { bool enabled = false; const char *promote_on_fail = NULL; /* We only care about explicit on-fail (if promote uses default, so * can demote) */ promote_on_fail = crm_element_value(operation, PCMK_META_ON_FAIL); if (promote_on_fail == NULL) { continue; } // We only care about recurring monitors for the promoted role name = crm_element_value(operation, PCMK_XA_NAME); role = crm_element_value(operation, PCMK_XA_ROLE); if (!pcmk__str_eq(name, PCMK_ACTION_MONITOR, pcmk__str_none) || !pcmk__strcase_any_of(role, PCMK_ROLE_PROMOTED, PCMK__ROLE_PROMOTED_LEGACY, NULL)) { continue; } interval_spec = crm_element_value(operation, PCMK_META_INTERVAL); pcmk_parse_interval_spec(interval_spec, &interval_ms); if (interval_ms == 0U) { continue; } // We only care about enabled monitors if ((pcmk__xe_get_bool_attr(operation, PCMK_META_ENABLED, &enabled) == pcmk_rc_ok) && !enabled) { continue; } /* Demote actions can't default to * PCMK_META_ON_FAIL=PCMK_VALUE_DEMOTE */ if (pcmk__str_eq(promote_on_fail, PCMK_VALUE_DEMOTE, pcmk__str_casei)) { continue; } // Use value from first applicable promote action found pcmk__insert_dup(meta, PCMK_META_ON_FAIL, promote_on_fail); } return; } if (pcmk__str_eq(action_name, PCMK_ACTION_LRM_DELETE, pcmk__str_none) && !pcmk__str_eq(value, PCMK_VALUE_IGNORE, pcmk__str_casei)) { pcmk__insert_dup(meta, PCMK_META_ON_FAIL, PCMK_VALUE_IGNORE); return; } // PCMK_META_ON_FAIL=PCMK_VALUE_DEMOTE is allowed only for certain actions if (pcmk__str_eq(value, PCMK_VALUE_DEMOTE, pcmk__str_casei)) { name = crm_element_value(action_config, PCMK_XA_NAME); role = crm_element_value(action_config, PCMK_XA_ROLE); interval_spec = crm_element_value(action_config, PCMK_META_INTERVAL); pcmk_parse_interval_spec(interval_spec, &interval_ms); if (!pcmk__str_eq(name, PCMK_ACTION_PROMOTE, pcmk__str_none) && ((interval_ms == 0U) || !pcmk__str_eq(name, PCMK_ACTION_MONITOR, pcmk__str_none) || !pcmk__strcase_any_of(role, PCMK_ROLE_PROMOTED, PCMK__ROLE_PROMOTED_LEGACY, NULL))) { pcmk__config_err("Resetting '" PCMK_META_ON_FAIL "' for %s %s " "action to default value because 'demote' is not " "allowed for it", rsc->id, name); g_hash_table_remove(meta, PCMK_META_ON_FAIL); return; } } } static int unpack_timeout(const char *value) { long long timeout_ms = crm_get_msec(value); if (timeout_ms <= 0) { timeout_ms = PCMK_DEFAULT_ACTION_TIMEOUT_MS; } return (int) QB_MIN(timeout_ms, INT_MAX); } // true if value contains valid, non-NULL interval origin for recurring op static bool unpack_interval_origin(const char *value, const xmlNode *xml_obj, guint interval_ms, const crm_time_t *now, long long *start_delay) { long long result = 0; guint interval_sec = pcmk__timeout_ms2s(interval_ms); crm_time_t *origin = NULL; // Ignore unspecified values and non-recurring operations if ((value == NULL) || (interval_ms == 0) || (now == NULL)) { return false; } // Parse interval origin from text origin = crm_time_new(value); if (origin == NULL) { pcmk__config_err("Ignoring '" PCMK_META_INTERVAL_ORIGIN "' for " "operation '%s' because '%s' is not valid", pcmk__s(pcmk__xe_id(xml_obj), "(missing ID)"), value); return false; } // Get seconds since origin (negative if origin is in the future) result = crm_time_get_seconds(now) - crm_time_get_seconds(origin); crm_time_free(origin); // Calculate seconds from closest interval to now result = result % interval_sec; // Calculate seconds remaining until next interval result = ((result <= 0)? 0 : interval_sec) - result; crm_info("Calculated a start delay of %llds for operation '%s'", result, pcmk__s(pcmk__xe_id(xml_obj), "(unspecified)")); if (start_delay != NULL) { *start_delay = result * 1000; // milliseconds } return true; } static int unpack_start_delay(const char *value, GHashTable *meta) { long long start_delay_ms = 0; if (value == NULL) { return 0; } start_delay_ms = crm_get_msec(value); start_delay_ms = QB_MIN(start_delay_ms, INT_MAX); if (start_delay_ms < 0) { start_delay_ms = 0; } if (meta != NULL) { g_hash_table_replace(meta, strdup(PCMK_META_START_DELAY), pcmk__itoa(start_delay_ms)); } return (int) start_delay_ms; } /*! * \internal * \brief Find a resource's most frequent recurring monitor * * \param[in] rsc Resource to check * * \return Operation XML configured for most frequent recurring monitor for * \p rsc (if any) */ static xmlNode * most_frequent_monitor(const pcmk_resource_t *rsc) { guint min_interval_ms = G_MAXUINT; xmlNode *op = NULL; for (xmlNode *operation = pcmk__xe_first_child(rsc->priv->ops_xml, PCMK_XE_OP, NULL, NULL); operation != NULL; operation = pcmk__xe_next(operation, PCMK_XE_OP)) { bool enabled = false; guint interval_ms = 0U; const char *interval_spec = crm_element_value(operation, PCMK_META_INTERVAL); // We only care about enabled recurring monitors if (!pcmk__str_eq(crm_element_value(operation, PCMK_XA_NAME), PCMK_ACTION_MONITOR, pcmk__str_none)) { continue; } pcmk_parse_interval_spec(interval_spec, &interval_ms); if (interval_ms == 0U) { continue; } // @TODO This does not consider meta-attributes, rules, defaults, etc. if ((pcmk__xe_get_bool_attr(operation, PCMK_META_ENABLED, &enabled) == pcmk_rc_ok) && !enabled) { continue; } if (interval_ms < min_interval_ms) { min_interval_ms = interval_ms; op = operation; } } return op; } /*! * \internal * \brief Unpack action meta-attributes * * \param[in,out] rsc Resource that action is for * \param[in] node Node that action is on * \param[in] action_name Action name * \param[in] interval_ms Action interval (in milliseconds) * \param[in] action_config Action XML configuration from CIB (if any) * * Unpack a resource action's meta-attributes (normalizing the interval, * timeout, and start delay values as integer milliseconds) from its CIB XML * configuration (including defaults). * * \return Newly allocated hash table with normalized action meta-attributes */ GHashTable * pcmk__unpack_action_meta(pcmk_resource_t *rsc, const pcmk_node_t *node, const char *action_name, guint interval_ms, const xmlNode *action_config) { GHashTable *meta = NULL; const char *timeout_spec = NULL; const char *str = NULL; const pcmk_rule_input_t rule_input = { /* Node attributes are not set because node expressions are not allowed * for meta-attributes */ .now = rsc->priv->scheduler->priv->now, .rsc_standard = crm_element_value(rsc->priv->xml, PCMK_XA_CLASS), .rsc_provider = crm_element_value(rsc->priv->xml, PCMK_XA_PROVIDER), .rsc_agent = crm_element_value(rsc->priv->xml, PCMK_XA_TYPE), .op_name = action_name, .op_interval_ms = interval_ms, }; meta = pcmk__strkey_table(free, free); if (action_config != NULL) { // take precedence over defaults pe__unpack_dataset_nvpairs(action_config, PCMK_XE_META_ATTRIBUTES, &rule_input, meta, NULL, rsc->priv->scheduler); /* Anything set as an XML property has highest precedence. * This ensures we use the name and interval from the tag. * (See below for the only exception, fence device start/probe timeout.) */ for (xmlAttrPtr attr = action_config->properties; attr != NULL; attr = attr->next) { pcmk__insert_dup(meta, (const char *) attr->name, pcmk__xml_attr_value(attr)); } } // Derive default timeout for probes from recurring monitor timeouts if (pcmk_is_probe(action_name, interval_ms) && (g_hash_table_lookup(meta, PCMK_META_TIMEOUT) == NULL)) { xmlNode *min_interval_mon = most_frequent_monitor(rsc); if (min_interval_mon != NULL) { /* @TODO This does not consider timeouts set in * PCMK_XE_META_ATTRIBUTES blocks (which may also have rules that * need to be evaluated). */ timeout_spec = crm_element_value(min_interval_mon, PCMK_META_TIMEOUT); if (timeout_spec != NULL) { pcmk__rsc_trace(rsc, "Setting default timeout for %s probe to " "most frequent monitor's timeout '%s'", rsc->id, timeout_spec); pcmk__insert_dup(meta, PCMK_META_TIMEOUT, timeout_spec); } } } // Cluster-wide pe__unpack_dataset_nvpairs(rsc->priv->scheduler->priv->op_defaults, PCMK_XE_META_ATTRIBUTES, &rule_input, meta, NULL, rsc->priv->scheduler); g_hash_table_remove(meta, PCMK_XA_ID); // Normalize interval to milliseconds if (interval_ms > 0) { g_hash_table_insert(meta, pcmk__str_copy(PCMK_META_INTERVAL), crm_strdup_printf("%u", interval_ms)); } else { g_hash_table_remove(meta, PCMK_META_INTERVAL); } /* Timeout order of precedence (highest to lowest): * 1. pcmk_monitor_timeout resource parameter (only for starts and probes * when rsc has pcmk_ra_cap_fence_params; this gets used for recurring * monitors via the executor instead) * 2. timeout configured in (with taking precedence over * ) * 3. timeout configured in * 4. PCMK_DEFAULT_ACTION_TIMEOUT_MS */ // Check for pcmk_monitor_timeout if (pcmk_is_set(pcmk_get_ra_caps(rule_input.rsc_standard), pcmk_ra_cap_fence_params) && (pcmk__str_eq(action_name, PCMK_ACTION_START, pcmk__str_none) || pcmk_is_probe(action_name, interval_ms))) { GHashTable *params = pe_rsc_params(rsc, node, rsc->priv->scheduler); timeout_spec = g_hash_table_lookup(params, "pcmk_monitor_timeout"); if (timeout_spec != NULL) { pcmk__rsc_trace(rsc, "Setting timeout for %s %s to " "pcmk_monitor_timeout (%s)", rsc->id, action_name, timeout_spec); pcmk__insert_dup(meta, PCMK_META_TIMEOUT, timeout_spec); } } // Normalize timeout to positive milliseconds timeout_spec = g_hash_table_lookup(meta, PCMK_META_TIMEOUT); g_hash_table_insert(meta, pcmk__str_copy(PCMK_META_TIMEOUT), pcmk__itoa(unpack_timeout(timeout_spec))); // Ensure on-fail has a valid value validate_on_fail(rsc, action_name, action_config, meta); // Normalize PCMK_META_START_DELAY str = g_hash_table_lookup(meta, PCMK_META_START_DELAY); if (str != NULL) { unpack_start_delay(str, meta); } else { long long start_delay = 0; str = g_hash_table_lookup(meta, PCMK_META_INTERVAL_ORIGIN); if (unpack_interval_origin(str, action_config, interval_ms, rsc->priv->scheduler->priv->now, &start_delay)) { g_hash_table_insert(meta, pcmk__str_copy(PCMK_META_START_DELAY), crm_strdup_printf("%lld", start_delay)); } } return meta; } /*! * \internal * \brief Determine an action's quorum and fencing dependency * * \param[in] rsc Resource that action is for * \param[in] action_name Name of action being unpacked * * \return Quorum and fencing dependency appropriate to action */ enum pcmk__requires pcmk__action_requires(const pcmk_resource_t *rsc, const char *action_name) { const char *value = NULL; enum pcmk__requires requires = pcmk__requires_nothing; CRM_CHECK((rsc != NULL) && (action_name != NULL), return requires); if (!pcmk__strcase_any_of(action_name, PCMK_ACTION_START, PCMK_ACTION_PROMOTE, NULL)) { value = "nothing (not start or promote)"; } else if (pcmk_is_set(rsc->flags, pcmk__rsc_needs_fencing)) { requires = pcmk__requires_fencing; value = "fencing"; } else if (pcmk_is_set(rsc->flags, pcmk__rsc_needs_quorum)) { requires = pcmk__requires_quorum; value = "quorum"; } else { value = "nothing"; } pcmk__rsc_trace(rsc, "%s of %s requires %s", action_name, rsc->id, value); return requires; } /*! * \internal * \brief Parse action failure response from a user-provided string * * \param[in] rsc Resource that action is for * \param[in] action_name Name of action * \param[in] interval_ms Action interval (in milliseconds) * \param[in] value User-provided configuration value for on-fail * * \return Action failure response parsed from \p text */ enum pcmk__on_fail pcmk__parse_on_fail(const pcmk_resource_t *rsc, const char *action_name, guint interval_ms, const char *value) { const char *desc = NULL; bool needs_remote_reset = false; enum pcmk__on_fail on_fail = pcmk__on_fail_ignore; const pcmk_scheduler_t *scheduler = NULL; // There's no enum value for unknown or invalid, so assert pcmk__assert((rsc != NULL) && (action_name != NULL)); scheduler = rsc->priv->scheduler; if (value == NULL) { // Use default } else if (pcmk__str_eq(value, PCMK_VALUE_BLOCK, pcmk__str_casei)) { on_fail = pcmk__on_fail_block; desc = "block"; } else if (pcmk__str_eq(value, PCMK_VALUE_FENCE, pcmk__str_casei)) { if (pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled)) { on_fail = pcmk__on_fail_fence_node; desc = "node fencing"; } else { pcmk__config_err("Resetting '" PCMK_META_ON_FAIL "' for " "%s of %s to 'stop' because 'fence' is not " "valid when fencing is disabled", action_name, rsc->id); on_fail = pcmk__on_fail_stop; desc = "stop resource"; } } else if (pcmk__str_eq(value, PCMK_VALUE_STANDBY, pcmk__str_casei)) { on_fail = pcmk__on_fail_standby_node; desc = "node standby"; } else if (pcmk__strcase_any_of(value, PCMK_VALUE_IGNORE, PCMK_VALUE_NOTHING, NULL)) { desc = "ignore"; } else if (pcmk__str_eq(value, "migrate", pcmk__str_casei)) { on_fail = pcmk__on_fail_ban; desc = "force migration"; } else if (pcmk__str_eq(value, PCMK_VALUE_STOP, pcmk__str_casei)) { on_fail = pcmk__on_fail_stop; desc = "stop resource"; } else if (pcmk__str_eq(value, PCMK_VALUE_RESTART, pcmk__str_casei)) { on_fail = pcmk__on_fail_restart; desc = "restart (and possibly migrate)"; } else if (pcmk__str_eq(value, PCMK_VALUE_RESTART_CONTAINER, pcmk__str_casei)) { if (rsc->priv->launcher == NULL) { pcmk__rsc_debug(rsc, "Using default " PCMK_META_ON_FAIL " for %s " "of %s because it does not have a launcher", action_name, rsc->id); } else { on_fail = pcmk__on_fail_restart_container; desc = "restart container (and possibly migrate)"; } } else if (pcmk__str_eq(value, PCMK_VALUE_DEMOTE, pcmk__str_casei)) { on_fail = pcmk__on_fail_demote; desc = "demote instance"; } else { pcmk__config_err("Using default '" PCMK_META_ON_FAIL "' for " "%s of %s because '%s' is not valid", action_name, rsc->id, value); } /* Remote node connections are handled specially. Failures that result * in dropping an active connection must result in fencing. The only * failures that don't are probes and starts. The user can explicitly set * PCMK_META_ON_FAIL=PCMK_VALUE_FENCE to fence after start failures. */ if (pcmk_is_set(rsc->flags, pcmk__rsc_is_remote_connection) && pcmk__is_remote_node(pcmk_find_node(scheduler, rsc->id)) && !pcmk_is_probe(action_name, interval_ms) && !pcmk__str_eq(action_name, PCMK_ACTION_START, pcmk__str_none)) { needs_remote_reset = true; if (!pcmk_is_set(rsc->flags, pcmk__rsc_managed)) { desc = NULL; // Force default for unmanaged connections } } if (desc != NULL) { // Explicit value used, default not needed } else if (rsc->priv->launcher != NULL) { on_fail = pcmk__on_fail_restart_container; desc = "restart container (and possibly migrate) (default)"; } else if (needs_remote_reset) { if (pcmk_is_set(rsc->flags, pcmk__rsc_managed)) { if (pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled)) { desc = "fence remote node (default)"; } else { desc = "recover remote node connection (default)"; } on_fail = pcmk__on_fail_reset_remote; } else { on_fail = pcmk__on_fail_stop; desc = "stop unmanaged remote node (enforcing default)"; } } else if (pcmk__str_eq(action_name, PCMK_ACTION_STOP, pcmk__str_none)) { if (pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled)) { on_fail = pcmk__on_fail_fence_node; desc = "resource fence (default)"; } else { on_fail = pcmk__on_fail_block; desc = "resource block (default)"; } } else { on_fail = pcmk__on_fail_restart; desc = "restart (and possibly migrate) (default)"; } pcmk__rsc_trace(rsc, "Failure handling for %s-interval %s of %s: %s", pcmk__readable_interval(interval_ms), action_name, rsc->id, desc); return on_fail; } /*! * \internal * \brief Determine a resource's role after failure of an action * * \param[in] rsc Resource that action is for * \param[in] action_name Action name * \param[in] on_fail Failure handling for action * \param[in] meta Unpacked action meta-attributes * * \return Resource role that results from failure of action */ enum rsc_role_e pcmk__role_after_failure(const pcmk_resource_t *rsc, const char *action_name, enum pcmk__on_fail on_fail, GHashTable *meta) { enum rsc_role_e role = pcmk_role_unknown; // Set default for role after failure specially in certain circumstances switch (on_fail) { case pcmk__on_fail_stop: role = pcmk_role_stopped; break; case pcmk__on_fail_reset_remote: if (rsc->priv->remote_reconnect_ms != 0U) { role = pcmk_role_stopped; } break; default: break; } if (role == pcmk_role_unknown) { // Use default if (pcmk__str_eq(action_name, PCMK_ACTION_PROMOTE, pcmk__str_none)) { role = pcmk_role_unpromoted; } else { role = pcmk_role_started; } } pcmk__rsc_trace(rsc, "Role after %s %s failure is: %s", rsc->id, action_name, pcmk_role_text(role)); return role; } /*! * \internal * \brief Unpack action configuration * * Unpack a resource action's meta-attributes (normalizing the interval, * timeout, and start delay values as integer milliseconds), requirements, and * failure policy from its CIB XML configuration (including defaults). * * \param[in,out] action Resource action to unpack into * \param[in] xml_obj Action configuration XML (NULL for defaults only) * \param[in] interval_ms How frequently to perform the operation */ static void unpack_operation(pcmk_action_t *action, const xmlNode *xml_obj, guint interval_ms) { const char *value = NULL; action->meta = pcmk__unpack_action_meta(action->rsc, action->node, action->task, interval_ms, xml_obj); action->needs = pcmk__action_requires(action->rsc, action->task); value = g_hash_table_lookup(action->meta, PCMK_META_ON_FAIL); action->on_fail = pcmk__parse_on_fail(action->rsc, action->task, interval_ms, value); action->fail_role = pcmk__role_after_failure(action->rsc, action->task, action->on_fail, action->meta); } /*! * \brief Create or update an action object * * \param[in,out] rsc Resource that action is for (if any) * \param[in,out] key Action key (must be non-NULL) * \param[in] task Action name (must be non-NULL) * \param[in] on_node Node that action is on (if any) * \param[in] optional Whether action should be considered optional * \param[in,out] scheduler Scheduler data * * \return Action object corresponding to arguments (guaranteed not to be * \c NULL) * \note This function takes ownership of (and might free) \p key, and * \p scheduler takes ownership of the returned action (the caller should * not free it). */ pcmk_action_t * custom_action(pcmk_resource_t *rsc, char *key, const char *task, const pcmk_node_t *on_node, gboolean optional, pcmk_scheduler_t *scheduler) { pcmk_action_t *action = NULL; pcmk__assert((key != NULL) && (task != NULL) && (scheduler != NULL)); action = find_existing_action(key, rsc, on_node, scheduler); if (action == NULL) { action = new_action(key, task, rsc, on_node, optional, scheduler); } else { free(key); } update_action_optional(action, optional); if (rsc != NULL) { /* An action can be initially created with a NULL node, and later have * the node added via find_existing_action() (above) -> find_actions(). * That is why the extra parameters are unpacked here rather than in * new_action(). */ if ((action->node != NULL) && (action->op_entry != NULL) && !pcmk_is_set(action->flags, pcmk__action_attrs_evaluated)) { GHashTable *attrs = action->node->priv->attrs; if (action->extra != NULL) { g_hash_table_destroy(action->extra); } action->extra = pcmk__unpack_action_rsc_params(action->op_entry, attrs, scheduler); pcmk__set_action_flags(action, pcmk__action_attrs_evaluated); } update_resource_action_runnable(action, scheduler); } if (action->extra == NULL) { action->extra = pcmk__strkey_table(free, free); } return action; } pcmk_action_t * get_pseudo_op(const char *name, pcmk_scheduler_t *scheduler) { pcmk_action_t *op = lookup_singleton(scheduler, name); if (op == NULL) { op = custom_action(NULL, strdup(name), name, NULL, TRUE, scheduler); pcmk__set_action_flags(op, pcmk__action_pseudo|pcmk__action_runnable); } return op; } static GList * find_unfencing_devices(GList *candidates, GList *matches) { for (GList *gIter = candidates; gIter != NULL; gIter = gIter->next) { pcmk_resource_t *candidate = gIter->data; if (candidate->priv->children != NULL) { matches = find_unfencing_devices(candidate->priv->children, matches); } else if (!pcmk_is_set(candidate->flags, pcmk__rsc_fence_device)) { continue; } else if (pcmk_is_set(candidate->flags, pcmk__rsc_needs_unfencing)) { matches = g_list_prepend(matches, candidate); } else if (pcmk__str_eq(g_hash_table_lookup(candidate->priv->meta, PCMK_STONITH_PROVIDES), PCMK_VALUE_UNFENCING, pcmk__str_casei)) { matches = g_list_prepend(matches, candidate); } } return matches; } static int node_priority_fencing_delay(const pcmk_node_t *node, const pcmk_scheduler_t *scheduler) { int member_count = 0; int online_count = 0; int top_priority = 0; int lowest_priority = 0; GList *gIter = NULL; // PCMK_OPT_PRIORITY_FENCING_DELAY is disabled if (scheduler->priv->priority_fencing_ms == 0U) { return 0; } /* No need to request a delay if the fencing target is not a normal cluster * member, for example if it's a remote node or a guest node. */ if (node->priv->variant != pcmk__node_variant_cluster) { return 0; } // No need to request a delay if the fencing target is in our partition if (node->details->online) { return 0; } for (gIter = scheduler->nodes; gIter != NULL; gIter = gIter->next) { pcmk_node_t *n = gIter->data; if (n->priv->variant != pcmk__node_variant_cluster) { continue; } member_count ++; if (n->details->online) { online_count++; } if (member_count == 1 || n->priv->priority > top_priority) { top_priority = n->priv->priority; } if (member_count == 1 || n->priv->priority < lowest_priority) { lowest_priority = n->priv->priority; } } // No need to delay if we have more than half of the cluster members if (online_count > member_count / 2) { return 0; } /* All the nodes have equal priority. * Any configured corresponding `pcmk_delay_base/max` will be applied. */ if (lowest_priority == top_priority) { return 0; } if (node->priv->priority < top_priority) { return 0; } return pcmk__timeout_ms2s(scheduler->priv->priority_fencing_ms); } pcmk_action_t * pe_fence_op(pcmk_node_t *node, const char *op, bool optional, const char *reason, bool priority_delay, pcmk_scheduler_t *scheduler) { char *op_key = NULL; pcmk_action_t *stonith_op = NULL; if(op == NULL) { op = scheduler->priv->fence_action; } op_key = crm_strdup_printf("%s-%s-%s", PCMK_ACTION_STONITH, node->priv->name, op); stonith_op = lookup_singleton(scheduler, op_key); if(stonith_op == NULL) { stonith_op = custom_action(NULL, op_key, PCMK_ACTION_STONITH, node, TRUE, scheduler); pcmk__insert_meta(stonith_op, PCMK__META_ON_NODE, node->priv->name); pcmk__insert_meta(stonith_op, PCMK__META_ON_NODE_UUID, node->priv->id); pcmk__insert_meta(stonith_op, PCMK__META_STONITH_ACTION, op); if (pcmk_is_set(scheduler->flags, pcmk__sched_enable_unfencing)) { /* Extra work to detect device changes */ GString *digests_all = g_string_sized_new(1024); GString *digests_secure = g_string_sized_new(1024); GList *matches = find_unfencing_devices(scheduler->priv->resources, NULL); for (GList *gIter = matches; gIter != NULL; gIter = gIter->next) { pcmk_resource_t *match = gIter->data; const char *agent = g_hash_table_lookup(match->priv->meta, PCMK_XA_TYPE); pcmk__op_digest_t *data = NULL; data = pe__compare_fencing_digest(match, agent, node, scheduler); if (data->rc == pcmk__digest_mismatch) { optional = FALSE; crm_notice("Unfencing node %s because the definition of " "%s changed", pcmk__node_name(node), match->id); if (!pcmk__is_daemon && (scheduler->priv->out != NULL)) { pcmk__output_t *out = scheduler->priv->out; out->info(out, "notice: Unfencing node %s because the " "definition of %s changed", pcmk__node_name(node), match->id); } } pcmk__g_strcat(digests_all, match->id, ":", agent, ":", data->digest_all_calc, ",", NULL); pcmk__g_strcat(digests_secure, match->id, ":", agent, ":", data->digest_secure_calc, ",", NULL); } pcmk__insert_dup(stonith_op->meta, PCMK__META_DIGESTS_ALL, digests_all->str); g_string_free(digests_all, TRUE); pcmk__insert_dup(stonith_op->meta, PCMK__META_DIGESTS_SECURE, digests_secure->str); g_string_free(digests_secure, TRUE); g_list_free(matches); } } else { free(op_key); } if ((scheduler->priv->priority_fencing_ms > 0U) /* It's a suitable case where PCMK_OPT_PRIORITY_FENCING_DELAY * applies. At least add PCMK_OPT_PRIORITY_FENCING_DELAY field as * an indicator. */ && (priority_delay /* The priority delay needs to be recalculated if this function has * been called by schedule_fencing_and_shutdowns() after node * priority has already been calculated by native_add_running(). */ || g_hash_table_lookup(stonith_op->meta, PCMK_OPT_PRIORITY_FENCING_DELAY) != NULL)) { /* Add PCMK_OPT_PRIORITY_FENCING_DELAY to the fencing op even if * it's 0 for the targeting node. So that it takes precedence over * any possible `pcmk_delay_base/max`. */ char *delay_s = pcmk__itoa(node_priority_fencing_delay(node, scheduler)); g_hash_table_insert(stonith_op->meta, strdup(PCMK_OPT_PRIORITY_FENCING_DELAY), delay_s); } if(optional == FALSE && pe_can_fence(scheduler, node)) { pcmk__clear_action_flags(stonith_op, pcmk__action_optional); pe_action_set_reason(stonith_op, reason, false); } else if(reason && stonith_op->reason == NULL) { stonith_op->reason = strdup(reason); } return stonith_op; } -void -pe_free_action(pcmk_action_t *action) -{ - if (action == NULL) { - return; - } - g_list_free_full(action->actions_before, free); - g_list_free_full(action->actions_after, free); - if (action->extra) { - g_hash_table_destroy(action->extra); - } - if (action->meta) { - g_hash_table_destroy(action->meta); - } - pcmk__free_node_copy(action->node); - free(action->cancel_task); - free(action->reason); - free(action->task); - free(action->uuid); - free(action); -} - enum pcmk__action_type get_complex_task(const pcmk_resource_t *rsc, const char *name) { enum pcmk__action_type task = pcmk__parse_action(name); if (pcmk__is_primitive(rsc)) { switch (task) { case pcmk__action_stopped: case pcmk__action_started: case pcmk__action_demoted: case pcmk__action_promoted: crm_trace("Folding %s back into its atomic counterpart for %s", name, rsc->id); --task; break; default: break; } } return task; } /*! * \internal * \brief Find first matching action in a list * * \param[in] input List of actions to search * \param[in] uuid If not NULL, action must have this UUID * \param[in] task If not NULL, action must have this action name * \param[in] on_node If not NULL, action must be on this node * * \return First action in list that matches criteria, or NULL if none */ pcmk_action_t * find_first_action(const GList *input, const char *uuid, const char *task, const pcmk_node_t *on_node) { CRM_CHECK(uuid || task, return NULL); for (const GList *gIter = input; gIter != NULL; gIter = gIter->next) { pcmk_action_t *action = (pcmk_action_t *) gIter->data; if (uuid != NULL && !pcmk__str_eq(uuid, action->uuid, pcmk__str_casei)) { continue; } else if (task != NULL && !pcmk__str_eq(task, action->task, pcmk__str_casei)) { continue; } else if (on_node == NULL) { return action; } else if (action->node == NULL) { continue; } else if (pcmk__same_node(on_node, action->node)) { return action; } } return NULL; } GList * find_actions(GList *input, const char *key, const pcmk_node_t *on_node) { GList *gIter = input; GList *result = NULL; CRM_CHECK(key != NULL, return NULL); for (; gIter != NULL; gIter = gIter->next) { pcmk_action_t *action = (pcmk_action_t *) gIter->data; if (!pcmk__str_eq(key, action->uuid, pcmk__str_casei)) { continue; } else if (on_node == NULL) { crm_trace("Action %s matches (ignoring node)", key); result = g_list_prepend(result, action); } else if (action->node == NULL) { crm_trace("Action %s matches (unallocated, assigning to %s)", key, pcmk__node_name(on_node)); action->node = pe__copy_node(on_node); result = g_list_prepend(result, action); } else if (pcmk__same_node(on_node, action->node)) { crm_trace("Action %s on %s matches", key, pcmk__node_name(on_node)); result = g_list_prepend(result, action); } } return result; } GList * find_actions_exact(GList *input, const char *key, const pcmk_node_t *on_node) { GList *result = NULL; CRM_CHECK(key != NULL, return NULL); if (on_node == NULL) { return NULL; } for (GList *gIter = input; gIter != NULL; gIter = gIter->next) { pcmk_action_t *action = (pcmk_action_t *) gIter->data; if ((action->node != NULL) && pcmk__str_eq(key, action->uuid, pcmk__str_casei) && pcmk__same_node(on_node, action->node)) { crm_trace("Action %s on %s matches", key, pcmk__node_name(on_node)); result = g_list_prepend(result, action); } } return result; } /*! * \brief Find all actions of given type for a resource * * \param[in] rsc Resource to search * \param[in] node Find only actions scheduled on this node * \param[in] task Action name to search for * \param[in] require_node If TRUE, NULL node or action node will not match * * \return List of actions found (or NULL if none) * \note If node is not NULL and require_node is FALSE, matching actions * without a node will be assigned to node. */ GList * pe__resource_actions(const pcmk_resource_t *rsc, const pcmk_node_t *node, const char *task, bool require_node) { GList *result = NULL; char *key = pcmk__op_key(rsc->id, task, 0); if (require_node) { result = find_actions_exact(rsc->priv->actions, key, node); } else { result = find_actions(rsc->priv->actions, key, node); } free(key); return result; } /*! * \internal * \brief Create an action reason string based on the action itself * * \param[in] action Action to create reason string for * \param[in] flag Action flag that was cleared * * \return Newly allocated string suitable for use as action reason * \note It is the caller's responsibility to free() the result. */ char * pe__action2reason(const pcmk_action_t *action, enum pcmk__action_flags flag) { const char *change = NULL; switch (flag) { case pcmk__action_runnable: change = "unrunnable"; break; case pcmk__action_migratable: change = "unmigrateable"; break; case pcmk__action_optional: change = "required"; break; default: // Bug: caller passed unsupported flag CRM_CHECK(change != NULL, change = ""); break; } return crm_strdup_printf("%s%s%s %s", change, (action->rsc == NULL)? "" : " ", (action->rsc == NULL)? "" : action->rsc->id, action->task); } void pe_action_set_reason(pcmk_action_t *action, const char *reason, bool overwrite) { if (action->reason != NULL && overwrite) { pcmk__rsc_trace(action->rsc, "Changing %s reason from '%s' to '%s'", action->uuid, action->reason, pcmk__s(reason, "(none)")); } else if (action->reason == NULL) { pcmk__rsc_trace(action->rsc, "Set %s reason to '%s'", action->uuid, pcmk__s(reason, "(none)")); } else { // crm_assert(action->reason != NULL && !overwrite); return; } pcmk__str_update(&action->reason, reason); } /*! * \internal * \brief Create an action to clear a resource's history from CIB * * \param[in,out] rsc Resource to clear * \param[in] node Node to clear history on */ void pe__clear_resource_history(pcmk_resource_t *rsc, const pcmk_node_t *node) { pcmk__assert((rsc != NULL) && (node != NULL)); custom_action(rsc, pcmk__op_key(rsc->id, PCMK_ACTION_LRM_DELETE, 0), PCMK_ACTION_LRM_DELETE, node, FALSE, rsc->priv->scheduler); } #define sort_return(an_int, why) do { \ free(a_uuid); \ free(b_uuid); \ crm_trace("%s (%d) %c %s (%d) : %s", \ a_xml_id, a_call_id, an_int>0?'>':an_int<0?'<':'=', \ b_xml_id, b_call_id, why); \ return an_int; \ } while(0) int pe__is_newer_op(const xmlNode *xml_a, const xmlNode *xml_b) { int a_call_id = -1; int b_call_id = -1; char *a_uuid = NULL; char *b_uuid = NULL; const char *a_xml_id = crm_element_value(xml_a, PCMK_XA_ID); const char *b_xml_id = crm_element_value(xml_b, PCMK_XA_ID); const char *a_node = crm_element_value(xml_a, PCMK__META_ON_NODE); const char *b_node = crm_element_value(xml_b, PCMK__META_ON_NODE); bool same_node = pcmk__str_eq(a_node, b_node, pcmk__str_casei); if (same_node && pcmk__str_eq(a_xml_id, b_xml_id, pcmk__str_none)) { /* We have duplicate PCMK__XE_LRM_RSC_OP entries in the status * section which is unlikely to be a good thing * - we can handle it easily enough, but we need to get * to the bottom of why it's happening. */ pcmk__config_err("Duplicate " PCMK__XE_LRM_RSC_OP " entries named %s", a_xml_id); sort_return(0, "duplicate"); } crm_element_value_int(xml_a, PCMK__XA_CALL_ID, &a_call_id); crm_element_value_int(xml_b, PCMK__XA_CALL_ID, &b_call_id); if (a_call_id == -1 && b_call_id == -1) { /* both are pending ops so it doesn't matter since * stops are never pending */ sort_return(0, "pending"); } else if (same_node && a_call_id >= 0 && a_call_id < b_call_id) { sort_return(-1, "call id"); } else if (same_node && b_call_id >= 0 && a_call_id > b_call_id) { sort_return(1, "call id"); } else if (a_call_id >= 0 && b_call_id >= 0 && (!same_node || a_call_id == b_call_id)) { /* The op and last_failed_op are the same. Order on * PCMK_XA_LAST_RC_CHANGE. */ time_t last_a = -1; time_t last_b = -1; crm_element_value_epoch(xml_a, PCMK_XA_LAST_RC_CHANGE, &last_a); crm_element_value_epoch(xml_b, PCMK_XA_LAST_RC_CHANGE, &last_b); crm_trace("rc-change: %lld vs %lld", (long long) last_a, (long long) last_b); if (last_a >= 0 && last_a < last_b) { sort_return(-1, "rc-change"); } else if (last_b >= 0 && last_a > last_b) { sort_return(1, "rc-change"); } sort_return(0, "rc-change"); } else { /* One of the inputs is a pending operation. * Attempt to use PCMK__XA_TRANSITION_MAGIC to determine its age relative * to the other. */ int a_id = -1; int b_id = -1; const char *a_magic = crm_element_value(xml_a, PCMK__XA_TRANSITION_MAGIC); const char *b_magic = crm_element_value(xml_b, PCMK__XA_TRANSITION_MAGIC); CRM_CHECK(a_magic != NULL && b_magic != NULL, sort_return(0, "No magic")); if (!decode_transition_magic(a_magic, &a_uuid, &a_id, NULL, NULL, NULL, NULL)) { sort_return(0, "bad magic a"); } if (!decode_transition_magic(b_magic, &b_uuid, &b_id, NULL, NULL, NULL, NULL)) { sort_return(0, "bad magic b"); } /* try to determine the relative age of the operation... * some pending operations (e.g. a start) may have been superseded * by a subsequent stop * * [a|b]_id == -1 means it's a shutdown operation and _always_ comes last */ if (!pcmk__str_eq(a_uuid, b_uuid, pcmk__str_casei) || a_id == b_id) { /* * some of the logic in here may be redundant... * * if the UUID from the TE doesn't match then one better * be a pending operation. * pending operations don't survive between elections and joins * because we query the LRM directly */ if (b_call_id == -1) { sort_return(-1, "transition + call"); } else if (a_call_id == -1) { sort_return(1, "transition + call"); } } else if ((a_id >= 0 && a_id < b_id) || b_id == -1) { sort_return(-1, "transition"); } else if ((b_id >= 0 && a_id > b_id) || a_id == -1) { sort_return(1, "transition"); } } /* we should never end up here */ CRM_CHECK(FALSE, sort_return(0, "default")); } gint sort_op_by_callid(gconstpointer a, gconstpointer b) { return pe__is_newer_op((const xmlNode *) a, (const xmlNode *) b); } /*! * \internal * \brief Create a new pseudo-action for a resource * * \param[in,out] rsc Resource to create action for * \param[in] task Action name * \param[in] optional Whether action should be considered optional * \param[in] runnable Whethe action should be considered runnable * * \return New action object corresponding to arguments */ pcmk_action_t * pe__new_rsc_pseudo_action(pcmk_resource_t *rsc, const char *task, bool optional, bool runnable) { pcmk_action_t *action = NULL; pcmk__assert((rsc != NULL) && (task != NULL)); action = custom_action(rsc, pcmk__op_key(rsc->id, task, 0), task, NULL, optional, rsc->priv->scheduler); pcmk__set_action_flags(action, pcmk__action_pseudo); if (runnable) { pcmk__set_action_flags(action, pcmk__action_runnable); } return action; } /*! * \internal * \brief Add the expected result to an action * * \param[in,out] action Action to add expected result to * \param[in] expected_result Expected result to add * * \note This is more efficient than calling pcmk__insert_meta(). */ void pe__add_action_expected_result(pcmk_action_t *action, int expected_result) { pcmk__assert((action != NULL) && (action->meta != NULL)); g_hash_table_insert(action->meta, pcmk__str_copy(PCMK__META_OP_TARGET_RC), pcmk__itoa(expected_result)); } diff --git a/lib/pengine/remote.c b/lib/pengine/remote.c index 395fa473e6..b3820a742d 100644 --- a/lib/pengine/remote.c +++ b/lib/pengine/remote.c @@ -1,242 +1,181 @@ /* * Copyright 2013-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include /*! * \internal * \brief Check whether a resource creates a guest node * * If a given resource contains a launched resource that is a remote connection, * return that launched resource (or NULL if none is found). * * \param[in] scheduler Scheduler data * \param[in] rsc Resource to check * * \return Launched remote connection, or NULL if none found */ pcmk_resource_t * pe__resource_contains_guest_node(const pcmk_scheduler_t *scheduler, const pcmk_resource_t *rsc) { if ((rsc != NULL) && (scheduler != NULL) && pcmk_is_set(scheduler->flags, pcmk__sched_have_remote_nodes)) { for (GList *gIter = rsc->priv->launched; gIter != NULL; gIter = gIter->next) { pcmk_resource_t *launched = gIter->data; if (pcmk_is_set(launched->flags, pcmk__rsc_is_remote_connection)) { return launched; } } } return NULL; } bool xml_contains_remote_node(xmlNode *xml) { const char *value = NULL; if (xml == NULL) { return false; } value = crm_element_value(xml, PCMK_XA_TYPE); if (!pcmk__str_eq(value, "remote", pcmk__str_casei)) { return false; } value = crm_element_value(xml, PCMK_XA_CLASS); if (!pcmk__str_eq(value, PCMK_RESOURCE_CLASS_OCF, pcmk__str_casei)) { return false; } value = crm_element_value(xml, PCMK_XA_PROVIDER); if (!pcmk__str_eq(value, "pacemaker", pcmk__str_casei)) { return false; } return true; } /*! * \internal * \brief Execute a supplied function for each guest node running on a host * * \param[in] scheduler Scheduler data * \param[in] host Host node to check * \param[in] helper Function to call for each guest node * \param[in,out] user_data Pointer to pass to helper function */ void pe_foreach_guest_node(const pcmk_scheduler_t *scheduler, const pcmk_node_t *host, void (*helper)(const pcmk_node_t*, void*), void *user_data) { GList *iter; CRM_CHECK(scheduler && host && host->details && helper, return); if (!pcmk_is_set(scheduler->flags, pcmk__sched_have_remote_nodes)) { return; } for (iter = host->details->running_rsc; iter != NULL; iter = iter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) iter->data; if (pcmk_is_set(rsc->flags, pcmk__rsc_is_remote_connection) && (rsc->priv->launcher != NULL)) { pcmk_node_t *guest_node = pcmk_find_node(scheduler, rsc->id); if (guest_node) { (*helper)(guest_node, user_data); } } } } /*! * \internal * \brief Create CIB XML for an implicit remote connection * * \param[in,out] parent If not \c NULL, use as parent XML element * \param[in] uname Name of Pacemaker Remote node * \param[in] container_id If not \c NULL, use this as connection container * \param[in] migrateable If not \c NULL, use as remote * \c PCMK_META_ALLOW_MIGRATE value * \param[in] is_managed If not \c NULL, use as remote * \c PCMK_META_IS_MANAGED value * \param[in] start_timeout If not \c NULL, use as remote connect timeout * \param[in] server If not \c NULL, use as \c PCMK_REMOTE_RA_ADDR * \param[in] port If not \c NULL, use as \c PCMK_REMOTE_RA_PORT * * \return Newly created XML */ xmlNode * pe_create_remote_xml(xmlNode *parent, const char *uname, const char *container_id, const char *migrateable, const char *is_managed, const char *start_timeout, const char *server, const char *port) { xmlNode *remote; xmlNode *xml_sub; remote = pcmk__xe_create(parent, PCMK_XE_PRIMITIVE); // Add identity crm_xml_add(remote, PCMK_XA_ID, uname); crm_xml_add(remote, PCMK_XA_CLASS, PCMK_RESOURCE_CLASS_OCF); crm_xml_add(remote, PCMK_XA_PROVIDER, "pacemaker"); crm_xml_add(remote, PCMK_XA_TYPE, "remote"); // Add meta-attributes xml_sub = pcmk__xe_create(remote, PCMK_XE_META_ATTRIBUTES); pcmk__xe_set_id(xml_sub, "%s-%s", uname, PCMK_XE_META_ATTRIBUTES); crm_create_nvpair_xml(xml_sub, NULL, PCMK__META_INTERNAL_RSC, PCMK_VALUE_TRUE); if (container_id) { crm_create_nvpair_xml(xml_sub, NULL, PCMK__META_CONTAINER, container_id); } if (migrateable) { crm_create_nvpair_xml(xml_sub, NULL, PCMK_META_ALLOW_MIGRATE, migrateable); } if (is_managed) { crm_create_nvpair_xml(xml_sub, NULL, PCMK_META_IS_MANAGED, is_managed); } // Add instance attributes if (port || server) { xml_sub = pcmk__xe_create(remote, PCMK_XE_INSTANCE_ATTRIBUTES); pcmk__xe_set_id(xml_sub, "%s-%s", uname, PCMK_XE_INSTANCE_ATTRIBUTES); if (server) { crm_create_nvpair_xml(xml_sub, NULL, PCMK_REMOTE_RA_ADDR, server); } if (port) { crm_create_nvpair_xml(xml_sub, NULL, PCMK_REMOTE_RA_PORT, port); } } // Add operations xml_sub = pcmk__xe_create(remote, PCMK_XE_OPERATIONS); crm_create_op_xml(xml_sub, uname, PCMK_ACTION_MONITOR, "30s", "30s"); if (start_timeout) { crm_create_op_xml(xml_sub, uname, PCMK_ACTION_START, "0", start_timeout); } return remote; } - -// History entry to be checked for fail count clearing -struct check_op { - const xmlNode *rsc_op; // History entry XML - pcmk_resource_t *rsc; // Known resource corresponding to history entry - pcmk_node_t *node; // Known node corresponding to history entry - enum pcmk__check_parameters check_type; // What needs checking -}; - -void -pe__add_param_check(const xmlNode *rsc_op, pcmk_resource_t *rsc, - pcmk_node_t *node, enum pcmk__check_parameters flag, - pcmk_scheduler_t *scheduler) -{ - struct check_op *check_op = NULL; - - CRM_CHECK(scheduler && rsc_op && rsc && node, return); - - check_op = pcmk__assert_alloc(1, sizeof(struct check_op)); - - crm_trace("Deferring checks of %s until after allocation", - pcmk__xe_id(rsc_op)); - check_op->rsc_op = rsc_op; - check_op->rsc = rsc; - check_op->node = node; - check_op->check_type = flag; - scheduler->priv->param_check = g_list_prepend(scheduler->priv->param_check, - check_op); -} - -/*! - * \internal - * \brief Call a function for each action to be checked for addr substitution - * - * \param[in,out] scheduler Scheduler data - * \param[in] cb Function to be called - */ -void -pe__foreach_param_check(pcmk_scheduler_t *scheduler, - void (*cb)(pcmk_resource_t*, pcmk_node_t*, - const xmlNode*, enum pcmk__check_parameters)) -{ - CRM_CHECK(scheduler && cb, return); - - for (GList *item = scheduler->priv->param_check; - item != NULL; item = item->next) { - struct check_op *check_op = item->data; - - cb(check_op->rsc, check_op->node, check_op->rsc_op, - check_op->check_type); - } -} - -void -pe__free_param_checks(pcmk_scheduler_t *scheduler) -{ - if ((scheduler != NULL) && (scheduler->priv->param_check != NULL)) { - g_list_free_full(scheduler->priv->param_check, free); - scheduler->priv->param_check = NULL; - } -} diff --git a/lib/pengine/status.c b/lib/pengine/status.c index 2e4deb07e1..beb92786e0 100644 --- a/lib/pengine/status.c +++ b/lib/pengine/status.c @@ -1,546 +1,354 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include -/*! - * \brief Create a new object to hold scheduler data - * - * \return New, initialized scheduler data on success, else NULL (and set errno) - * \note Only pcmk_scheduler_t objects created with this function (as opposed - * to statically declared or directly allocated) should be used with the - * functions in this library, to allow for future extensions to the - * data type. The caller is responsible for freeing the memory with - * pe_free_working_set() when the instance is no longer needed. - */ -pcmk_scheduler_t * -pe_new_working_set(void) -{ - pcmk_scheduler_t *scheduler = calloc(1, sizeof(pcmk_scheduler_t)); - - if (scheduler == NULL) { - return NULL; - } - scheduler->priv = calloc(1, sizeof(pcmk__scheduler_private_t)); - if (scheduler->priv == NULL) { - free(scheduler); - return NULL; - } - set_working_set_defaults(scheduler); - return scheduler; -} - -/*! - * \brief Free scheduler data - * - * \param[in,out] scheduler Scheduler data to free - */ -void -pe_free_working_set(pcmk_scheduler_t *scheduler) -{ - if (scheduler != NULL) { - pe_reset_working_set(scheduler); - free(scheduler->priv->local_node_name); - free(scheduler->priv); - free(scheduler); - } -} - #define XPATH_DEPRECATED_RULES \ "//" PCMK_XE_OP_DEFAULTS "//" PCMK_XE_EXPRESSION \ "|//" PCMK_XE_OP "//" PCMK_XE_EXPRESSION /*! * \internal * \brief Log a warning for deprecated rule syntax in operations * * \param[in] scheduler Scheduler data */ static void check_for_deprecated_rules(pcmk_scheduler_t *scheduler) { // @COMPAT Drop this function when support for the syntax is dropped xmlNode *deprecated = get_xpath_object(XPATH_DEPRECATED_RULES, scheduler->input, LOG_NEVER); if (deprecated != NULL) { pcmk__warn_once(pcmk__wo_op_attr_expr, "Support for rules with node attribute expressions in " PCMK_XE_OP " or " PCMK_XE_OP_DEFAULTS " is deprecated " "and will be dropped in a future release"); } } /* * Unpack everything * At the end you'll have: * - A list of nodes * - A list of resources (each with any dependencies on other resources) * - A list of constraints between resources and nodes * - A list of constraints between start/stop actions * - A list of nodes that need to be stonith'd * - A list of nodes that need to be shutdown * - A list of the possible stop/start actions (without dependencies) */ gboolean cluster_status(pcmk_scheduler_t * scheduler) { const char *new_version = NULL; xmlNode *section = NULL; if ((scheduler == NULL) || (scheduler->input == NULL)) { return FALSE; } new_version = crm_element_value(scheduler->input, PCMK_XA_CRM_FEATURE_SET); if (pcmk__check_feature_set(new_version) != pcmk_rc_ok) { pcmk__config_err("Can't process CIB with feature set '%s' greater than our own '%s'", new_version, CRM_FEATURE_SET); return FALSE; } crm_trace("Beginning unpack"); if (scheduler->priv->failed != NULL) { pcmk__xml_free(scheduler->priv->failed); } scheduler->priv->failed = pcmk__xe_create(NULL, "failed-ops"); if (scheduler->priv->now == NULL) { scheduler->priv->now = crm_time_new(NULL); } if (pcmk__xe_attr_is_true(scheduler->input, PCMK_XA_HAVE_QUORUM)) { pcmk__set_scheduler_flags(scheduler, pcmk__sched_quorate); } else { pcmk__clear_scheduler_flags(scheduler, pcmk__sched_quorate); } scheduler->priv->op_defaults = get_xpath_object("//" PCMK_XE_OP_DEFAULTS, scheduler->input, LOG_NEVER); check_for_deprecated_rules(scheduler); scheduler->priv->rsc_defaults = get_xpath_object("//" PCMK_XE_RSC_DEFAULTS, scheduler->input, LOG_NEVER); section = get_xpath_object("//" PCMK_XE_CRM_CONFIG, scheduler->input, LOG_TRACE); unpack_config(section, scheduler); if (!pcmk_any_flags_set(scheduler->flags, pcmk__sched_location_only|pcmk__sched_quorate) && (scheduler->no_quorum_policy != pcmk_no_quorum_ignore)) { pcmk__sched_warn(scheduler, "Fencing and resource management disabled " "due to lack of quorum"); } section = get_xpath_object("//" PCMK_XE_NODES, scheduler->input, LOG_TRACE); unpack_nodes(section, scheduler); section = get_xpath_object("//" PCMK_XE_RESOURCES, scheduler->input, LOG_TRACE); if (!pcmk_is_set(scheduler->flags, pcmk__sched_location_only)) { unpack_remote_nodes(section, scheduler); } unpack_resources(section, scheduler); section = get_xpath_object("//" PCMK_XE_FENCING_TOPOLOGY, scheduler->input, LOG_TRACE); pcmk__validate_fencing_topology(section); section = get_xpath_object("//" PCMK_XE_TAGS, scheduler->input, LOG_NEVER); unpack_tags(section, scheduler); if (!pcmk_is_set(scheduler->flags, pcmk__sched_location_only)) { section = get_xpath_object("//" PCMK_XE_STATUS, scheduler->input, LOG_TRACE); unpack_status(section, scheduler); } if (!pcmk_is_set(scheduler->flags, pcmk__sched_no_counts)) { for (GList *item = scheduler->priv->resources; item != NULL; item = item->next) { pcmk_resource_t *rsc = item->data; rsc->priv->fns->count(item->data); } crm_trace("Cluster resource count: %d (%d disabled, %d blocked)", scheduler->priv->ninstances, scheduler->priv->disabled_resources, scheduler->priv->blocked_resources); } if ((scheduler->priv->local_node_name != NULL) && (pcmk_find_node(scheduler, scheduler->priv->local_node_name) == NULL)) { crm_info("Creating a fake local node for %s", scheduler->priv->local_node_name); pe_create_node(scheduler->priv->local_node_name, scheduler->priv->local_node_name, NULL, 0, scheduler); } pcmk__set_scheduler_flags(scheduler, pcmk__sched_have_status); return TRUE; } -/*! - * \internal - * \brief Free a list of pcmk_resource_t - * - * \param[in,out] resources List to free - * - * \note When the scheduler's resource list is freed, that includes the original - * storage for the uname and id of any Pacemaker Remote nodes in the - * scheduler's node list, so take care not to use those afterward. - * \todo Refactor pcmk_node_t to strdup() the node name. - */ -static void -pe_free_resources(GList *resources) +pcmk_resource_t * +pe_find_resource(GList *rsc_list, const char *id) { - pcmk_resource_t *rsc = NULL; - GList *iterator = resources; + return pe_find_resource_with_flags(rsc_list, id, pcmk_rsc_match_history); +} - while (iterator != NULL) { - rsc = (pcmk_resource_t *) iterator->data; - iterator = iterator->next; - rsc->priv->fns->free(rsc); - } - if (resources != NULL) { - g_list_free(resources); +pcmk_resource_t * +pe_find_resource_with_flags(GList *rsc_list, const char *id, enum pe_find flags) +{ + GList *rIter = NULL; + + for (rIter = rsc_list; id && rIter; rIter = rIter->next) { + pcmk_resource_t *parent = rIter->data; + pcmk_resource_t *match = parent->priv->fns->find_rsc(parent, id, NULL, + flags); + + if (match != NULL) { + return match; + } } + crm_trace("No match for %s", id); + return NULL; } -static void -pe_free_actions(GList *actions) +/*! + * \brief Find a node by name or ID in a list of nodes + * + * \param[in] nodes List of nodes (as pcmk_node_t*) + * \param[in] id If not NULL, ID of node to find + * \param[in] node_name If not NULL, name of node to find + * + * \return Node from \p nodes that matches \p id if any, + * otherwise node from \p nodes that matches \p uname if any, + * otherwise NULL + */ +pcmk_node_t * +pe_find_node_any(const GList *nodes, const char *id, const char *uname) { - GList *iterator = actions; + pcmk_node_t *match = NULL; - while (iterator != NULL) { - pe_free_action(iterator->data); - iterator = iterator->next; + if (id != NULL) { + match = pe_find_node_id(nodes, id); } - if (actions != NULL) { - g_list_free(actions); + if ((match == NULL) && (uname != NULL)) { + match = pcmk__find_node_in_list(nodes, uname); } + return match; } -static void -pe_free_nodes(GList *nodes) +/*! + * \brief Find a node by ID in a list of nodes + * + * \param[in] nodes List of nodes (as pcmk_node_t*) + * \param[in] id ID of node to find + * + * \return Node from \p nodes that matches \p id if any, otherwise NULL + */ +pcmk_node_t * +pe_find_node_id(const GList *nodes, const char *id) { - for (GList *iterator = nodes; iterator != NULL; iterator = iterator->next) { - pcmk_node_t *node = (pcmk_node_t *) iterator->data; - - // Shouldn't be possible, but to be safe ... - if (node == NULL) { - continue; - } - if (node->details == NULL) { - free(node); - continue; - } + for (const GList *iter = nodes; iter != NULL; iter = iter->next) { + pcmk_node_t *node = (pcmk_node_t *) iter->data; - /* This is called after pe_free_resources(), which means that we can't - * use node->private->name for Pacemaker Remote nodes. + /* @TODO Whether node IDs should be considered case-sensitive should + * probably depend on the node type, so functionizing the comparison + * would be worthwhile */ - crm_trace("Freeing node %s", (pcmk__is_pacemaker_remote_node(node)? - "(guest or remote)" : pcmk__node_name(node))); - - if (node->priv->attrs != NULL) { - g_hash_table_destroy(node->priv->attrs); - } - if (node->priv->utilization != NULL) { - g_hash_table_destroy(node->priv->utilization); - } - if (node->priv->digest_cache != NULL) { - g_hash_table_destroy(node->priv->digest_cache); + if (pcmk__str_eq(node->priv->id, id, pcmk__str_casei)) { + return node; } - g_list_free(node->details->running_rsc); - g_list_free(node->priv->assigned_resources); - free(node->priv); - free(node->details); - free(node->assign); - free(node); - } - if (nodes != NULL) { - g_list_free(nodes); } + return NULL; } -static void -pe__free_ordering(GList *constraints) -{ - GList *iterator = constraints; - - while (iterator != NULL) { - pcmk__action_relation_t *order = iterator->data; +// Deprecated functions kept only for backward API compatibility +// LCOV_EXCL_START - iterator = iterator->next; +#include - free(order->task1); - free(order->task2); - free(order); - } - if (constraints != NULL) { - g_list_free(constraints); - } +/*! + * \brief Create a new object to hold scheduler data + * + * \return New, initialized scheduler data on success, else NULL (and set errno) + * \note Only pcmk_scheduler_t objects created with this function (as opposed + * to statically declared or directly allocated) should be used with the + * functions in this library, to allow for future extensions to the + * data type. The caller is responsible for freeing the memory with + * pcmk_free_scheduler() when the instance is no longer needed. + */ +pcmk_scheduler_t * +pe_new_working_set(void) +{ + return pcmk_new_scheduler(); } -static void -pe__free_location(GList *constraints) +void +pe_reset_working_set(pcmk_scheduler_t *scheduler) { - GList *iterator = constraints; - - while (iterator != NULL) { - pcmk__location_t *cons = iterator->data; - - iterator = iterator->next; - - g_list_free_full(cons->nodes, pcmk__free_node_copy); - free(cons->id); - free(cons); - } - if (constraints != NULL) { - g_list_free(constraints); + if (scheduler == NULL) { + return; } + pcmk_reset_scheduler(scheduler); } -/*! - * \brief Reset scheduler data to defaults without freeing it or constraints - * - * \param[in,out] scheduler Scheduler data to reset - * - * \deprecated This function is deprecated as part of the API; - * pe_reset_working_set() should be used instead. - */ void cleanup_calculations(pcmk_scheduler_t *scheduler) { if (scheduler == NULL) { return; } pcmk__clear_scheduler_flags(scheduler, pcmk__sched_have_status); if (scheduler->priv->options != NULL) { g_hash_table_destroy(scheduler->priv->options); } if (scheduler->priv->singletons != NULL) { g_hash_table_destroy(scheduler->priv->singletons); } if (scheduler->priv->ticket_constraints != NULL) { g_hash_table_destroy(scheduler->priv->ticket_constraints); } if (scheduler->priv->templates != NULL) { g_hash_table_destroy(scheduler->priv->templates); } if (scheduler->priv->tags != NULL) { g_hash_table_destroy(scheduler->priv->tags); } crm_trace("deleting resources"); - pe_free_resources(scheduler->priv->resources); + g_list_free_full(scheduler->priv->resources, pcmk__free_resource); crm_trace("deleting actions"); - pe_free_actions(scheduler->priv->actions); + g_list_free_full(scheduler->priv->actions, pcmk__free_action); crm_trace("deleting nodes"); - pe_free_nodes(scheduler->nodes); + g_list_free_full(scheduler->nodes, pcmk__free_node); + scheduler->nodes = NULL; - pe__free_param_checks(scheduler); + pcmk__free_param_checks(scheduler); g_list_free(scheduler->priv->stop_needed); crm_time_free(scheduler->priv->now); pcmk__xml_free(scheduler->input); pcmk__xml_free(scheduler->priv->failed); pcmk__xml_free(scheduler->priv->graph); set_working_set_defaults(scheduler); CRM_LOG_ASSERT((scheduler->priv->location_constraints == NULL) && (scheduler->priv->ordering_constraints == NULL)); } -/*! - * \brief Reset scheduler data to default state without freeing it - * - * \param[in,out] scheduler Scheduler data to reset - */ -void -pe_reset_working_set(pcmk_scheduler_t *scheduler) -{ - if (scheduler == NULL) { - return; - } - - crm_trace("Deleting %d ordering constraints", - g_list_length(scheduler->priv->ordering_constraints)); - pe__free_ordering(scheduler->priv->ordering_constraints); - scheduler->priv->ordering_constraints = NULL; - - crm_trace("Deleting %d location constraints", - g_list_length(scheduler->priv->location_constraints)); - pe__free_location(scheduler->priv->location_constraints); - scheduler->priv->location_constraints = NULL; - - crm_trace("Deleting %d colocation constraints", - g_list_length(scheduler->priv->colocation_constraints)); - g_list_free_full(scheduler->priv->colocation_constraints, free); - scheduler->priv->colocation_constraints = NULL; - - cleanup_calculations(scheduler); -} - void set_working_set_defaults(pcmk_scheduler_t *scheduler) { // These members must be preserved pcmk__scheduler_private_t *priv = scheduler->priv; pcmk__output_t *out = priv->out; char *local_node_name = scheduler->priv->local_node_name; // Wipe the main structs (any other members must have previously been freed) memset(scheduler, 0, sizeof(pcmk_scheduler_t)); memset(priv, 0, sizeof(pcmk__scheduler_private_t)); // Restore the members to preserve scheduler->priv = priv; scheduler->priv->out = out; scheduler->priv->local_node_name = local_node_name; // Set defaults for everything else - scheduler->priv->next_ordering_id = 1; - scheduler->priv->next_action_id = 1; - scheduler->no_quorum_policy = pcmk_no_quorum_stop; -#if PCMK__CONCURRENT_FENCING_DEFAULT_TRUE - pcmk__set_scheduler_flags(scheduler, - pcmk__sched_symmetric_cluster - |pcmk__sched_concurrent_fencing - |pcmk__sched_stop_removed_resources - |pcmk__sched_cancel_removed_actions); -#else - pcmk__set_scheduler_flags(scheduler, - pcmk__sched_symmetric_cluster - |pcmk__sched_stop_removed_resources - |pcmk__sched_cancel_removed_actions); -#endif + pcmk__set_scheduler_defaults(scheduler); } -pcmk_resource_t * -pe_find_resource(GList *rsc_list, const char *id) -{ - return pe_find_resource_with_flags(rsc_list, id, pcmk_rsc_match_history); -} - -pcmk_resource_t * -pe_find_resource_with_flags(GList *rsc_list, const char *id, enum pe_find flags) -{ - GList *rIter = NULL; - - for (rIter = rsc_list; id && rIter; rIter = rIter->next) { - pcmk_resource_t *parent = rIter->data; - pcmk_resource_t *match = parent->priv->fns->find_rsc(parent, id, NULL, - flags); - - if (match != NULL) { - return match; - } - } - crm_trace("No match for %s", id); - return NULL; -} - -/*! - * \brief Find a node by name or ID in a list of nodes - * - * \param[in] nodes List of nodes (as pcmk_node_t*) - * \param[in] id If not NULL, ID of node to find - * \param[in] node_name If not NULL, name of node to find - * - * \return Node from \p nodes that matches \p id if any, - * otherwise node from \p nodes that matches \p uname if any, - * otherwise NULL - */ -pcmk_node_t * -pe_find_node_any(const GList *nodes, const char *id, const char *uname) -{ - pcmk_node_t *match = NULL; - - if (id != NULL) { - match = pe_find_node_id(nodes, id); - } - if ((match == NULL) && (uname != NULL)) { - match = pcmk__find_node_in_list(nodes, uname); - } - return match; -} - -/*! - * \brief Find a node by ID in a list of nodes - * - * \param[in] nodes List of nodes (as pcmk_node_t*) - * \param[in] id ID of node to find - * - * \return Node from \p nodes that matches \p id if any, otherwise NULL - */ -pcmk_node_t * -pe_find_node_id(const GList *nodes, const char *id) +void +pe_free_working_set(pcmk_scheduler_t *scheduler) { - for (const GList *iter = nodes; iter != NULL; iter = iter->next) { - pcmk_node_t *node = (pcmk_node_t *) iter->data; - - /* @TODO Whether node IDs should be considered case-sensitive should - * probably depend on the node type, so functionizing the comparison - * would be worthwhile - */ - if (pcmk__str_eq(node->priv->id, id, pcmk__str_casei)) { - return node; - } - } - return NULL; + pcmk_free_scheduler(scheduler); } -// Deprecated functions kept only for backward API compatibility -// LCOV_EXCL_START - -#include - -/*! - * \brief Find a node by name in a list of nodes - * - * \param[in] nodes List of nodes (as pcmk_node_t*) - * \param[in] node_name Name of node to find - * - * \return Node from \p nodes that matches \p node_name if any, otherwise NULL - */ pcmk_node_t * pe_find_node(const GList *nodes, const char *node_name) { return pcmk__find_node_in_list(nodes, node_name); } // LCOV_EXCL_STOP // End deprecated API diff --git a/lib/pengine/tests/native/native_find_rsc_test.c b/lib/pengine/tests/native/native_find_rsc_test.c index 676d22a88d..070b355064 100644 --- a/lib/pengine/tests/native/native_find_rsc_test.c +++ b/lib/pengine/tests/native/native_find_rsc_test.c @@ -1,933 +1,933 @@ /* * Copyright 2022-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include xmlNode *input = NULL; pcmk_scheduler_t *scheduler = NULL; pcmk_node_t *cluster01, *cluster02, *httpd_bundle_0; pcmk_resource_t *exim_group, *inactive_group; pcmk_resource_t *promotable_clone, *inactive_clone; pcmk_resource_t *httpd_bundle, *mysql_clone_group; static int setup(void **state) { char *path = NULL; pcmk__xml_init(); path = crm_strdup_printf("%s/crm_mon.xml", getenv("PCMK_CTS_CLI_DIR")); input = pcmk__xml_read(path); free(path); if (input == NULL) { return 1; } - scheduler = pe_new_working_set(); - + scheduler = pcmk_new_scheduler(); if (scheduler == NULL) { return 1; } pcmk__set_scheduler_flags(scheduler, pcmk__sched_no_counts); scheduler->input = input; cluster_status(scheduler); /* Get references to the cluster nodes so we don't have to find them repeatedly. */ cluster01 = pcmk_find_node(scheduler, "cluster01"); cluster02 = pcmk_find_node(scheduler, "cluster02"); httpd_bundle_0 = pcmk_find_node(scheduler, "httpd-bundle-0"); /* Get references to several resources we use frequently. */ for (GList *iter = scheduler->priv->resources; iter != NULL; iter = iter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) iter->data; if (strcmp(rsc->id, "exim-group") == 0) { exim_group = rsc; } else if (strcmp(rsc->id, "httpd-bundle") == 0) { httpd_bundle = rsc; } else if (strcmp(rsc->id, "inactive-clone") == 0) { inactive_clone = rsc; } else if (strcmp(rsc->id, "inactive-group") == 0) { inactive_group = rsc; } else if (strcmp(rsc->id, "mysql-clone-group") == 0) { mysql_clone_group = rsc; } else if (strcmp(rsc->id, "promotable-clone") == 0) { promotable_clone = rsc; } } return 0; } static int -teardown(void **state) { - pe_free_working_set(scheduler); +teardown(void **state) +{ + pcmk_free_scheduler(scheduler); pcmk__xml_cleanup(); return 0; } static void bad_args(void **state) { pcmk_resource_t *rsc = g_list_first(scheduler->priv->resources)->data; char *id = rsc->id; char *name = NULL; assert_non_null(rsc); assert_null(native_find_rsc(NULL, "dummy", NULL, 0)); assert_null(native_find_rsc(rsc, NULL, NULL, 0)); /* No resources exist with these names. */ name = crm_strdup_printf("%sX", rsc->id); assert_null(native_find_rsc(rsc, name, NULL, 0)); free(name); name = crm_strdup_printf("x%s", rsc->id); assert_null(native_find_rsc(rsc, name, NULL, 0)); free(name); name = g_ascii_strup(rsc->id, -1); assert_null(native_find_rsc(rsc, name, NULL, 0)); g_free(name); /* Fails because resource ID is NULL. */ rsc->id = NULL; assert_null(native_find_rsc(rsc, id, NULL, 0)); rsc->id = id; } static void primitive_rsc(void **state) { pcmk_resource_t *dummy = NULL; /* Find the "dummy" resource, which is the only one with that ID in the set. */ for (GList *iter = scheduler->priv->resources; iter != NULL; iter = iter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) iter->data; if (strcmp(rsc->id, "dummy") == 0) { dummy = rsc; break; } } assert_non_null(dummy); /* Passes because NULL was passed for node, regardless of flags. */ assert_ptr_equal(dummy, native_find_rsc(dummy, "dummy", NULL, 0)); assert_ptr_equal(dummy, native_find_rsc(dummy, "dummy", NULL, pcmk_rsc_match_current_node)); /* Fails because resource is not a clone (nor cloned). */ assert_null(native_find_rsc(dummy, "dummy", NULL, pcmk_rsc_match_clone_only)); assert_null(native_find_rsc(dummy, "dummy", cluster02, pcmk_rsc_match_clone_only)); /* Fails because dummy is not running on cluster01, even with the right flags. */ assert_null(native_find_rsc(dummy, "dummy", cluster01, pcmk_rsc_match_current_node)); // Fails because pcmk_rsc_match_current_node is required if a node is given assert_null(native_find_rsc(dummy, "dummy", cluster02, 0)); /* Passes because dummy is running on cluster02. */ assert_ptr_equal(dummy, native_find_rsc(dummy, "dummy", cluster02, pcmk_rsc_match_current_node)); } static void group_rsc(void **state) { assert_non_null(exim_group); /* Passes because NULL was passed for node, regardless of flags. */ assert_ptr_equal(exim_group, native_find_rsc(exim_group, "exim-group", NULL, 0)); assert_ptr_equal(exim_group, native_find_rsc(exim_group, "exim-group", NULL, pcmk_rsc_match_current_node)); /* Fails because resource is not a clone (nor cloned). */ assert_null(native_find_rsc(exim_group, "exim-group", NULL, pcmk_rsc_match_clone_only)); assert_null(native_find_rsc(exim_group, "exim-group", cluster01, pcmk_rsc_match_clone_only)); /* Fails because none of exim-group's children are running on cluster01, even with the right flags. */ assert_null(native_find_rsc(exim_group, "exim-group", cluster01, pcmk_rsc_match_current_node)); // Fails because pcmk_rsc_match_current_node is required if a node is given assert_null(native_find_rsc(exim_group, "exim-group", cluster01, 0)); /* Passes because one of exim-group's children is running on cluster02. */ assert_ptr_equal(exim_group, native_find_rsc(exim_group, "exim-group", cluster02, pcmk_rsc_match_current_node)); } static void inactive_group_rsc(void **state) { assert_non_null(inactive_group); /* Passes because NULL was passed for node, regardless of flags. */ assert_ptr_equal(inactive_group, native_find_rsc(inactive_group, "inactive-group", NULL, 0)); assert_ptr_equal(inactive_group, native_find_rsc(inactive_group, "inactive-group", NULL, pcmk_rsc_match_current_node)); /* Fails because resource is not a clone (nor cloned). */ assert_null(native_find_rsc(inactive_group, "inactive-group", NULL, pcmk_rsc_match_clone_only)); assert_null(native_find_rsc(inactive_group, "inactive-group", cluster01, pcmk_rsc_match_clone_only)); /* Fails because none of inactive-group's children are running. */ assert_null(native_find_rsc(inactive_group, "inactive-group", cluster01, pcmk_rsc_match_current_node)); assert_null(native_find_rsc(inactive_group, "inactive-group", cluster02, pcmk_rsc_match_current_node)); } static void group_member_rsc(void **state) { pcmk_resource_t *public_ip = NULL; /* Find the "Public-IP" resource, a member of "exim-group". */ for (GList *iter = exim_group->priv->children; iter != NULL; iter = iter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) iter->data; if (strcmp(rsc->id, "Public-IP") == 0) { public_ip = rsc; break; } } assert_non_null(public_ip); /* Passes because NULL was passed for node, regardless of flags. */ assert_ptr_equal(public_ip, native_find_rsc(public_ip, "Public-IP", NULL, 0)); assert_ptr_equal(public_ip, native_find_rsc(public_ip, "Public-IP", NULL, pcmk_rsc_match_current_node)); /* Fails because resource is not a clone (nor cloned). */ assert_null(native_find_rsc(public_ip, "Public-IP", NULL, pcmk_rsc_match_clone_only)); assert_null(native_find_rsc(public_ip, "Public-IP", cluster02, pcmk_rsc_match_clone_only)); /* Fails because Public-IP is not running on cluster01, even with the right flags. */ assert_null(native_find_rsc(public_ip, "Public-IP", cluster01, pcmk_rsc_match_current_node)); // Fails because pcmk_rsc_match_current_node is required if a node is given assert_null(native_find_rsc(public_ip, "Public-IP", cluster02, 0)); /* Passes because Public-IP is running on cluster02. */ assert_ptr_equal(public_ip, native_find_rsc(public_ip, "Public-IP", cluster02, pcmk_rsc_match_current_node)); } static void inactive_group_member_rsc(void **state) { pcmk_resource_t *inactive_dummy_1 = NULL; /* Find the "inactive-dummy-1" resource, a member of "inactive-group". */ for (GList *iter = inactive_group->priv->children; iter != NULL; iter = iter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) iter->data; if (strcmp(rsc->id, "inactive-dummy-1") == 0) { inactive_dummy_1 = rsc; break; } } assert_non_null(inactive_dummy_1); /* Passes because NULL was passed for node, regardless of flags. */ assert_ptr_equal(inactive_dummy_1, native_find_rsc(inactive_dummy_1, "inactive-dummy-1", NULL, 0)); assert_ptr_equal(inactive_dummy_1, native_find_rsc(inactive_dummy_1, "inactive-dummy-1", NULL, pcmk_rsc_match_current_node)); /* Fails because resource is not a clone (nor cloned). */ assert_null(native_find_rsc(inactive_dummy_1, "inactive-dummy-1", NULL, pcmk_rsc_match_clone_only)); assert_null(native_find_rsc(inactive_dummy_1, "inactive-dummy-1", cluster01, pcmk_rsc_match_clone_only)); /* Fails because inactive-dummy-1 is not running. */ assert_null(native_find_rsc(inactive_dummy_1, "inactive-dummy-1", cluster01, pcmk_rsc_match_current_node)); assert_null(native_find_rsc(inactive_dummy_1, "inactive-dummy-1", cluster02, pcmk_rsc_match_current_node)); } static void clone_rsc(void **state) { assert_non_null(promotable_clone); /* Passes because NULL was passed for node, regardless of flags. */ assert_ptr_equal(promotable_clone, native_find_rsc(promotable_clone, "promotable-clone", NULL, 0)); assert_ptr_equal(promotable_clone, native_find_rsc(promotable_clone, "promotable-clone", NULL, pcmk_rsc_match_current_node)); assert_ptr_equal(promotable_clone, native_find_rsc(promotable_clone, "promotable-clone", NULL, pcmk_rsc_match_clone_only)); // Fails because pcmk_rsc_match_current_node is required if a node is given assert_null(native_find_rsc(promotable_clone, "promotable-clone", cluster01, 0)); /* Passes because one of ping-clone's children is running on cluster01. */ assert_ptr_equal(promotable_clone, native_find_rsc(promotable_clone, "promotable-clone", cluster01, pcmk_rsc_match_current_node)); // Fails because pcmk_rsc_match_current_node is required if a node is given assert_null(native_find_rsc(promotable_clone, "promotable-clone", cluster02, 0)); /* Passes because one of ping_clone's children is running on cluster02. */ assert_ptr_equal(promotable_clone, native_find_rsc(promotable_clone, "promotable-clone", cluster02, pcmk_rsc_match_current_node)); // Passes for previous reasons, plus includes pcmk_rsc_match_clone_only assert_ptr_equal(promotable_clone, native_find_rsc(promotable_clone, "promotable-clone", cluster01, pcmk_rsc_match_clone_only |pcmk_rsc_match_current_node)); assert_ptr_equal(promotable_clone, native_find_rsc(promotable_clone, "promotable-clone", cluster02, pcmk_rsc_match_clone_only |pcmk_rsc_match_current_node)); } static void inactive_clone_rsc(void **state) { assert_non_null(inactive_clone); /* Passes because NULL was passed for node, regardless of flags. */ assert_ptr_equal(inactive_clone, native_find_rsc(inactive_clone, "inactive-clone", NULL, 0)); assert_ptr_equal(inactive_clone, native_find_rsc(inactive_clone, "inactive-clone", NULL, pcmk_rsc_match_current_node)); assert_ptr_equal(inactive_clone, native_find_rsc(inactive_clone, "inactive-clone", NULL, pcmk_rsc_match_clone_only)); /* Fails because none of inactive-clone's children are running. */ assert_null(native_find_rsc(inactive_clone, "inactive-clone", cluster01, pcmk_rsc_match_current_node |pcmk_rsc_match_clone_only)); assert_null(native_find_rsc(inactive_clone, "inactive-clone", cluster02, pcmk_rsc_match_current_node |pcmk_rsc_match_clone_only)); } static void clone_instance_rsc(void **state) { pcmk_resource_t *promotable_0 = NULL; pcmk_resource_t *promotable_1 = NULL; /* Find the "promotable-rsc:0" and "promotable-rsc:1" resources, members of "promotable-clone". */ for (GList *iter = promotable_clone->priv->children; iter != NULL; iter = iter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) iter->data; if (strcmp(rsc->id, "promotable-rsc:0") == 0) { promotable_0 = rsc; } else if (strcmp(rsc->id, "promotable-rsc:1") == 0) { promotable_1 = rsc; } } assert_non_null(promotable_0); assert_non_null(promotable_1); /* Passes because NULL was passed for node, regardless of flags. */ assert_ptr_equal(promotable_0, native_find_rsc(promotable_0, "promotable-rsc:0", NULL, 0)); assert_ptr_equal(promotable_0, native_find_rsc(promotable_0, "promotable-rsc:0", NULL, pcmk_rsc_match_current_node)); assert_ptr_equal(promotable_1, native_find_rsc(promotable_1, "promotable-rsc:1", NULL, 0)); assert_ptr_equal(promotable_1, native_find_rsc(promotable_1, "promotable-rsc:1", NULL, pcmk_rsc_match_current_node)); // Fails because pcmk_rsc_match_current_node is required if a node is given assert_null(native_find_rsc(promotable_0, "promotable-rsc:0", cluster02, 0)); assert_null(native_find_rsc(promotable_1, "promotable-rsc:1", cluster01, 0)); /* Check that the resource is running on the node we expect. */ assert_ptr_equal(promotable_0, native_find_rsc(promotable_0, "promotable-rsc:0", cluster02, pcmk_rsc_match_current_node)); assert_null(native_find_rsc(promotable_0, "promotable-rsc:0", cluster01, pcmk_rsc_match_current_node)); assert_ptr_equal(promotable_1, native_find_rsc(promotable_1, "promotable-rsc:1", cluster01, pcmk_rsc_match_current_node)); assert_null(native_find_rsc(promotable_1, "promotable-rsc:1", cluster02, pcmk_rsc_match_current_node)); /* Passes because NULL was passed for node and primitive name was given, with correct flags. */ assert_ptr_equal(promotable_0, native_find_rsc(promotable_0, "promotable-rsc", NULL, pcmk_rsc_match_clone_only)); // Passes because pcmk_rsc_match_basename matches any instance's base name assert_ptr_equal(promotable_0, native_find_rsc(promotable_0, "promotable-rsc", NULL, pcmk_rsc_match_basename)); assert_ptr_equal(promotable_1, native_find_rsc(promotable_1, "promotable-rsc", NULL, pcmk_rsc_match_basename)); // Passes because pcmk_rsc_match_anon_basename matches assert_ptr_equal(promotable_0, native_find_rsc(promotable_0, "promotable-rsc", NULL, pcmk_rsc_match_anon_basename)); assert_ptr_equal(promotable_1, native_find_rsc(promotable_1, "promotable-rsc", NULL, pcmk_rsc_match_anon_basename)); /* Check that the resource is running on the node we expect. */ assert_ptr_equal(promotable_0, native_find_rsc(promotable_0, "promotable-rsc", cluster02, pcmk_rsc_match_basename |pcmk_rsc_match_current_node)); assert_ptr_equal(promotable_0, native_find_rsc(promotable_0, "promotable-rsc", cluster02, pcmk_rsc_match_anon_basename |pcmk_rsc_match_current_node)); assert_null(native_find_rsc(promotable_0, "promotable-rsc", cluster01, pcmk_rsc_match_basename |pcmk_rsc_match_current_node)); assert_null(native_find_rsc(promotable_0, "promotable-rsc", cluster01, pcmk_rsc_match_anon_basename |pcmk_rsc_match_current_node)); assert_ptr_equal(promotable_1, native_find_rsc(promotable_1, "promotable-rsc", cluster01, pcmk_rsc_match_basename |pcmk_rsc_match_current_node)); assert_ptr_equal(promotable_1, native_find_rsc(promotable_1, "promotable-rsc", cluster01, pcmk_rsc_match_anon_basename |pcmk_rsc_match_current_node)); assert_null(native_find_rsc(promotable_1, "promotable-rsc", cluster02, pcmk_rsc_match_basename |pcmk_rsc_match_current_node)); assert_null(native_find_rsc(promotable_1, "promotable-rsc", cluster02, pcmk_rsc_match_anon_basename |pcmk_rsc_match_current_node)); /* Fails because incorrect flags were given along with primitive name. */ assert_null(native_find_rsc(promotable_0, "promotable-rsc", NULL, pcmk_rsc_match_current_node)); assert_null(native_find_rsc(promotable_1, "promotable-rsc", NULL, pcmk_rsc_match_current_node)); /* And then we check failure possibilities again, except passing promotable_clone * instead of promotable_X as the first argument to native_find_rsc. */ // Fails because pcmk_rsc_match_current_node is required if a node is given assert_null(native_find_rsc(promotable_clone, "promotable-rsc:0", cluster02, 0)); assert_null(native_find_rsc(promotable_clone, "promotable-rsc:1", cluster01, 0)); /* Check that the resource is running on the node we expect. */ assert_ptr_equal(promotable_0, native_find_rsc(promotable_clone, "promotable-rsc:0", cluster02, pcmk_rsc_match_current_node)); assert_ptr_equal(promotable_0, native_find_rsc(promotable_clone, "promotable-rsc", cluster02, pcmk_rsc_match_basename |pcmk_rsc_match_current_node)); assert_ptr_equal(promotable_0, native_find_rsc(promotable_clone, "promotable-rsc", cluster02, pcmk_rsc_match_anon_basename |pcmk_rsc_match_current_node)); assert_ptr_equal(promotable_1, native_find_rsc(promotable_clone, "promotable-rsc:1", cluster01, pcmk_rsc_match_current_node)); assert_ptr_equal(promotable_1, native_find_rsc(promotable_clone, "promotable-rsc", cluster01, pcmk_rsc_match_basename |pcmk_rsc_match_current_node)); assert_ptr_equal(promotable_1, native_find_rsc(promotable_clone, "promotable-rsc", cluster01, pcmk_rsc_match_anon_basename |pcmk_rsc_match_current_node)); } static void renamed_rsc(void **state) { pcmk_resource_t *promotable_0 = NULL; pcmk_resource_t *promotable_1 = NULL; /* Find the "promotable-rsc:0" and "promotable-rsc:1" resources, members of "promotable-clone". */ for (GList *iter = promotable_clone->priv->children; iter != NULL; iter = iter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) iter->data; if (strcmp(rsc->id, "promotable-rsc:0") == 0) { promotable_0 = rsc; } else if (strcmp(rsc->id, "promotable-rsc:1") == 0) { promotable_1 = rsc; } } assert_non_null(promotable_0); assert_non_null(promotable_1); // Passes because pcmk_rsc_match_history means base name matches history_id assert_ptr_equal(promotable_0, native_find_rsc(promotable_0, "promotable-rsc", NULL, pcmk_rsc_match_history)); assert_ptr_equal(promotable_1, native_find_rsc(promotable_1, "promotable-rsc", NULL, pcmk_rsc_match_history)); } static void bundle_rsc(void **state) { assert_non_null(httpd_bundle); /* Passes because NULL was passed for node, regardless of flags. */ assert_ptr_equal(httpd_bundle, native_find_rsc(httpd_bundle, "httpd-bundle", NULL, 0)); assert_ptr_equal(httpd_bundle, native_find_rsc(httpd_bundle, "httpd-bundle", NULL, pcmk_rsc_match_current_node)); /* Fails because resource is not a clone (nor cloned). */ assert_null(native_find_rsc(httpd_bundle, "httpd-bundle", NULL, pcmk_rsc_match_clone_only)); assert_null(native_find_rsc(httpd_bundle, "httpd-bundle", cluster01, pcmk_rsc_match_clone_only)); // Fails because pcmk_rsc_match_current_node is required if a node is given assert_null(native_find_rsc(httpd_bundle, "httpd-bundle", cluster01, 0)); /* Passes because one of httpd_bundle's children is running on cluster01. */ assert_ptr_equal(httpd_bundle, native_find_rsc(httpd_bundle, "httpd-bundle", cluster01, pcmk_rsc_match_current_node)); } static bool bundle_first_replica(pcmk__bundle_replica_t *replica, void *user_data) { pcmk_resource_t *ip_0 = replica->ip; pcmk_resource_t *child_0 = replica->child; pcmk_resource_t *container_0 = replica->container; pcmk_resource_t *remote_0 = replica->remote; assert_non_null(ip_0); assert_non_null(child_0); assert_non_null(container_0); assert_non_null(remote_0); /* Passes because NULL was passed for node, regardless of flags. */ assert_ptr_equal(ip_0, native_find_rsc(ip_0, "httpd-bundle-ip-192.168.122.131", NULL, 0)); assert_ptr_equal(child_0, native_find_rsc(child_0, "httpd:0", NULL, 0)); assert_ptr_equal(container_0, native_find_rsc(container_0, "httpd-bundle-docker-0", NULL, 0)); assert_ptr_equal(remote_0, native_find_rsc(remote_0, "httpd-bundle-0", NULL, 0)); // Fails because pcmk_rsc_match_current_node is required if a node is given assert_null(native_find_rsc(ip_0, "httpd-bundle-ip-192.168.122.131", cluster01, 0)); assert_null(native_find_rsc(child_0, "httpd:0", httpd_bundle_0, 0)); assert_null(native_find_rsc(container_0, "httpd-bundle-docker-0", cluster01, 0)); assert_null(native_find_rsc(remote_0, "httpd-bundle-0", cluster01, 0)); /* Check that the resource is running on the node we expect. */ assert_ptr_equal(ip_0, native_find_rsc(ip_0, "httpd-bundle-ip-192.168.122.131", cluster01, pcmk_rsc_match_current_node)); assert_null(native_find_rsc(ip_0, "httpd-bundle-ip-192.168.122.131", cluster02, pcmk_rsc_match_current_node)); assert_null(native_find_rsc(ip_0, "httpd-bundle-ip-192.168.122.131", httpd_bundle_0, pcmk_rsc_match_current_node)); assert_ptr_equal(child_0, native_find_rsc(child_0, "httpd:0", httpd_bundle_0, pcmk_rsc_match_current_node)); assert_null(native_find_rsc(child_0, "httpd:0", cluster01, pcmk_rsc_match_current_node)); assert_null(native_find_rsc(child_0, "httpd:0", cluster02, pcmk_rsc_match_current_node)); assert_ptr_equal(container_0, native_find_rsc(container_0, "httpd-bundle-docker-0", cluster01, pcmk_rsc_match_current_node)); assert_null(native_find_rsc(container_0, "httpd-bundle-docker-0", cluster02, pcmk_rsc_match_current_node)); assert_null(native_find_rsc(container_0, "httpd-bundle-docker-0", httpd_bundle_0, pcmk_rsc_match_current_node)); assert_ptr_equal(remote_0, native_find_rsc(remote_0, "httpd-bundle-0", cluster01, pcmk_rsc_match_current_node)); assert_null(native_find_rsc(remote_0, "httpd-bundle-0", cluster02, pcmk_rsc_match_current_node)); assert_null(native_find_rsc(remote_0, "httpd-bundle-0", httpd_bundle_0, pcmk_rsc_match_current_node)); // Passes because pcmk_rsc_match_basename matches any replica's base name assert_ptr_equal(child_0, native_find_rsc(child_0, "httpd", NULL, pcmk_rsc_match_basename)); // Passes because pcmk_rsc_match_anon_basename matches assert_ptr_equal(child_0, native_find_rsc(child_0, "httpd", NULL, pcmk_rsc_match_anon_basename)); /* Check that the resource is running on the node we expect. */ assert_ptr_equal(child_0, native_find_rsc(child_0, "httpd", httpd_bundle_0, pcmk_rsc_match_basename |pcmk_rsc_match_current_node)); assert_ptr_equal(child_0, native_find_rsc(child_0, "httpd", httpd_bundle_0, pcmk_rsc_match_anon_basename |pcmk_rsc_match_current_node)); assert_null(native_find_rsc(child_0, "httpd", cluster01, pcmk_rsc_match_basename |pcmk_rsc_match_current_node)); assert_null(native_find_rsc(child_0, "httpd", cluster01, pcmk_rsc_match_anon_basename |pcmk_rsc_match_current_node)); assert_null(native_find_rsc(child_0, "httpd", cluster02, pcmk_rsc_match_basename |pcmk_rsc_match_current_node)); assert_null(native_find_rsc(child_0, "httpd", cluster02, pcmk_rsc_match_anon_basename |pcmk_rsc_match_current_node)); /* Fails because incorrect flags were given along with base name. */ assert_null(native_find_rsc(child_0, "httpd", NULL, pcmk_rsc_match_current_node)); /* And then we check failure possibilities again, except passing httpd-bundle * instead of X_0 as the first argument to native_find_rsc. */ // Fails because pcmk_rsc_match_current_node is required if a node is given assert_null(native_find_rsc(httpd_bundle, "httpd-bundle-ip-192.168.122.131", cluster01, 0)); assert_null(native_find_rsc(httpd_bundle, "httpd:0", httpd_bundle_0, 0)); assert_null(native_find_rsc(httpd_bundle, "httpd-bundle-docker-0", cluster01, 0)); assert_null(native_find_rsc(httpd_bundle, "httpd-bundle-0", cluster01, 0)); /* Check that the resource is running on the node we expect. */ assert_ptr_equal(ip_0, native_find_rsc(httpd_bundle, "httpd-bundle-ip-192.168.122.131", cluster01, pcmk_rsc_match_current_node)); assert_ptr_equal(child_0, native_find_rsc(httpd_bundle, "httpd:0", httpd_bundle_0, pcmk_rsc_match_current_node)); assert_ptr_equal(container_0, native_find_rsc(httpd_bundle, "httpd-bundle-docker-0", cluster01, pcmk_rsc_match_current_node)); assert_ptr_equal(remote_0, native_find_rsc(httpd_bundle, "httpd-bundle-0", cluster01, pcmk_rsc_match_current_node)); return false; // Do not iterate through any further replicas } static void bundle_replica_rsc(void **state) { pe__foreach_bundle_replica(httpd_bundle, bundle_first_replica, NULL); } static void clone_group_rsc(void **rsc) { assert_non_null(mysql_clone_group); /* Passes because NULL was passed for node, regardless of flags. */ assert_ptr_equal(mysql_clone_group, native_find_rsc(mysql_clone_group, "mysql-clone-group", NULL, 0)); assert_ptr_equal(mysql_clone_group, native_find_rsc(mysql_clone_group, "mysql-clone-group", NULL, pcmk_rsc_match_current_node)); assert_ptr_equal(mysql_clone_group, native_find_rsc(mysql_clone_group, "mysql-clone-group", NULL, pcmk_rsc_match_clone_only)); // Fails because pcmk_rsc_match_current_node is required if a node is given assert_null(native_find_rsc(mysql_clone_group, "mysql-clone-group", cluster01, 0)); /* Passes because one of mysql-clone-group's children is running on cluster01. */ assert_ptr_equal(mysql_clone_group, native_find_rsc(mysql_clone_group, "mysql-clone-group", cluster01, pcmk_rsc_match_current_node)); // Fails because pcmk_rsc_match_current_node is required if a node is given assert_null(native_find_rsc(mysql_clone_group, "mysql-clone-group", cluster02, 0)); /* Passes because one of mysql-clone-group's children is running on cluster02. */ assert_ptr_equal(mysql_clone_group, native_find_rsc(mysql_clone_group, "mysql-clone-group", cluster02, pcmk_rsc_match_current_node)); // Passes for previous reasons, plus includes pcmk_rsc_match_clone_only assert_ptr_equal(mysql_clone_group, native_find_rsc(mysql_clone_group, "mysql-clone-group", cluster01, pcmk_rsc_match_clone_only |pcmk_rsc_match_current_node)); assert_ptr_equal(mysql_clone_group, native_find_rsc(mysql_clone_group, "mysql-clone-group", cluster02, pcmk_rsc_match_clone_only |pcmk_rsc_match_current_node)); } static void clone_group_instance_rsc(void **rsc) { pcmk_resource_t *mysql_group_0 = NULL; pcmk_resource_t *mysql_group_1 = NULL; /* Find the "mysql-group:0" and "mysql-group:1" resources, members of "mysql-clone-group". */ for (GList *iter = mysql_clone_group->priv->children; iter != NULL; iter = iter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) iter->data; if (strcmp(rsc->id, "mysql-group:0") == 0) { mysql_group_0 = rsc; } else if (strcmp(rsc->id, "mysql-group:1") == 0) { mysql_group_1 = rsc; } } assert_non_null(mysql_group_0); assert_non_null(mysql_group_1); /* Passes because NULL was passed for node, regardless of flags. */ assert_ptr_equal(mysql_group_0, native_find_rsc(mysql_group_0, "mysql-group:0", NULL, 0)); assert_ptr_equal(mysql_group_0, native_find_rsc(mysql_group_0, "mysql-group:0", NULL, pcmk_rsc_match_current_node)); assert_ptr_equal(mysql_group_1, native_find_rsc(mysql_group_1, "mysql-group:1", NULL, 0)); assert_ptr_equal(mysql_group_1, native_find_rsc(mysql_group_1, "mysql-group:1", NULL, pcmk_rsc_match_current_node)); // Fails because pcmk_rsc_match_current_node is required if a node is given assert_null(native_find_rsc(mysql_group_0, "mysql-group:0", cluster02, 0)); assert_null(native_find_rsc(mysql_group_1, "mysql-group:1", cluster01, 0)); /* Check that the resource is running on the node we expect. */ assert_ptr_equal(mysql_group_0, native_find_rsc(mysql_group_0, "mysql-group:0", cluster02, pcmk_rsc_match_current_node)); assert_null(native_find_rsc(mysql_group_0, "mysql-group:0", cluster01, pcmk_rsc_match_current_node)); assert_ptr_equal(mysql_group_1, native_find_rsc(mysql_group_1, "mysql-group:1", cluster01, pcmk_rsc_match_current_node)); assert_null(native_find_rsc(mysql_group_1, "mysql-group:1", cluster02, pcmk_rsc_match_current_node)); /* Passes because NULL was passed for node and base name was given, with correct flags. */ assert_ptr_equal(mysql_group_0, native_find_rsc(mysql_group_0, "mysql-group" , NULL, pcmk_rsc_match_clone_only)); // Passes because pcmk_rsc_match_basename matches any base name assert_ptr_equal(mysql_group_0, native_find_rsc(mysql_group_0, "mysql-group" , NULL, pcmk_rsc_match_basename)); assert_ptr_equal(mysql_group_1, native_find_rsc(mysql_group_1, "mysql-group" , NULL, pcmk_rsc_match_basename)); // Passes because pcmk_rsc_match_anon_basename matches assert_ptr_equal(mysql_group_0, native_find_rsc(mysql_group_0, "mysql-group" , NULL, pcmk_rsc_match_anon_basename)); assert_ptr_equal(mysql_group_1, native_find_rsc(mysql_group_1, "mysql-group" , NULL, pcmk_rsc_match_anon_basename)); /* Check that the resource is running on the node we expect. */ assert_ptr_equal(mysql_group_0, native_find_rsc(mysql_group_0, "mysql-group", cluster02, pcmk_rsc_match_basename |pcmk_rsc_match_current_node)); assert_ptr_equal(mysql_group_0, native_find_rsc(mysql_group_0, "mysql-group", cluster02, pcmk_rsc_match_anon_basename |pcmk_rsc_match_current_node)); assert_null(native_find_rsc(mysql_group_0, "mysql-group", cluster01, pcmk_rsc_match_basename |pcmk_rsc_match_current_node)); assert_null(native_find_rsc(mysql_group_0, "mysql-group", cluster01, pcmk_rsc_match_anon_basename |pcmk_rsc_match_current_node)); assert_ptr_equal(mysql_group_1, native_find_rsc(mysql_group_1, "mysql-group", cluster01, pcmk_rsc_match_basename |pcmk_rsc_match_current_node)); assert_ptr_equal(mysql_group_1, native_find_rsc(mysql_group_1, "mysql-group", cluster01, pcmk_rsc_match_anon_basename |pcmk_rsc_match_current_node)); assert_null(native_find_rsc(mysql_group_1, "mysql-group", cluster02, pcmk_rsc_match_basename |pcmk_rsc_match_current_node)); assert_null(native_find_rsc(mysql_group_1, "mysql-group", cluster02, pcmk_rsc_match_anon_basename |pcmk_rsc_match_current_node)); /* Fails because incorrect flags were given along with base name. */ assert_null(native_find_rsc(mysql_group_0, "mysql-group", NULL, pcmk_rsc_match_current_node)); assert_null(native_find_rsc(mysql_group_1, "mysql-group", NULL, pcmk_rsc_match_current_node)); /* And then we check failure possibilities again, except passing mysql_clone_group * instead of mysql_group_X as the first argument to native_find_rsc. */ // Fails because pcmk_rsc_match_current_node is required if a node is given assert_null(native_find_rsc(mysql_clone_group, "mysql-group:0", cluster02, 0)); assert_null(native_find_rsc(mysql_clone_group, "mysql-group:1", cluster01, 0)); /* Check that the resource is running on the node we expect. */ assert_ptr_equal(mysql_group_0, native_find_rsc(mysql_clone_group, "mysql-group:0", cluster02, pcmk_rsc_match_current_node)); assert_ptr_equal(mysql_group_0, native_find_rsc(mysql_clone_group, "mysql-group", cluster02, pcmk_rsc_match_basename |pcmk_rsc_match_current_node)); assert_ptr_equal(mysql_group_0, native_find_rsc(mysql_clone_group, "mysql-group", cluster02, pcmk_rsc_match_anon_basename |pcmk_rsc_match_current_node)); assert_ptr_equal(mysql_group_1, native_find_rsc(mysql_clone_group, "mysql-group:1", cluster01, pcmk_rsc_match_current_node)); assert_ptr_equal(mysql_group_1, native_find_rsc(mysql_clone_group, "mysql-group", cluster01, pcmk_rsc_match_basename |pcmk_rsc_match_current_node)); assert_ptr_equal(mysql_group_1, native_find_rsc(mysql_clone_group, "mysql-group", cluster01, pcmk_rsc_match_anon_basename |pcmk_rsc_match_current_node)); } static void clone_group_member_rsc(void **state) { pcmk_resource_t *mysql_proxy = NULL; /* Find the "mysql-proxy" resource, a member of "mysql-group". */ for (GList *iter = mysql_clone_group->priv->children; iter != NULL; iter = iter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) iter->data; if (strcmp(rsc->id, "mysql-group:0") == 0) { for (GList *iter2 = rsc->priv->children; iter2 != NULL; iter2 = iter2->next) { pcmk_resource_t *child = (pcmk_resource_t *) iter2->data; if (strcmp(child->id, "mysql-proxy:0") == 0) { mysql_proxy = child; break; } } break; } } assert_non_null(mysql_proxy); /* Passes because NULL was passed for node, regardless of flags. */ assert_ptr_equal(mysql_proxy, native_find_rsc(mysql_proxy, "mysql-proxy:0", NULL, 0)); assert_ptr_equal(mysql_proxy, native_find_rsc(mysql_proxy, "mysql-proxy:0", NULL, pcmk_rsc_match_current_node)); /* Passes because resource's parent is a clone. */ assert_ptr_equal(mysql_proxy, native_find_rsc(mysql_proxy, "mysql-proxy:0", NULL, pcmk_rsc_match_clone_only)); assert_ptr_equal(mysql_proxy, native_find_rsc(mysql_proxy, "mysql-proxy:0", cluster02, pcmk_rsc_match_clone_only |pcmk_rsc_match_current_node)); /* Fails because mysql-proxy:0 is not running on cluster01, even with the right flags. */ assert_null(native_find_rsc(mysql_proxy, "mysql-proxy:0", cluster01, pcmk_rsc_match_current_node)); // Fails because pcmk_rsc_match_current_node is required if a node is given assert_null(native_find_rsc(mysql_proxy, "mysql-proxy:0", cluster02, 0)); /* Passes because mysql-proxy:0 is running on cluster02. */ assert_ptr_equal(mysql_proxy, native_find_rsc(mysql_proxy, "mysql-proxy:0", cluster02, pcmk_rsc_match_current_node)); } /* TODO: Add tests for finding on assigned node (passing a node without * pcmk_rsc_match_current_node, after scheduling, for a resource that is * starting/stopping/moving. */ PCMK__UNIT_TEST(setup, teardown, cmocka_unit_test(bad_args), cmocka_unit_test(primitive_rsc), cmocka_unit_test(group_rsc), cmocka_unit_test(inactive_group_rsc), cmocka_unit_test(group_member_rsc), cmocka_unit_test(inactive_group_member_rsc), cmocka_unit_test(clone_rsc), cmocka_unit_test(inactive_clone_rsc), cmocka_unit_test(clone_instance_rsc), cmocka_unit_test(renamed_rsc), cmocka_unit_test(bundle_rsc), cmocka_unit_test(bundle_replica_rsc), cmocka_unit_test(clone_group_rsc), cmocka_unit_test(clone_group_instance_rsc), cmocka_unit_test(clone_group_member_rsc)) diff --git a/lib/pengine/tests/native/pe_base_name_eq_test.c b/lib/pengine/tests/native/pe_base_name_eq_test.c index 1a08480974..51e4af3787 100644 --- a/lib/pengine/tests/native/pe_base_name_eq_test.c +++ b/lib/pengine/tests/native/pe_base_name_eq_test.c @@ -1,159 +1,159 @@ /* * Copyright 2022-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include xmlNode *input = NULL; pcmk_scheduler_t *scheduler = NULL; pcmk_resource_t *exim_group, *promotable_0, *promotable_1, *dummy; pcmk_resource_t *httpd_bundle, *mysql_group_0, *mysql_group_1; static int setup(void **state) { char *path = NULL; pcmk__xml_init(); path = crm_strdup_printf("%s/crm_mon.xml", getenv("PCMK_CTS_CLI_DIR")); input = pcmk__xml_read(path); free(path); if (input == NULL) { return 1; } - scheduler = pe_new_working_set(); - + scheduler = pcmk_new_scheduler(); if (scheduler == NULL) { return 1; } pcmk__set_scheduler_flags(scheduler, pcmk__sched_no_counts); scheduler->input = input; cluster_status(scheduler); /* Get references to several resources we use frequently. */ for (GList *iter = scheduler->priv->resources; iter != NULL; iter = iter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) iter->data; if (strcmp(rsc->id, "dummy") == 0) { dummy = rsc; } else if (strcmp(rsc->id, "exim-group") == 0) { exim_group = rsc; } else if (strcmp(rsc->id, "httpd-bundle") == 0) { httpd_bundle = rsc; } else if (strcmp(rsc->id, "mysql-clone-group") == 0) { for (GList *iter = rsc->priv->children; iter != NULL; iter = iter->next) { pcmk_resource_t *child = (pcmk_resource_t *) iter->data; if (strcmp(child->id, "mysql-group:0") == 0) { mysql_group_0 = child; } else if (strcmp(child->id, "mysql-group:1") == 0) { mysql_group_1 = child; } } } else if (strcmp(rsc->id, "promotable-clone") == 0) { for (GList *iter = rsc->priv->children; iter != NULL; iter = iter->next) { pcmk_resource_t *child = (pcmk_resource_t *) iter->data; if (strcmp(child->id, "promotable-rsc:0") == 0) { promotable_0 = child; } else if (strcmp(child->id, "promotable-rsc:1") == 0) { promotable_1 = child; } } } } return 0; } static int -teardown(void **state) { - pe_free_working_set(scheduler); +teardown(void **state) +{ + pcmk_free_scheduler(scheduler); pcmk__xml_cleanup(); return 0; } static void bad_args(void **state) { char *id = dummy->id; assert_false(pe_base_name_eq(NULL, "dummy")); assert_false(pe_base_name_eq(dummy, NULL)); dummy->id = NULL; assert_false(pe_base_name_eq(dummy, "dummy")); dummy->id = id; } static void primitive_rsc(void **state) { assert_true(pe_base_name_eq(dummy, "dummy")); assert_false(pe_base_name_eq(dummy, "DUMMY")); assert_false(pe_base_name_eq(dummy, "dUmMy")); assert_false(pe_base_name_eq(dummy, "dummy0")); assert_false(pe_base_name_eq(dummy, "dummy:0")); } static void group_rsc(void **state) { assert_true(pe_base_name_eq(exim_group, "exim-group")); assert_false(pe_base_name_eq(exim_group, "EXIM-GROUP")); assert_false(pe_base_name_eq(exim_group, "exim-group0")); assert_false(pe_base_name_eq(exim_group, "exim-group:0")); assert_false(pe_base_name_eq(exim_group, "Public-IP")); } static void clone_rsc(void **state) { assert_true(pe_base_name_eq(promotable_0, "promotable-rsc")); assert_true(pe_base_name_eq(promotable_1, "promotable-rsc")); assert_false(pe_base_name_eq(promotable_0, "promotable-rsc:0")); assert_false(pe_base_name_eq(promotable_1, "promotable-rsc:1")); assert_false(pe_base_name_eq(promotable_0, "PROMOTABLE-RSC")); assert_false(pe_base_name_eq(promotable_1, "PROMOTABLE-RSC")); assert_false(pe_base_name_eq(promotable_0, "Promotable-rsc")); assert_false(pe_base_name_eq(promotable_1, "Promotable-rsc")); } static void bundle_rsc(void **state) { assert_true(pe_base_name_eq(httpd_bundle, "httpd-bundle")); assert_false(pe_base_name_eq(httpd_bundle, "HTTPD-BUNDLE")); assert_false(pe_base_name_eq(httpd_bundle, "httpd")); assert_false(pe_base_name_eq(httpd_bundle, "httpd-docker-0")); } PCMK__UNIT_TEST(setup, teardown, cmocka_unit_test(bad_args), cmocka_unit_test(primitive_rsc), cmocka_unit_test(group_rsc), cmocka_unit_test(clone_rsc), cmocka_unit_test(bundle_rsc)) diff --git a/lib/pengine/tests/status/Makefile.am b/lib/pengine/tests/status/Makefile.am index e95f51a74a..79c877db59 100644 --- a/lib/pengine/tests/status/Makefile.am +++ b/lib/pengine/tests/status/Makefile.am @@ -1,22 +1,20 @@ # # Copyright 2022-2024 the Pacemaker project contributors # # The version control history for this file may have further details. # # This source code is licensed under the GNU General Public License version 2 # or later (GPLv2+) WITHOUT ANY WARRANTY. # include $(top_srcdir)/mk/common.mk include $(top_srcdir)/mk/tap.mk include $(top_srcdir)/mk/unittest.mk LDADD += $(top_builddir)/lib/pengine/libpe_status_test.la # Add "_test" to the end of all test program names to simplify .gitignore. check_PROGRAMS = pe_find_node_any_test \ - pe_find_node_id_test \ - pe_new_working_set_test \ - set_working_set_defaults_test + pe_find_node_id_test TESTS = $(check_PROGRAMS) diff --git a/lib/pengine/unpack.c b/lib/pengine/unpack.c index 1fe58ab9c2..25857450f3 100644 --- a/lib/pengine/unpack.c +++ b/lib/pengine/unpack.c @@ -1,5099 +1,5099 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include #include #include CRM_TRACE_INIT_DATA(pe_status); // A (parsed) resource action history entry struct action_history { pcmk_resource_t *rsc; // Resource that history is for pcmk_node_t *node; // Node that history is for xmlNode *xml; // History entry XML // Parsed from entry XML const char *id; // XML ID of history entry const char *key; // Operation key of action const char *task; // Action name const char *exit_reason; // Exit reason given for result guint interval_ms; // Action interval int call_id; // Call ID of action int expected_exit_status; // Expected exit status of action int exit_status; // Actual exit status of action int execution_status; // Execution status of action }; /* This uses pcmk__set_flags_as()/pcmk__clear_flags_as() directly rather than * use pcmk__set_scheduler_flags()/pcmk__clear_scheduler_flags() so that the * flag is stringified more readably in log messages. */ #define set_config_flag(scheduler, option, flag) do { \ GHashTable *config_hash = (scheduler)->priv->options; \ const char *scf_value = pcmk__cluster_option(config_hash, (option)); \ \ if (scf_value != NULL) { \ if (crm_is_true(scf_value)) { \ (scheduler)->flags = pcmk__set_flags_as(__func__, __LINE__, \ LOG_TRACE, "Scheduler", \ crm_system_name, (scheduler)->flags, \ (flag), #flag); \ } else { \ (scheduler)->flags = pcmk__clear_flags_as(__func__, __LINE__, \ LOG_TRACE, "Scheduler", \ crm_system_name, (scheduler)->flags, \ (flag), #flag); \ } \ } \ } while(0) static void unpack_rsc_op(pcmk_resource_t *rsc, pcmk_node_t *node, xmlNode *xml_op, xmlNode **last_failure, enum pcmk__on_fail *failed); static void determine_remote_online_status(pcmk_scheduler_t *scheduler, pcmk_node_t *this_node); static void add_node_attrs(const xmlNode *xml_obj, pcmk_node_t *node, bool overwrite, pcmk_scheduler_t *scheduler); static void determine_online_status(const xmlNode *node_state, pcmk_node_t *this_node, pcmk_scheduler_t *scheduler); static void unpack_node_lrm(pcmk_node_t *node, const xmlNode *xml, pcmk_scheduler_t *scheduler); /*! * \internal * \brief Check whether a node is a dangling guest node * * \param[in] node Node to check * * \return true if \p node had a Pacemaker Remote connection resource with a * launcher that was removed from the CIB, otherwise false. */ static bool is_dangling_guest_node(pcmk_node_t *node) { return pcmk__is_pacemaker_remote_node(node) && (node->priv->remote != NULL) && (node->priv->remote->priv->launcher == NULL) && pcmk_is_set(node->priv->remote->flags, pcmk__rsc_removed_launched); } /*! * \brief Schedule a fence action for a node * * \param[in,out] scheduler Scheduler data * \param[in,out] node Node to fence * \param[in] reason Text description of why fencing is needed * \param[in] priority_delay Whether to consider * \c PCMK_OPT_PRIORITY_FENCING_DELAY */ void pe_fence_node(pcmk_scheduler_t *scheduler, pcmk_node_t *node, const char *reason, bool priority_delay) { CRM_CHECK(node, return); if (pcmk__is_guest_or_bundle_node(node)) { // Fence a guest or bundle node by marking its launcher as failed pcmk_resource_t *rsc = node->priv->remote->priv->launcher; if (!pcmk_is_set(rsc->flags, pcmk__rsc_failed)) { if (!pcmk_is_set(rsc->flags, pcmk__rsc_managed)) { crm_notice("Not fencing guest node %s " "(otherwise would because %s): " "its guest resource %s is unmanaged", pcmk__node_name(node), reason, rsc->id); } else { pcmk__sched_warn(scheduler, "Guest node %s will be fenced " "(by recovering its guest resource %s): %s", pcmk__node_name(node), rsc->id, reason); /* We don't mark the node as unclean because that would prevent the * node from running resources. We want to allow it to run resources * in this transition if the recovery succeeds. */ pcmk__set_node_flags(node, pcmk__node_remote_reset); pcmk__set_rsc_flags(rsc, pcmk__rsc_failed|pcmk__rsc_stop_if_failed); } } } else if (is_dangling_guest_node(node)) { crm_info("Cleaning up dangling connection for guest node %s: " "fencing was already done because %s, " "and guest resource no longer exists", pcmk__node_name(node), reason); pcmk__set_rsc_flags(node->priv->remote, pcmk__rsc_failed|pcmk__rsc_stop_if_failed); } else if (pcmk__is_remote_node(node)) { pcmk_resource_t *rsc = node->priv->remote; if ((rsc != NULL) && !pcmk_is_set(rsc->flags, pcmk__rsc_managed)) { crm_notice("Not fencing remote node %s " "(otherwise would because %s): connection is unmanaged", pcmk__node_name(node), reason); } else if (!pcmk_is_set(node->priv->flags, pcmk__node_remote_reset)) { pcmk__set_node_flags(node, pcmk__node_remote_reset); pcmk__sched_warn(scheduler, "Remote node %s %s: %s", pcmk__node_name(node), pe_can_fence(scheduler, node)? "will be fenced" : "is unclean", reason); } node->details->unclean = TRUE; // No need to apply PCMK_OPT_PRIORITY_FENCING_DELAY for remote nodes pe_fence_op(node, NULL, TRUE, reason, FALSE, scheduler); } else if (node->details->unclean) { crm_trace("Cluster node %s %s because %s", pcmk__node_name(node), pe_can_fence(scheduler, node)? "would also be fenced" : "also is unclean", reason); } else { pcmk__sched_warn(scheduler, "Cluster node %s %s: %s", pcmk__node_name(node), pe_can_fence(scheduler, node)? "will be fenced" : "is unclean", reason); node->details->unclean = TRUE; pe_fence_op(node, NULL, TRUE, reason, priority_delay, scheduler); } } // @TODO xpaths can't handle templates, rules, or id-refs // nvpair with provides or requires set to unfencing #define XPATH_UNFENCING_NVPAIR PCMK_XE_NVPAIR \ "[(@" PCMK_XA_NAME "='" PCMK_STONITH_PROVIDES "'" \ "or @" PCMK_XA_NAME "='" PCMK_META_REQUIRES "') " \ "and @" PCMK_XA_VALUE "='" PCMK_VALUE_UNFENCING "']" // unfencing in rsc_defaults or any resource #define XPATH_ENABLE_UNFENCING \ "/" PCMK_XE_CIB "/" PCMK_XE_CONFIGURATION "/" PCMK_XE_RESOURCES \ "//" PCMK_XE_META_ATTRIBUTES "/" XPATH_UNFENCING_NVPAIR \ "|/" PCMK_XE_CIB "/" PCMK_XE_CONFIGURATION "/" PCMK_XE_RSC_DEFAULTS \ "/" PCMK_XE_META_ATTRIBUTES "/" XPATH_UNFENCING_NVPAIR static void set_if_xpath(uint64_t flag, const char *xpath, pcmk_scheduler_t *scheduler) { xmlXPathObjectPtr result = NULL; if (!pcmk_is_set(scheduler->flags, flag)) { result = xpath_search(scheduler->input, xpath); if (result && (numXpathResults(result) > 0)) { pcmk__set_scheduler_flags(scheduler, flag); } freeXpathObject(result); } } gboolean unpack_config(xmlNode *config, pcmk_scheduler_t *scheduler) { const char *value = NULL; GHashTable *config_hash = pcmk__strkey_table(free, free); const pcmk_rule_input_t rule_input = { .now = scheduler->priv->now, }; scheduler->priv->options = config_hash; pe__unpack_dataset_nvpairs(config, PCMK_XE_CLUSTER_PROPERTY_SET, &rule_input, config_hash, PCMK_VALUE_CIB_BOOTSTRAP_OPTIONS, scheduler); pcmk__validate_cluster_options(config_hash); set_config_flag(scheduler, PCMK_OPT_ENABLE_STARTUP_PROBES, pcmk__sched_probe_resources); if (!pcmk_is_set(scheduler->flags, pcmk__sched_probe_resources)) { crm_info("Startup probes: disabled (dangerous)"); } value = pcmk__cluster_option(config_hash, PCMK_OPT_HAVE_WATCHDOG); if (value && crm_is_true(value)) { crm_info("Watchdog-based self-fencing will be performed via SBD if " "fencing is required and " PCMK_OPT_STONITH_WATCHDOG_TIMEOUT " is nonzero"); pcmk__set_scheduler_flags(scheduler, pcmk__sched_have_fencing); } /* Set certain flags via xpath here, so they can be used before the relevant * configuration sections are unpacked. */ set_if_xpath(pcmk__sched_enable_unfencing, XPATH_ENABLE_UNFENCING, scheduler); value = pcmk__cluster_option(config_hash, PCMK_OPT_STONITH_TIMEOUT); pcmk_parse_interval_spec(value, &(scheduler->priv->fence_timeout_ms)); crm_debug("Default fencing action timeout: %s", pcmk__readable_interval(scheduler->priv->fence_timeout_ms)); set_config_flag(scheduler, PCMK_OPT_STONITH_ENABLED, pcmk__sched_fencing_enabled); if (pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled)) { crm_debug("STONITH of failed nodes is enabled"); } else { crm_debug("STONITH of failed nodes is disabled"); } scheduler->priv->fence_action = pcmk__cluster_option(config_hash, PCMK_OPT_STONITH_ACTION); crm_trace("STONITH will %s nodes", scheduler->priv->fence_action); set_config_flag(scheduler, PCMK_OPT_CONCURRENT_FENCING, pcmk__sched_concurrent_fencing); if (pcmk_is_set(scheduler->flags, pcmk__sched_concurrent_fencing)) { crm_debug("Concurrent fencing is enabled"); } else { crm_debug("Concurrent fencing is disabled"); } value = pcmk__cluster_option(config_hash, PCMK_OPT_PRIORITY_FENCING_DELAY); if (value) { pcmk_parse_interval_spec(value, &(scheduler->priv->priority_fencing_ms)); crm_trace("Priority fencing delay is %s", pcmk__readable_interval(scheduler->priv->priority_fencing_ms)); } set_config_flag(scheduler, PCMK_OPT_STOP_ALL_RESOURCES, pcmk__sched_stop_all); crm_debug("Stop all active resources: %s", pcmk__flag_text(scheduler->flags, pcmk__sched_stop_all)); set_config_flag(scheduler, PCMK_OPT_SYMMETRIC_CLUSTER, pcmk__sched_symmetric_cluster); if (pcmk_is_set(scheduler->flags, pcmk__sched_symmetric_cluster)) { crm_debug("Cluster is symmetric" " - resources can run anywhere by default"); } value = pcmk__cluster_option(config_hash, PCMK_OPT_NO_QUORUM_POLICY); if (pcmk__str_eq(value, PCMK_VALUE_IGNORE, pcmk__str_casei)) { scheduler->no_quorum_policy = pcmk_no_quorum_ignore; } else if (pcmk__str_eq(value, PCMK_VALUE_FREEZE, pcmk__str_casei)) { scheduler->no_quorum_policy = pcmk_no_quorum_freeze; } else if (pcmk__str_eq(value, PCMK_VALUE_DEMOTE, pcmk__str_casei)) { scheduler->no_quorum_policy = pcmk_no_quorum_demote; } else if (pcmk__strcase_any_of(value, PCMK_VALUE_FENCE, PCMK_VALUE_FENCE_LEGACY, NULL)) { if (pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled)) { int do_panic = 0; crm_element_value_int(scheduler->input, PCMK_XA_NO_QUORUM_PANIC, &do_panic); if (do_panic || pcmk_is_set(scheduler->flags, pcmk__sched_quorate)) { scheduler->no_quorum_policy = pcmk_no_quorum_fence; } else { crm_notice("Resetting " PCMK_OPT_NO_QUORUM_POLICY " to 'stop': cluster has never had quorum"); scheduler->no_quorum_policy = pcmk_no_quorum_stop; } } else { pcmk__config_err("Resetting " PCMK_OPT_NO_QUORUM_POLICY " to 'stop' because fencing is disabled"); scheduler->no_quorum_policy = pcmk_no_quorum_stop; } } else { scheduler->no_quorum_policy = pcmk_no_quorum_stop; } switch (scheduler->no_quorum_policy) { case pcmk_no_quorum_freeze: crm_debug("On loss of quorum: Freeze resources"); break; case pcmk_no_quorum_stop: crm_debug("On loss of quorum: Stop ALL resources"); break; case pcmk_no_quorum_demote: crm_debug("On loss of quorum: " "Demote promotable resources and stop other resources"); break; case pcmk_no_quorum_fence: crm_notice("On loss of quorum: Fence all remaining nodes"); break; case pcmk_no_quorum_ignore: crm_notice("On loss of quorum: Ignore"); break; } set_config_flag(scheduler, PCMK_OPT_STOP_ORPHAN_RESOURCES, pcmk__sched_stop_removed_resources); if (pcmk_is_set(scheduler->flags, pcmk__sched_stop_removed_resources)) { crm_trace("Orphan resources are stopped"); } else { crm_trace("Orphan resources are ignored"); } set_config_flag(scheduler, PCMK_OPT_STOP_ORPHAN_ACTIONS, pcmk__sched_cancel_removed_actions); if (pcmk_is_set(scheduler->flags, pcmk__sched_cancel_removed_actions)) { crm_trace("Orphan resource actions are stopped"); } else { crm_trace("Orphan resource actions are ignored"); } set_config_flag(scheduler, PCMK_OPT_MAINTENANCE_MODE, pcmk__sched_in_maintenance); crm_trace("Maintenance mode: %s", pcmk__flag_text(scheduler->flags, pcmk__sched_in_maintenance)); set_config_flag(scheduler, PCMK_OPT_START_FAILURE_IS_FATAL, pcmk__sched_start_failure_fatal); if (pcmk_is_set(scheduler->flags, pcmk__sched_start_failure_fatal)) { crm_trace("Start failures are always fatal"); } else { crm_trace("Start failures are handled by failcount"); } if (pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled)) { set_config_flag(scheduler, PCMK_OPT_STARTUP_FENCING, pcmk__sched_startup_fencing); } if (pcmk_is_set(scheduler->flags, pcmk__sched_startup_fencing)) { crm_trace("Unseen nodes will be fenced"); } else { pcmk__warn_once(pcmk__wo_blind, "Blind faith: not fencing unseen nodes"); } pe__unpack_node_health_scores(scheduler); scheduler->priv->placement_strategy = pcmk__cluster_option(config_hash, PCMK_OPT_PLACEMENT_STRATEGY); crm_trace("Placement strategy: %s", scheduler->priv->placement_strategy); set_config_flag(scheduler, PCMK_OPT_SHUTDOWN_LOCK, pcmk__sched_shutdown_lock); if (pcmk_is_set(scheduler->flags, pcmk__sched_shutdown_lock)) { value = pcmk__cluster_option(config_hash, PCMK_OPT_SHUTDOWN_LOCK_LIMIT); pcmk_parse_interval_spec(value, &(scheduler->priv->shutdown_lock_ms)); crm_trace("Resources will be locked to nodes that were cleanly " "shut down (locks expire after %s)", pcmk__readable_interval(scheduler->priv->shutdown_lock_ms)); } else { crm_trace("Resources will not be locked to nodes that were cleanly " "shut down"); } value = pcmk__cluster_option(config_hash, PCMK_OPT_NODE_PENDING_TIMEOUT); pcmk_parse_interval_spec(value, &(scheduler->priv->node_pending_ms)); if (scheduler->priv->node_pending_ms == 0U) { crm_trace("Do not fence pending nodes"); } else { crm_trace("Fence pending nodes after %s", pcmk__readable_interval(scheduler->priv->node_pending_ms)); } return TRUE; } /*! * \internal * \brief Create a new node object in scheduler data * * \param[in] id ID of new node * \param[in] uname Name of new node * \param[in] type Type of new node * \param[in] score Score of new node * \param[in,out] scheduler Scheduler data * * \return Newly created node object * \note The returned object is part of the scheduler data and should not be * freed separately. */ pcmk_node_t * pe_create_node(const char *id, const char *uname, const char *type, int score, pcmk_scheduler_t *scheduler) { enum pcmk__node_variant variant = pcmk__node_variant_cluster; pcmk_node_t *new_node = NULL; if (pcmk_find_node(scheduler, uname) != NULL) { pcmk__config_warn("More than one node entry has name '%s'", uname); } if (pcmk__str_eq(type, PCMK_VALUE_MEMBER, pcmk__str_null_matches|pcmk__str_casei)) { variant = pcmk__node_variant_cluster; } else if (pcmk__str_eq(type, PCMK_VALUE_REMOTE, pcmk__str_casei)) { variant = pcmk__node_variant_remote; } else { pcmk__config_err("Ignoring node %s with unrecognized type '%s'", pcmk__s(uname, "without name"), type); return NULL; } new_node = calloc(1, sizeof(pcmk_node_t)); if (new_node == NULL) { pcmk__sched_err(scheduler, "Could not allocate memory for node %s", uname); return NULL; } new_node->assign = calloc(1, sizeof(struct pcmk__node_assignment)); new_node->details = calloc(1, sizeof(struct pcmk__node_details)); new_node->priv = calloc(1, sizeof(pcmk__node_private_t)); if ((new_node->assign == NULL) || (new_node->details == NULL) || (new_node->priv == NULL)) { free(new_node->assign); free(new_node->details); free(new_node->priv); free(new_node); pcmk__sched_err(scheduler, "Could not allocate memory for node %s", uname); return NULL; } crm_trace("Creating node for entry %s/%s", uname, id); new_node->assign->score = score; new_node->priv->id = id; new_node->priv->name = uname; new_node->priv->flags = pcmk__node_probes_allowed; new_node->details->online = FALSE; new_node->details->shutdown = FALSE; new_node->details->running_rsc = NULL; new_node->priv->scheduler = scheduler; new_node->priv->variant = variant; new_node->priv->attrs = pcmk__strkey_table(free, free); new_node->priv->utilization = pcmk__strkey_table(free, free); new_node->priv->digest_cache = pcmk__strkey_table(free, pe__free_digests); if (pcmk__is_pacemaker_remote_node(new_node)) { pcmk__insert_dup(new_node->priv->attrs, CRM_ATTR_KIND, "remote"); pcmk__set_scheduler_flags(scheduler, pcmk__sched_have_remote_nodes); } else { pcmk__insert_dup(new_node->priv->attrs, CRM_ATTR_KIND, "cluster"); } scheduler->nodes = g_list_insert_sorted(scheduler->nodes, new_node, pe__cmp_node_name); return new_node; } static const char * expand_remote_rsc_meta(xmlNode *xml_obj, xmlNode *parent, pcmk_scheduler_t *data) { xmlNode *attr_set = NULL; xmlNode *attr = NULL; const char *container_id = pcmk__xe_id(xml_obj); const char *remote_name = NULL; const char *remote_server = NULL; const char *remote_port = NULL; const char *connect_timeout = "60s"; const char *remote_allow_migrate=NULL; const char *is_managed = NULL; for (attr_set = pcmk__xe_first_child(xml_obj, PCMK_XE_META_ATTRIBUTES, NULL, NULL); attr_set != NULL; attr_set = pcmk__xe_next(attr_set, PCMK_XE_META_ATTRIBUTES)) { for (attr = pcmk__xe_first_child(attr_set, NULL, NULL, NULL); attr != NULL; attr = pcmk__xe_next(attr, NULL)) { const char *value = crm_element_value(attr, PCMK_XA_VALUE); const char *name = crm_element_value(attr, PCMK_XA_NAME); if (name == NULL) { // Sanity continue; } if (strcmp(name, PCMK_META_REMOTE_NODE) == 0) { remote_name = value; } else if (strcmp(name, PCMK_META_REMOTE_ADDR) == 0) { remote_server = value; } else if (strcmp(name, PCMK_META_REMOTE_PORT) == 0) { remote_port = value; } else if (strcmp(name, PCMK_META_REMOTE_CONNECT_TIMEOUT) == 0) { connect_timeout = value; } else if (strcmp(name, PCMK_META_REMOTE_ALLOW_MIGRATE) == 0) { remote_allow_migrate = value; } else if (strcmp(name, PCMK_META_IS_MANAGED) == 0) { is_managed = value; } } } if (remote_name == NULL) { return NULL; } if (pe_find_resource(data->priv->resources, remote_name) != NULL) { return NULL; } pe_create_remote_xml(parent, remote_name, container_id, remote_allow_migrate, is_managed, connect_timeout, remote_server, remote_port); return remote_name; } static void handle_startup_fencing(pcmk_scheduler_t *scheduler, pcmk_node_t *new_node) { if ((new_node->priv->variant == pcmk__node_variant_remote) && (new_node->priv->remote == NULL)) { /* Ignore fencing for remote nodes that don't have a connection resource * associated with them. This happens when remote node entries get left * in the nodes section after the connection resource is removed. */ return; } if (pcmk_is_set(scheduler->flags, pcmk__sched_startup_fencing)) { // All nodes are unclean until we've seen their status entry new_node->details->unclean = TRUE; } else { // Blind faith ... new_node->details->unclean = FALSE; } } gboolean unpack_nodes(xmlNode *xml_nodes, pcmk_scheduler_t *scheduler) { xmlNode *xml_obj = NULL; pcmk_node_t *new_node = NULL; const char *id = NULL; const char *uname = NULL; const char *type = NULL; for (xml_obj = pcmk__xe_first_child(xml_nodes, PCMK_XE_NODE, NULL, NULL); xml_obj != NULL; xml_obj = pcmk__xe_next(xml_obj, PCMK_XE_NODE)) { int score = 0; int rc = pcmk__xe_get_score(xml_obj, PCMK_XA_SCORE, &score, 0); new_node = NULL; id = crm_element_value(xml_obj, PCMK_XA_ID); uname = crm_element_value(xml_obj, PCMK_XA_UNAME); type = crm_element_value(xml_obj, PCMK_XA_TYPE); crm_trace("Processing node %s/%s", uname, id); if (id == NULL) { pcmk__config_err("Ignoring <" PCMK_XE_NODE "> entry in configuration without id"); continue; } if (rc != pcmk_rc_ok) { // Not possible with schema validation enabled pcmk__config_warn("Using 0 as score for node %s " "because '%s' is not a valid score: %s", pcmk__s(uname, "without name"), crm_element_value(xml_obj, PCMK_XA_SCORE), pcmk_rc_str(rc)); } new_node = pe_create_node(id, uname, type, score, scheduler); if (new_node == NULL) { return FALSE; } handle_startup_fencing(scheduler, new_node); add_node_attrs(xml_obj, new_node, FALSE, scheduler); crm_trace("Done with node %s", crm_element_value(xml_obj, PCMK_XA_UNAME)); } return TRUE; } static void unpack_launcher(pcmk_resource_t *rsc, pcmk_scheduler_t *scheduler) { const char *launcher_id = NULL; if (rsc->priv->children != NULL) { g_list_foreach(rsc->priv->children, (GFunc) unpack_launcher, scheduler); return; } launcher_id = g_hash_table_lookup(rsc->priv->meta, PCMK__META_CONTAINER); if ((launcher_id != NULL) && !pcmk__str_eq(launcher_id, rsc->id, pcmk__str_none)) { pcmk_resource_t *launcher = pe_find_resource(scheduler->priv->resources, launcher_id); if (launcher != NULL) { rsc->priv->launcher = launcher; launcher->priv->launched = g_list_append(launcher->priv->launched, rsc); pcmk__rsc_trace(rsc, "Resource %s's launcher is %s", rsc->id, launcher_id); } else { pcmk__config_err("Resource %s: Unknown " PCMK__META_CONTAINER " %s", rsc->id, launcher_id); } } } gboolean unpack_remote_nodes(xmlNode *xml_resources, pcmk_scheduler_t *scheduler) { xmlNode *xml_obj = NULL; /* Create remote nodes and guest nodes from the resource configuration * before unpacking resources. */ for (xml_obj = pcmk__xe_first_child(xml_resources, NULL, NULL, NULL); xml_obj != NULL; xml_obj = pcmk__xe_next(xml_obj, NULL)) { const char *new_node_id = NULL; /* Check for remote nodes, which are defined by ocf:pacemaker:remote * primitives. */ if (xml_contains_remote_node(xml_obj)) { new_node_id = pcmk__xe_id(xml_obj); /* The pcmk_find_node() check ensures we don't iterate over an * expanded node that has already been added to the node list */ if (new_node_id && (pcmk_find_node(scheduler, new_node_id) == NULL)) { crm_trace("Found remote node %s defined by resource %s", new_node_id, pcmk__xe_id(xml_obj)); pe_create_node(new_node_id, new_node_id, PCMK_VALUE_REMOTE, 0, scheduler); } continue; } /* Check for guest nodes, which are defined by special meta-attributes * of a primitive of any type (for example, VirtualDomain or Xen). */ if (pcmk__xe_is(xml_obj, PCMK_XE_PRIMITIVE)) { /* This will add an ocf:pacemaker:remote primitive to the * configuration for the guest node's connection, to be unpacked * later. */ new_node_id = expand_remote_rsc_meta(xml_obj, xml_resources, scheduler); if (new_node_id && (pcmk_find_node(scheduler, new_node_id) == NULL)) { crm_trace("Found guest node %s in resource %s", new_node_id, pcmk__xe_id(xml_obj)); pe_create_node(new_node_id, new_node_id, PCMK_VALUE_REMOTE, 0, scheduler); } continue; } /* Check for guest nodes inside a group. Clones are currently not * supported as guest nodes. */ if (pcmk__xe_is(xml_obj, PCMK_XE_GROUP)) { xmlNode *xml_obj2 = NULL; for (xml_obj2 = pcmk__xe_first_child(xml_obj, NULL, NULL, NULL); xml_obj2 != NULL; xml_obj2 = pcmk__xe_next(xml_obj2, NULL)) { new_node_id = expand_remote_rsc_meta(xml_obj2, xml_resources, scheduler); if (new_node_id && (pcmk_find_node(scheduler, new_node_id) == NULL)) { crm_trace("Found guest node %s in resource %s inside group %s", new_node_id, pcmk__xe_id(xml_obj2), pcmk__xe_id(xml_obj)); pe_create_node(new_node_id, new_node_id, PCMK_VALUE_REMOTE, 0, scheduler); } } } } return TRUE; } /* Call this after all the nodes and resources have been * unpacked, but before the status section is read. * * A remote node's online status is reflected by the state * of the remote node's connection resource. We need to link * the remote node to this connection resource so we can have * easy access to the connection resource during the scheduler calculations. */ static void link_rsc2remotenode(pcmk_scheduler_t *scheduler, pcmk_resource_t *new_rsc) { pcmk_node_t *remote_node = NULL; if (!pcmk_is_set(new_rsc->flags, pcmk__rsc_is_remote_connection)) { return; } if (pcmk_is_set(scheduler->flags, pcmk__sched_location_only)) { /* remote_nodes and remote_resources are not linked in quick location calculations */ return; } remote_node = pcmk_find_node(scheduler, new_rsc->id); CRM_CHECK(remote_node != NULL, return); pcmk__rsc_trace(new_rsc, "Linking remote connection resource %s to %s", new_rsc->id, pcmk__node_name(remote_node)); remote_node->priv->remote = new_rsc; if (new_rsc->priv->launcher == NULL) { /* Handle start-up fencing for remote nodes (as opposed to guest nodes) * the same as is done for cluster nodes. */ handle_startup_fencing(scheduler, remote_node); } else { /* pe_create_node() marks the new node as "remote" or "cluster"; now * that we know the node is a guest node, update it correctly. */ pcmk__insert_dup(remote_node->priv->attrs, CRM_ATTR_KIND, "container"); } } /*! * \internal * \brief Parse configuration XML for resource information * * \param[in] xml_resources Top of resource configuration XML * \param[in,out] scheduler Scheduler data * * \return TRUE * * \note unpack_remote_nodes() MUST be called before this, so that the nodes can * be used when pe__unpack_resource() calls resource_location() */ gboolean unpack_resources(const xmlNode *xml_resources, pcmk_scheduler_t *scheduler) { xmlNode *xml_obj = NULL; GList *gIter = NULL; scheduler->priv->templates = pcmk__strkey_table(free, pcmk__free_idref); for (xml_obj = pcmk__xe_first_child(xml_resources, NULL, NULL, NULL); xml_obj != NULL; xml_obj = pcmk__xe_next(xml_obj, NULL)) { pcmk_resource_t *new_rsc = NULL; const char *id = pcmk__xe_id(xml_obj); if (pcmk__str_empty(id)) { pcmk__config_err("Ignoring <%s> resource without ID", xml_obj->name); continue; } if (pcmk__xe_is(xml_obj, PCMK_XE_TEMPLATE)) { if (g_hash_table_lookup_extended(scheduler->priv->templates, id, NULL, NULL) == FALSE) { /* Record the template's ID for the knowledge of its existence anyway. */ pcmk__insert_dup(scheduler->priv->templates, id, NULL); } continue; } crm_trace("Unpacking <%s " PCMK_XA_ID "='%s'>", xml_obj->name, id); if (pe__unpack_resource(xml_obj, &new_rsc, NULL, scheduler) == pcmk_rc_ok) { scheduler->priv->resources = g_list_append(scheduler->priv->resources, new_rsc); pcmk__rsc_trace(new_rsc, "Added resource %s", new_rsc->id); } else { pcmk__config_err("Ignoring <%s> resource '%s' " "because configuration is invalid", xml_obj->name, id); } } for (gIter = scheduler->priv->resources; gIter != NULL; gIter = gIter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) gIter->data; unpack_launcher(rsc, scheduler); link_rsc2remotenode(scheduler, rsc); } scheduler->priv->resources = g_list_sort(scheduler->priv->resources, pe__cmp_rsc_priority); if (pcmk_is_set(scheduler->flags, pcmk__sched_location_only)) { /* Ignore */ } else if (pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled) && !pcmk_is_set(scheduler->flags, pcmk__sched_have_fencing)) { pcmk__config_err("Resource start-up disabled since no STONITH resources have been defined"); pcmk__config_err("Either configure some or disable STONITH with the " PCMK_OPT_STONITH_ENABLED " option"); pcmk__config_err("NOTE: Clusters with shared data need STONITH to ensure data integrity"); } return TRUE; } /*! * \internal * \brief Validate the levels in a fencing topology * * \param[in] xml \c PCMK_XE_FENCING_TOPOLOGY element */ void pcmk__validate_fencing_topology(const xmlNode *xml) { if (xml == NULL) { return; } CRM_CHECK(pcmk__xe_is(xml, PCMK_XE_FENCING_TOPOLOGY), return); for (const xmlNode *level = pcmk__xe_first_child(xml, PCMK_XE_FENCING_LEVEL, NULL, NULL); level != NULL; level = pcmk__xe_next(level, PCMK_XE_FENCING_LEVEL)) { const char *id = pcmk__xe_id(level); int index = 0; if (pcmk__str_empty(id)) { pcmk__config_err("Ignoring fencing level without ID"); continue; } if (crm_element_value_int(level, PCMK_XA_INDEX, &index) != 0) { pcmk__config_err("Ignoring fencing level %s with invalid index", id); continue; } if ((index < ST__LEVEL_MIN) || (index > ST__LEVEL_MAX)) { pcmk__config_err("Ignoring fencing level %s with out-of-range " "index %d", id, index); } } } gboolean unpack_tags(xmlNode *xml_tags, pcmk_scheduler_t *scheduler) { xmlNode *xml_tag = NULL; scheduler->priv->tags = pcmk__strkey_table(free, pcmk__free_idref); for (xml_tag = pcmk__xe_first_child(xml_tags, PCMK_XE_TAG, NULL, NULL); xml_tag != NULL; xml_tag = pcmk__xe_next(xml_tag, PCMK_XE_TAG)) { xmlNode *xml_obj_ref = NULL; const char *tag_id = pcmk__xe_id(xml_tag); if (tag_id == NULL) { pcmk__config_err("Ignoring <%s> without " PCMK_XA_ID, (const char *) xml_tag->name); continue; } for (xml_obj_ref = pcmk__xe_first_child(xml_tag, PCMK_XE_OBJ_REF, NULL, NULL); xml_obj_ref != NULL; xml_obj_ref = pcmk__xe_next(xml_obj_ref, PCMK_XE_OBJ_REF)) { const char *obj_ref = pcmk__xe_id(xml_obj_ref); if (obj_ref == NULL) { pcmk__config_err("Ignoring <%s> for tag '%s' without " PCMK_XA_ID, xml_obj_ref->name, tag_id); continue; } pcmk__add_idref(scheduler->priv->tags, tag_id, obj_ref); } } return TRUE; } /*! * \internal * \brief Unpack a ticket state entry * * \param[in] xml_ticket XML ticket state to unpack * \param[in,out] userdata Scheduler data * * \return pcmk_rc_ok (to always continue unpacking further entries) */ static int unpack_ticket_state(xmlNode *xml_ticket, void *userdata) { pcmk_scheduler_t *scheduler = userdata; const char *ticket_id = NULL; const char *granted = NULL; const char *last_granted = NULL; const char *standby = NULL; xmlAttrPtr xIter = NULL; pcmk__ticket_t *ticket = NULL; ticket_id = pcmk__xe_id(xml_ticket); if (pcmk__str_empty(ticket_id)) { return pcmk_rc_ok; } crm_trace("Processing ticket state for %s", ticket_id); ticket = g_hash_table_lookup(scheduler->priv->ticket_constraints, ticket_id); if (ticket == NULL) { ticket = ticket_new(ticket_id, scheduler); if (ticket == NULL) { return pcmk_rc_ok; } } for (xIter = xml_ticket->properties; xIter; xIter = xIter->next) { const char *prop_name = (const char *)xIter->name; const char *prop_value = pcmk__xml_attr_value(xIter); if (pcmk__str_eq(prop_name, PCMK_XA_ID, pcmk__str_none)) { continue; } pcmk__insert_dup(ticket->state, prop_name, prop_value); } granted = g_hash_table_lookup(ticket->state, PCMK__XA_GRANTED); if (granted && crm_is_true(granted)) { pcmk__set_ticket_flags(ticket, pcmk__ticket_granted); crm_info("We have ticket '%s'", ticket->id); } else { pcmk__clear_ticket_flags(ticket, pcmk__ticket_granted); crm_info("We do not have ticket '%s'", ticket->id); } last_granted = g_hash_table_lookup(ticket->state, PCMK_XA_LAST_GRANTED); if (last_granted) { long long last_granted_ll = 0LL; int rc = pcmk__scan_ll(last_granted, &last_granted_ll, 0LL); if (rc != pcmk_rc_ok) { crm_warn("Using %lld instead of invalid " PCMK_XA_LAST_GRANTED " value '%s' in state for ticket %s: %s", last_granted_ll, last_granted, ticket->id, pcmk_rc_str(rc)); } ticket->last_granted = (time_t) last_granted_ll; } standby = g_hash_table_lookup(ticket->state, PCMK_XA_STANDBY); if (standby && crm_is_true(standby)) { pcmk__set_ticket_flags(ticket, pcmk__ticket_standby); if (pcmk_is_set(ticket->flags, pcmk__ticket_granted)) { crm_info("Granted ticket '%s' is in standby-mode", ticket->id); } } else { pcmk__clear_ticket_flags(ticket, pcmk__ticket_standby); } crm_trace("Done with ticket state for %s", ticket_id); return pcmk_rc_ok; } static void unpack_handle_remote_attrs(pcmk_node_t *this_node, const xmlNode *state, pcmk_scheduler_t *scheduler) { const char *discovery = NULL; const xmlNode *attrs = NULL; pcmk_resource_t *rsc = NULL; int maint = 0; if (!pcmk__xe_is(state, PCMK__XE_NODE_STATE)) { return; } if ((this_node == NULL) || !pcmk__is_pacemaker_remote_node(this_node)) { return; } crm_trace("Processing Pacemaker Remote node %s", pcmk__node_name(this_node)); pcmk__scan_min_int(crm_element_value(state, PCMK__XA_NODE_IN_MAINTENANCE), &maint, 0); if (maint) { pcmk__set_node_flags(this_node, pcmk__node_remote_maint); } else { pcmk__clear_node_flags(this_node, pcmk__node_remote_maint); } rsc = this_node->priv->remote; if (!pcmk_is_set(this_node->priv->flags, pcmk__node_remote_reset)) { this_node->details->unclean = FALSE; pcmk__set_node_flags(this_node, pcmk__node_seen); } attrs = pcmk__xe_first_child(state, PCMK__XE_TRANSIENT_ATTRIBUTES, NULL, NULL); add_node_attrs(attrs, this_node, TRUE, scheduler); if (pe__shutdown_requested(this_node)) { crm_info("%s is shutting down", pcmk__node_name(this_node)); this_node->details->shutdown = TRUE; } if (crm_is_true(pcmk__node_attr(this_node, PCMK_NODE_ATTR_STANDBY, NULL, pcmk__rsc_node_current))) { crm_info("%s is in standby mode", pcmk__node_name(this_node)); pcmk__set_node_flags(this_node, pcmk__node_standby); } if (crm_is_true(pcmk__node_attr(this_node, PCMK_NODE_ATTR_MAINTENANCE, NULL, pcmk__rsc_node_current)) || ((rsc != NULL) && !pcmk_is_set(rsc->flags, pcmk__rsc_managed))) { crm_info("%s is in maintenance mode", pcmk__node_name(this_node)); this_node->details->maintenance = TRUE; } discovery = pcmk__node_attr(this_node, PCMK__NODE_ATTR_RESOURCE_DISCOVERY_ENABLED, NULL, pcmk__rsc_node_current); if ((discovery != NULL) && !crm_is_true(discovery)) { pcmk__warn_once(pcmk__wo_rdisc_enabled, "Support for the " PCMK__NODE_ATTR_RESOURCE_DISCOVERY_ENABLED " node attribute is deprecated and will be removed" " (and behave as 'true') in a future release."); if (pcmk__is_remote_node(this_node) && !pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled)) { pcmk__config_warn("Ignoring " PCMK__NODE_ATTR_RESOURCE_DISCOVERY_ENABLED " attribute on Pacemaker Remote node %s" " because fencing is disabled", pcmk__node_name(this_node)); } else { /* This is either a remote node with fencing enabled, or a guest * node. We don't care whether fencing is enabled when fencing guest * nodes, because they are "fenced" by recovering their containing * resource. */ crm_info("%s has resource discovery disabled", pcmk__node_name(this_node)); pcmk__clear_node_flags(this_node, pcmk__node_probes_allowed); } } } /*! * \internal * \brief Unpack a cluster node's transient attributes * * \param[in] state CIB node state XML * \param[in,out] node Cluster node whose attributes are being unpacked * \param[in,out] scheduler Scheduler data */ static void unpack_transient_attributes(const xmlNode *state, pcmk_node_t *node, pcmk_scheduler_t *scheduler) { const char *discovery = NULL; const xmlNode *attrs = pcmk__xe_first_child(state, PCMK__XE_TRANSIENT_ATTRIBUTES, NULL, NULL); add_node_attrs(attrs, node, TRUE, scheduler); if (crm_is_true(pcmk__node_attr(node, PCMK_NODE_ATTR_STANDBY, NULL, pcmk__rsc_node_current))) { crm_info("%s is in standby mode", pcmk__node_name(node)); pcmk__set_node_flags(node, pcmk__node_standby); } if (crm_is_true(pcmk__node_attr(node, PCMK_NODE_ATTR_MAINTENANCE, NULL, pcmk__rsc_node_current))) { crm_info("%s is in maintenance mode", pcmk__node_name(node)); node->details->maintenance = TRUE; } discovery = pcmk__node_attr(node, PCMK__NODE_ATTR_RESOURCE_DISCOVERY_ENABLED, NULL, pcmk__rsc_node_current); if ((discovery != NULL) && !crm_is_true(discovery)) { pcmk__config_warn("Ignoring " PCMK__NODE_ATTR_RESOURCE_DISCOVERY_ENABLED " attribute for %s because disabling resource" " discovery is not allowed for cluster nodes", pcmk__node_name(node)); } } /*! * \internal * \brief Unpack a node state entry (first pass) * * Unpack one node state entry from status. This unpacks information from the * \C PCMK__XE_NODE_STATE element itself and node attributes inside it, but not * the resource history inside it. Multiple passes through the status are needed * to fully unpack everything. * * \param[in] state CIB node state XML * \param[in,out] scheduler Scheduler data */ static void unpack_node_state(const xmlNode *state, pcmk_scheduler_t *scheduler) { const char *id = NULL; const char *uname = NULL; pcmk_node_t *this_node = NULL; id = crm_element_value(state, PCMK_XA_ID); if (id == NULL) { pcmk__config_err("Ignoring invalid " PCMK__XE_NODE_STATE " entry without " PCMK_XA_ID); crm_log_xml_info(state, "missing-id"); return; } uname = crm_element_value(state, PCMK_XA_UNAME); if (uname == NULL) { /* If a joining peer makes the cluster acquire the quorum from Corosync * but has not joined the controller CPG membership yet, it's possible * that the created PCMK__XE_NODE_STATE entry doesn't have a * PCMK_XA_UNAME yet. Recognize the node as pending and wait for it to * join CPG. */ crm_trace("Handling " PCMK__XE_NODE_STATE " entry with id=\"%s\" " "without " PCMK_XA_UNAME, id); } this_node = pe_find_node_any(scheduler->nodes, id, uname); if (this_node == NULL) { crm_notice("Ignoring recorded state for removed node with name %s and " PCMK_XA_ID " %s", pcmk__s(uname, "unknown"), id); return; } if (pcmk__is_pacemaker_remote_node(this_node)) { int remote_fenced = 0; /* We can't determine the online status of Pacemaker Remote nodes until * after all resource history has been unpacked. In this first pass, we * do need to mark whether the node has been fenced, as this plays a * role during unpacking cluster node resource state. */ pcmk__scan_min_int(crm_element_value(state, PCMK__XA_NODE_FENCED), &remote_fenced, 0); if (remote_fenced) { pcmk__set_node_flags(this_node, pcmk__node_remote_fenced); } else { pcmk__clear_node_flags(this_node, pcmk__node_remote_fenced); } return; } unpack_transient_attributes(state, this_node, scheduler); /* Provisionally mark this cluster node as clean. We have at least seen it * in the current cluster's lifetime. */ this_node->details->unclean = FALSE; pcmk__set_node_flags(this_node, pcmk__node_seen); crm_trace("Determining online status of cluster node %s (id %s)", pcmk__node_name(this_node), id); determine_online_status(state, this_node, scheduler); if (!pcmk_is_set(scheduler->flags, pcmk__sched_quorate) && this_node->details->online && (scheduler->no_quorum_policy == pcmk_no_quorum_fence)) { /* Everything else should flow from this automatically * (at least until the scheduler becomes able to migrate off * healthy resources) */ pe_fence_node(scheduler, this_node, "cluster does not have quorum", FALSE); } } /*! * \internal * \brief Unpack nodes' resource history as much as possible * * Unpack as many nodes' resource history as possible in one pass through the * status. We need to process Pacemaker Remote nodes' connections/containers * before unpacking their history; the connection/container history will be * in another node's history, so it might take multiple passes to unpack * everything. * * \param[in] status CIB XML status section * \param[in] fence If true, treat any not-yet-unpacked nodes as unseen * \param[in,out] scheduler Scheduler data * * \return Standard Pacemaker return code (specifically pcmk_rc_ok if done, * or EAGAIN if more unpacking remains to be done) */ static int unpack_node_history(const xmlNode *status, bool fence, pcmk_scheduler_t *scheduler) { int rc = pcmk_rc_ok; // Loop through all PCMK__XE_NODE_STATE entries in CIB status for (const xmlNode *state = pcmk__xe_first_child(status, PCMK__XE_NODE_STATE, NULL, NULL); state != NULL; state = pcmk__xe_next(state, PCMK__XE_NODE_STATE)) { const char *id = pcmk__xe_id(state); const char *uname = crm_element_value(state, PCMK_XA_UNAME); pcmk_node_t *this_node = NULL; if ((id == NULL) || (uname == NULL)) { // Warning already logged in first pass through status section crm_trace("Not unpacking resource history from malformed " PCMK__XE_NODE_STATE " without id and/or uname"); continue; } this_node = pe_find_node_any(scheduler->nodes, id, uname); if (this_node == NULL) { // Warning already logged in first pass through status section crm_trace("Not unpacking resource history for node %s because " "no longer in configuration", id); continue; } if (pcmk_is_set(this_node->priv->flags, pcmk__node_unpacked)) { crm_trace("Not unpacking resource history for node %s because " "already unpacked", id); continue; } if (fence) { // We're processing all remaining nodes } else if (pcmk__is_guest_or_bundle_node(this_node)) { /* We can unpack a guest node's history only after we've unpacked * other resource history to the point that we know that the node's * connection and containing resource are both up. */ const pcmk_resource_t *remote = this_node->priv->remote; const pcmk_resource_t *launcher = remote->priv->launcher; if ((remote->priv->orig_role != pcmk_role_started) || (launcher->priv->orig_role != pcmk_role_started)) { crm_trace("Not unpacking resource history for guest node %s " "because launcher and connection are not known to " "be up", id); continue; } } else if (pcmk__is_remote_node(this_node)) { /* We can unpack a remote node's history only after we've unpacked * other resource history to the point that we know that the node's * connection is up, with the exception of when shutdown locks are * in use. */ pcmk_resource_t *rsc = this_node->priv->remote; if ((rsc == NULL) || (!pcmk_is_set(scheduler->flags, pcmk__sched_shutdown_lock) && (rsc->priv->orig_role != pcmk_role_started))) { crm_trace("Not unpacking resource history for remote node %s " "because connection is not known to be up", id); continue; } /* If fencing and shutdown locks are disabled and we're not processing * unseen nodes, then we don't want to unpack offline nodes until online * nodes have been unpacked. This allows us to number active clone * instances first. */ } else if (!pcmk_any_flags_set(scheduler->flags, pcmk__sched_fencing_enabled |pcmk__sched_shutdown_lock) && !this_node->details->online) { crm_trace("Not unpacking resource history for offline " "cluster node %s", id); continue; } if (pcmk__is_pacemaker_remote_node(this_node)) { determine_remote_online_status(scheduler, this_node); unpack_handle_remote_attrs(this_node, state, scheduler); } crm_trace("Unpacking resource history for %snode %s", (fence? "unseen " : ""), id); pcmk__set_node_flags(this_node, pcmk__node_unpacked); unpack_node_lrm(this_node, state, scheduler); rc = EAGAIN; // Other node histories might depend on this one } return rc; } /* remove nodes that are down, stopping */ /* create positive rsc_to_node constraints between resources and the nodes they are running on */ /* anything else? */ gboolean unpack_status(xmlNode *status, pcmk_scheduler_t *scheduler) { xmlNode *state = NULL; crm_trace("Beginning unpack"); if (scheduler->priv->ticket_constraints == NULL) { scheduler->priv->ticket_constraints = pcmk__strkey_table(free, destroy_ticket); } for (state = pcmk__xe_first_child(status, NULL, NULL, NULL); state != NULL; state = pcmk__xe_next(state, NULL)) { if (pcmk__xe_is(state, PCMK_XE_TICKETS)) { pcmk__xe_foreach_child(state, PCMK__XE_TICKET_STATE, unpack_ticket_state, scheduler); } else if (pcmk__xe_is(state, PCMK__XE_NODE_STATE)) { unpack_node_state(state, scheduler); } } while (unpack_node_history(status, FALSE, scheduler) == EAGAIN) { crm_trace("Another pass through node resource histories is needed"); } // Now catch any nodes we didn't see unpack_node_history(status, pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled), scheduler); /* Now that we know where resources are, we can schedule stops of containers * with failed bundle connections */ if (scheduler->priv->stop_needed != NULL) { for (GList *item = scheduler->priv->stop_needed; item != NULL; item = item->next) { pcmk_resource_t *container = item->data; pcmk_node_t *node = pcmk__current_node(container); if (node) { stop_action(container, node, FALSE); } } g_list_free(scheduler->priv->stop_needed); scheduler->priv->stop_needed = NULL; } /* Now that we know status of all Pacemaker Remote connections and nodes, * we can stop connections for node shutdowns, and check the online status * of remote/guest nodes that didn't have any node history to unpack. */ for (GList *gIter = scheduler->nodes; gIter != NULL; gIter = gIter->next) { pcmk_node_t *this_node = gIter->data; if (!pcmk__is_pacemaker_remote_node(this_node)) { continue; } if (this_node->details->shutdown && (this_node->priv->remote != NULL)) { pe__set_next_role(this_node->priv->remote, pcmk_role_stopped, "remote shutdown"); } if (!pcmk_is_set(this_node->priv->flags, pcmk__node_unpacked)) { determine_remote_online_status(scheduler, this_node); } } return TRUE; } /*! * \internal * \brief Unpack node's time when it became a member at the cluster layer * * \param[in] node_state Node's \c PCMK__XE_NODE_STATE entry * \param[in,out] scheduler Scheduler data * * \return Epoch time when node became a cluster member * (or scheduler effective time for legacy entries) if a member, * 0 if not a member, or -1 if no valid information available */ static long long unpack_node_member(const xmlNode *node_state, pcmk_scheduler_t *scheduler) { const char *member_time = crm_element_value(node_state, PCMK__XA_IN_CCM); int member = 0; if (member_time == NULL) { return -1LL; } else if (crm_str_to_boolean(member_time, &member) == 1) { /* If in_ccm=0, we'll return 0 here. If in_ccm=1, either the entry was * recorded as a boolean for a DC < 2.1.7, or the node is pending * shutdown and has left the CPG, in which case it was set to 1 to avoid * fencing for PCMK_OPT_NODE_PENDING_TIMEOUT. * * We return the effective time for in_ccm=1 because what's important to * avoid fencing is that effective time minus this value is less than * the pending node timeout. */ - return member? (long long) get_effective_time(scheduler) : 0LL; + return member? (long long) pcmk__scheduler_epoch_time(scheduler) : 0LL; } else { long long when_member = 0LL; if ((pcmk__scan_ll(member_time, &when_member, 0LL) != pcmk_rc_ok) || (when_member < 0LL)) { crm_warn("Unrecognized value '%s' for " PCMK__XA_IN_CCM " in " PCMK__XE_NODE_STATE " entry", member_time); return -1LL; } return when_member; } } /*! * \internal * \brief Unpack node's time when it became online in process group * * \param[in] node_state Node's \c PCMK__XE_NODE_STATE entry * * \return Epoch time when node became online in process group (or 0 if not * online, or 1 for legacy online entries) */ static long long unpack_node_online(const xmlNode *node_state) { const char *peer_time = crm_element_value(node_state, PCMK_XA_CRMD); // @COMPAT Entries recorded for DCs < 2.1.7 have "online" or "offline" if (pcmk__str_eq(peer_time, PCMK_VALUE_OFFLINE, pcmk__str_casei|pcmk__str_null_matches)) { return 0LL; } else if (pcmk__str_eq(peer_time, PCMK_VALUE_ONLINE, pcmk__str_casei)) { return 1LL; } else { long long when_online = 0LL; if ((pcmk__scan_ll(peer_time, &when_online, 0LL) != pcmk_rc_ok) || (when_online < 0)) { crm_warn("Unrecognized value '%s' for " PCMK_XA_CRMD " in " PCMK__XE_NODE_STATE " entry, assuming offline", peer_time); return 0LL; } return when_online; } } /*! * \internal * \brief Unpack node attribute for user-requested fencing * * \param[in] node Node to check * \param[in] node_state Node's \c PCMK__XE_NODE_STATE entry in CIB status * * \return \c true if fencing has been requested for \p node, otherwise \c false */ static bool unpack_node_terminate(const pcmk_node_t *node, const xmlNode *node_state) { long long value = 0LL; int value_i = 0; int rc = pcmk_rc_ok; const char *value_s = pcmk__node_attr(node, PCMK_NODE_ATTR_TERMINATE, NULL, pcmk__rsc_node_current); // Value may be boolean or an epoch time if (crm_str_to_boolean(value_s, &value_i) == 1) { return (value_i != 0); } rc = pcmk__scan_ll(value_s, &value, 0LL); if (rc == pcmk_rc_ok) { return (value > 0); } crm_warn("Ignoring unrecognized value '%s' for " PCMK_NODE_ATTR_TERMINATE "node attribute for %s: %s", value_s, pcmk__node_name(node), pcmk_rc_str(rc)); return false; } static gboolean determine_online_status_no_fencing(pcmk_scheduler_t *scheduler, const xmlNode *node_state, pcmk_node_t *this_node) { gboolean online = FALSE; const char *join = crm_element_value(node_state, PCMK__XA_JOIN); const char *exp_state = crm_element_value(node_state, PCMK_XA_EXPECTED); long long when_member = unpack_node_member(node_state, scheduler); long long when_online = unpack_node_online(node_state); if (when_member <= 0) { crm_trace("Node %s is %sdown", pcmk__node_name(this_node), ((when_member < 0)? "presumed " : "")); } else if (when_online > 0) { if (pcmk__str_eq(join, CRMD_JOINSTATE_MEMBER, pcmk__str_casei)) { online = TRUE; } else { crm_debug("Node %s is not ready to run resources: %s", pcmk__node_name(this_node), join); } } else if (!pcmk_is_set(this_node->priv->flags, pcmk__node_expected_up)) { crm_trace("Node %s controller is down: " "member@%lld online@%lld join=%s expected=%s", pcmk__node_name(this_node), when_member, when_online, pcmk__s(join, ""), pcmk__s(exp_state, "")); } else { /* mark it unclean */ pe_fence_node(scheduler, this_node, "peer is unexpectedly down", FALSE); crm_info("Node %s member@%lld online@%lld join=%s expected=%s", pcmk__node_name(this_node), when_member, when_online, pcmk__s(join, ""), pcmk__s(exp_state, "")); } return online; } /*! * \internal * \brief Check whether a node has taken too long to join controller group * * \param[in,out] scheduler Scheduler data * \param[in] node Node to check * \param[in] when_member Epoch time when node became a cluster member * \param[in] when_online Epoch time when node joined controller group * * \return true if node has been pending (on the way up) longer than * \c PCMK_OPT_NODE_PENDING_TIMEOUT, otherwise false * \note This will also update the cluster's recheck time if appropriate. */ static inline bool pending_too_long(pcmk_scheduler_t *scheduler, const pcmk_node_t *node, long long when_member, long long when_online) { if ((scheduler->priv->node_pending_ms > 0U) && (when_member > 0) && (when_online <= 0)) { // There is a timeout on pending nodes, and node is pending time_t timeout = when_member + pcmk__timeout_ms2s(scheduler->priv->node_pending_ms); - if (get_effective_time(node->priv->scheduler) >= timeout) { + if (pcmk__scheduler_epoch_time(node->priv->scheduler) >= timeout) { return true; // Node has timed out } // Node is pending, but still has time - pe__update_recheck_time(timeout, scheduler, "pending node timeout"); + pcmk__update_recheck_time(timeout, scheduler, "pending node timeout"); } return false; } static bool determine_online_status_fencing(pcmk_scheduler_t *scheduler, const xmlNode *node_state, pcmk_node_t *this_node) { bool termination_requested = unpack_node_terminate(this_node, node_state); const char *join = crm_element_value(node_state, PCMK__XA_JOIN); const char *exp_state = crm_element_value(node_state, PCMK_XA_EXPECTED); long long when_member = unpack_node_member(node_state, scheduler); long long when_online = unpack_node_online(node_state); /* - PCMK__XA_JOIN ::= member|down|pending|banned - PCMK_XA_EXPECTED ::= member|down @COMPAT with entries recorded for DCs < 2.1.7 - PCMK__XA_IN_CCM ::= true|false - PCMK_XA_CRMD ::= online|offline Since crm_feature_set 3.18.0 (pacemaker-2.1.7): - PCMK__XA_IN_CCM ::= |0 Since when node has been a cluster member. A value 0 of means the node is not a cluster member. - PCMK_XA_CRMD ::= |0 Since when peer has been online in CPG. A value 0 means the peer is offline in CPG. */ crm_trace("Node %s member@%lld online@%lld join=%s expected=%s%s", pcmk__node_name(this_node), when_member, when_online, pcmk__s(join, ""), pcmk__s(exp_state, ""), (termination_requested? " (termination requested)" : "")); if (this_node->details->shutdown) { crm_debug("%s is shutting down", pcmk__node_name(this_node)); /* Slightly different criteria since we can't shut down a dead peer */ return (when_online > 0); } if (when_member < 0) { pe_fence_node(scheduler, this_node, "peer has not been seen by the cluster", FALSE); return false; } if (pcmk__str_eq(join, CRMD_JOINSTATE_NACK, pcmk__str_none)) { pe_fence_node(scheduler, this_node, "peer failed Pacemaker membership criteria", FALSE); } else if (termination_requested) { if ((when_member <= 0) && (when_online <= 0) && pcmk__str_eq(join, CRMD_JOINSTATE_DOWN, pcmk__str_none)) { crm_info("%s was fenced as requested", pcmk__node_name(this_node)); return false; } pe_fence_node(scheduler, this_node, "fencing was requested", false); } else if (pcmk__str_eq(exp_state, CRMD_JOINSTATE_DOWN, pcmk__str_null_matches)) { if (pending_too_long(scheduler, this_node, when_member, when_online)) { pe_fence_node(scheduler, this_node, "peer pending timed out on joining the process group", FALSE); } else if ((when_member > 0) || (when_online > 0)) { crm_info("- %s is not ready to run resources", pcmk__node_name(this_node)); pcmk__set_node_flags(this_node, pcmk__node_standby); this_node->details->pending = TRUE; } else { crm_trace("%s is down or still coming up", pcmk__node_name(this_node)); } } else if (when_member <= 0) { // Consider PCMK_OPT_PRIORITY_FENCING_DELAY for lost nodes pe_fence_node(scheduler, this_node, "peer is no longer part of the cluster", TRUE); } else if (when_online <= 0) { pe_fence_node(scheduler, this_node, "peer process is no longer available", FALSE); /* Everything is running at this point, now check join state */ } else if (pcmk__str_eq(join, CRMD_JOINSTATE_MEMBER, pcmk__str_none)) { crm_info("%s is active", pcmk__node_name(this_node)); } else if (pcmk__str_any_of(join, CRMD_JOINSTATE_PENDING, CRMD_JOINSTATE_DOWN, NULL)) { crm_info("%s is not ready to run resources", pcmk__node_name(this_node)); pcmk__set_node_flags(this_node, pcmk__node_standby); this_node->details->pending = TRUE; } else { pe_fence_node(scheduler, this_node, "peer was in an unknown state", FALSE); } return (when_member > 0); } static void determine_remote_online_status(pcmk_scheduler_t *scheduler, pcmk_node_t *this_node) { pcmk_resource_t *rsc = this_node->priv->remote; pcmk_resource_t *launcher = NULL; pcmk_node_t *host = NULL; const char *node_type = "Remote"; if (rsc == NULL) { /* This is a leftover node state entry for a former Pacemaker Remote * node whose connection resource was removed. Consider it offline. */ crm_trace("Pacemaker Remote node %s is considered OFFLINE because " "its connection resource has been removed from the CIB", this_node->priv->id); this_node->details->online = FALSE; return; } launcher = rsc->priv->launcher; if (launcher != NULL) { node_type = "Guest"; if (pcmk__list_of_1(rsc->priv->active_nodes)) { host = rsc->priv->active_nodes->data; } } /* If the resource is currently started, mark it online. */ if (rsc->priv->orig_role == pcmk_role_started) { this_node->details->online = TRUE; } /* consider this node shutting down if transitioning start->stop */ if ((rsc->priv->orig_role == pcmk_role_started) && (rsc->priv->next_role == pcmk_role_stopped)) { crm_trace("%s node %s shutting down because connection resource is stopping", node_type, this_node->priv->id); this_node->details->shutdown = TRUE; } /* Now check all the failure conditions. */ if ((launcher != NULL) && pcmk_is_set(launcher->flags, pcmk__rsc_failed)) { crm_trace("Guest node %s UNCLEAN because guest resource failed", this_node->priv->id); this_node->details->online = FALSE; pcmk__set_node_flags(this_node, pcmk__node_remote_reset); } else if (pcmk_is_set(rsc->flags, pcmk__rsc_failed)) { crm_trace("%s node %s OFFLINE because connection resource failed", node_type, this_node->priv->id); this_node->details->online = FALSE; } else if ((rsc->priv->orig_role == pcmk_role_stopped) || ((launcher != NULL) && (launcher->priv->orig_role == pcmk_role_stopped))) { crm_trace("%s node %s OFFLINE because its resource is stopped", node_type, this_node->priv->id); this_node->details->online = FALSE; pcmk__clear_node_flags(this_node, pcmk__node_remote_reset); } else if (host && (host->details->online == FALSE) && host->details->unclean) { crm_trace("Guest node %s UNCLEAN because host is unclean", this_node->priv->id); this_node->details->online = FALSE; pcmk__set_node_flags(this_node, pcmk__node_remote_reset); } else { crm_trace("%s node %s is %s", node_type, this_node->priv->id, this_node->details->online? "ONLINE" : "OFFLINE"); } } static void determine_online_status(const xmlNode *node_state, pcmk_node_t *this_node, pcmk_scheduler_t *scheduler) { gboolean online = FALSE; const char *exp_state = crm_element_value(node_state, PCMK_XA_EXPECTED); CRM_CHECK(this_node != NULL, return); this_node->details->shutdown = FALSE; if (pe__shutdown_requested(this_node)) { this_node->details->shutdown = TRUE; } else if (pcmk__str_eq(exp_state, CRMD_JOINSTATE_MEMBER, pcmk__str_casei)) { pcmk__set_node_flags(this_node, pcmk__node_expected_up); } if (!pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled)) { online = determine_online_status_no_fencing(scheduler, node_state, this_node); } else { online = determine_online_status_fencing(scheduler, node_state, this_node); } if (online) { this_node->details->online = TRUE; } else { /* remove node from contention */ this_node->assign->score = -PCMK_SCORE_INFINITY; } if (online && this_node->details->shutdown) { /* don't run resources here */ this_node->assign->score = -PCMK_SCORE_INFINITY; } if (this_node->details->unclean) { pcmk__sched_warn(scheduler, "%s is unclean", pcmk__node_name(this_node)); } else if (!this_node->details->online) { crm_trace("%s is offline", pcmk__node_name(this_node)); } else if (this_node->details->shutdown) { crm_info("%s is shutting down", pcmk__node_name(this_node)); } else if (this_node->details->pending) { crm_info("%s is pending", pcmk__node_name(this_node)); } else if (pcmk_is_set(this_node->priv->flags, pcmk__node_standby)) { crm_info("%s is in standby", pcmk__node_name(this_node)); } else if (this_node->details->maintenance) { crm_info("%s is in maintenance", pcmk__node_name(this_node)); } else { crm_info("%s is online", pcmk__node_name(this_node)); } } /*! * \internal * \brief Find the end of a resource's name, excluding any clone suffix * * \param[in] id Resource ID to check * * \return Pointer to last character of resource's base name */ const char * pe_base_name_end(const char *id) { if (!pcmk__str_empty(id)) { const char *end = id + strlen(id) - 1; for (const char *s = end; s > id; --s) { switch (*s) { case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9': break; case ':': return (s == end)? s : (s - 1); default: return end; } } return end; } return NULL; } /*! * \internal * \brief Get a resource name excluding any clone suffix * * \param[in] last_rsc_id Resource ID to check * * \return Pointer to newly allocated string with resource's base name * \note It is the caller's responsibility to free() the result. * This asserts on error, so callers can assume result is not NULL. */ char * clone_strip(const char *last_rsc_id) { const char *end = pe_base_name_end(last_rsc_id); char *basename = NULL; pcmk__assert(end != NULL); basename = strndup(last_rsc_id, end - last_rsc_id + 1); pcmk__assert(basename != NULL); return basename; } /*! * \internal * \brief Get the name of the first instance of a cloned resource * * \param[in] last_rsc_id Resource ID to check * * \return Pointer to newly allocated string with resource's base name plus :0 * \note It is the caller's responsibility to free() the result. * This asserts on error, so callers can assume result is not NULL. */ char * clone_zero(const char *last_rsc_id) { const char *end = pe_base_name_end(last_rsc_id); size_t base_name_len = end - last_rsc_id + 1; char *zero = NULL; pcmk__assert(end != NULL); zero = pcmk__assert_alloc(base_name_len + 3, sizeof(char)); memcpy(zero, last_rsc_id, base_name_len); zero[base_name_len] = ':'; zero[base_name_len + 1] = '0'; return zero; } static pcmk_resource_t * create_fake_resource(const char *rsc_id, const xmlNode *rsc_entry, pcmk_scheduler_t *scheduler) { pcmk_resource_t *rsc = NULL; xmlNode *xml_rsc = pcmk__xe_create(NULL, PCMK_XE_PRIMITIVE); pcmk__xe_copy_attrs(xml_rsc, rsc_entry, pcmk__xaf_none); crm_xml_add(xml_rsc, PCMK_XA_ID, rsc_id); crm_log_xml_debug(xml_rsc, "Orphan resource"); if (pe__unpack_resource(xml_rsc, &rsc, NULL, scheduler) != pcmk_rc_ok) { return NULL; } if (xml_contains_remote_node(xml_rsc)) { pcmk_node_t *node; crm_debug("Detected orphaned remote node %s", rsc_id); node = pcmk_find_node(scheduler, rsc_id); if (node == NULL) { node = pe_create_node(rsc_id, rsc_id, PCMK_VALUE_REMOTE, 0, scheduler); } link_rsc2remotenode(scheduler, rsc); if (node) { crm_trace("Setting node %s as shutting down due to orphaned connection resource", rsc_id); node->details->shutdown = TRUE; } } if (crm_element_value(rsc_entry, PCMK__META_CONTAINER)) { // This removed resource needs to be mapped to a launcher crm_trace("Launched resource %s was removed from the configuration", rsc_id); pcmk__set_rsc_flags(rsc, pcmk__rsc_removed_launched); } pcmk__set_rsc_flags(rsc, pcmk__rsc_removed); scheduler->priv->resources = g_list_append(scheduler->priv->resources, rsc); return rsc; } /*! * \internal * \brief Create orphan instance for anonymous clone resource history * * \param[in,out] parent Clone resource that orphan will be added to * \param[in] rsc_id Orphan's resource ID * \param[in] node Where orphan is active (for logging only) * \param[in,out] scheduler Scheduler data * * \return Newly added orphaned instance of \p parent */ static pcmk_resource_t * create_anonymous_orphan(pcmk_resource_t *parent, const char *rsc_id, const pcmk_node_t *node, pcmk_scheduler_t *scheduler) { pcmk_resource_t *top = pe__create_clone_child(parent, scheduler); pcmk_resource_t *orphan = NULL; // find_rsc() because we might be a cloned group orphan = top->priv->fns->find_rsc(top, rsc_id, NULL, pcmk_rsc_match_clone_only); pcmk__rsc_debug(parent, "Created orphan %s for %s: %s on %s", top->id, parent->id, rsc_id, pcmk__node_name(node)); return orphan; } /*! * \internal * \brief Check a node for an instance of an anonymous clone * * Return a child instance of the specified anonymous clone, in order of * preference: (1) the instance running on the specified node, if any; * (2) an inactive instance (i.e. within the total of \c PCMK_META_CLONE_MAX * instances); (3) a newly created orphan (that is, \c PCMK_META_CLONE_MAX * instances are already active). * * \param[in,out] scheduler Scheduler data * \param[in] node Node on which to check for instance * \param[in,out] parent Clone to check * \param[in] rsc_id Name of cloned resource in history (no instance) */ static pcmk_resource_t * find_anonymous_clone(pcmk_scheduler_t *scheduler, const pcmk_node_t *node, pcmk_resource_t *parent, const char *rsc_id) { GList *rIter = NULL; pcmk_resource_t *rsc = NULL; pcmk_resource_t *inactive_instance = NULL; gboolean skip_inactive = FALSE; pcmk__assert(pcmk__is_anonymous_clone(parent)); // Check for active (or partially active, for cloned groups) instance pcmk__rsc_trace(parent, "Looking for %s on %s in %s", rsc_id, pcmk__node_name(node), parent->id); for (rIter = parent->priv->children; (rIter != NULL) && (rsc == NULL); rIter = rIter->next) { GList *locations = NULL; pcmk_resource_t *child = rIter->data; /* Check whether this instance is already known to be active or pending * anywhere, at this stage of unpacking. Because this function is called * for a resource before the resource's individual operation history * entries are unpacked, locations will generally not contain the * desired node. * * However, there are three exceptions: * (1) when child is a cloned group and we have already unpacked the * history of another member of the group on the same node; * (2) when we've already unpacked the history of another numbered * instance on the same node (which can happen if * PCMK_META_GLOBALLY_UNIQUE was flipped from true to false); and * (3) when we re-run calculations on the same scheduler data as part of * a simulation. */ child->priv->fns->location(child, &locations, pcmk__rsc_node_current |pcmk__rsc_node_pending); if (locations) { /* We should never associate the same numbered anonymous clone * instance with multiple nodes, and clone instances can't migrate, * so there must be only one location, regardless of history. */ CRM_LOG_ASSERT(locations->next == NULL); if (pcmk__same_node((pcmk_node_t *) locations->data, node)) { /* This child instance is active on the requested node, so check * for a corresponding configured resource. We use find_rsc() * instead of child because child may be a cloned group, and we * need the particular member corresponding to rsc_id. * * If the history entry is orphaned, rsc will be NULL. */ rsc = parent->priv->fns->find_rsc(child, rsc_id, NULL, pcmk_rsc_match_clone_only); if (rsc) { /* If there are multiple instance history entries for an * anonymous clone in a single node's history (which can * happen if PCMK_META_GLOBALLY_UNIQUE is switched from true * to false), we want to consider the instances beyond the * first as orphans, even if there are inactive instance * numbers available. */ if (rsc->priv->active_nodes != NULL) { crm_notice("Active (now-)anonymous clone %s has " "multiple (orphan) instance histories on %s", parent->id, pcmk__node_name(node)); skip_inactive = TRUE; rsc = NULL; } else { pcmk__rsc_trace(parent, "Resource %s, active", rsc->id); } } } g_list_free(locations); } else { pcmk__rsc_trace(parent, "Resource %s, skip inactive", child->id); if (!skip_inactive && !inactive_instance && !pcmk_is_set(child->flags, pcmk__rsc_blocked)) { // Remember one inactive instance in case we don't find active inactive_instance = parent->priv->fns->find_rsc(child, rsc_id, NULL, pcmk_rsc_match_clone_only); /* ... but don't use it if it was already associated with a * pending action on another node */ if (inactive_instance != NULL) { const pcmk_node_t *pending_node = NULL; pending_node = inactive_instance->priv->pending_node; if ((pending_node != NULL) && !pcmk__same_node(pending_node, node)) { inactive_instance = NULL; } } } } } if ((rsc == NULL) && !skip_inactive && (inactive_instance != NULL)) { pcmk__rsc_trace(parent, "Resource %s, empty slot", inactive_instance->id); rsc = inactive_instance; } /* If the resource has PCMK_META_REQUIRES set to PCMK_VALUE_QUORUM or * PCMK_VALUE_NOTHING, and we don't have a clone instance for every node, we * don't want to consume a valid instance number for unclean nodes. Such * instances may appear to be active according to the history, but should be * considered inactive, so we can start an instance elsewhere. Treat such * instances as orphans. * * An exception is instances running on guest nodes -- since guest node * "fencing" is actually just a resource stop, requires shouldn't apply. * * @TODO Ideally, we'd use an inactive instance number if it is not needed * for any clean instances. However, we don't know that at this point. */ if ((rsc != NULL) && !pcmk_is_set(rsc->flags, pcmk__rsc_needs_fencing) && (!node->details->online || node->details->unclean) && !pcmk__is_guest_or_bundle_node(node) && !pe__is_universal_clone(parent, scheduler)) { rsc = NULL; } if (rsc == NULL) { rsc = create_anonymous_orphan(parent, rsc_id, node, scheduler); pcmk__rsc_trace(parent, "Resource %s, orphan", rsc->id); } return rsc; } static pcmk_resource_t * unpack_find_resource(pcmk_scheduler_t *scheduler, const pcmk_node_t *node, const char *rsc_id) { pcmk_resource_t *rsc = NULL; pcmk_resource_t *parent = NULL; crm_trace("looking for %s", rsc_id); rsc = pe_find_resource(scheduler->priv->resources, rsc_id); if (rsc == NULL) { /* If we didn't find the resource by its name in the operation history, * check it again as a clone instance. Even when PCMK_META_CLONE_MAX=0, * we create a single :0 orphan to match against here. */ char *clone0_id = clone_zero(rsc_id); pcmk_resource_t *clone0 = pe_find_resource(scheduler->priv->resources, clone0_id); if (clone0 && !pcmk_is_set(clone0->flags, pcmk__rsc_unique)) { rsc = clone0; parent = uber_parent(clone0); crm_trace("%s found as %s (%s)", rsc_id, clone0_id, parent->id); } else { crm_trace("%s is not known as %s either (orphan)", rsc_id, clone0_id); } free(clone0_id); } else if (rsc->priv->variant > pcmk__rsc_variant_primitive) { crm_trace("Resource history for %s is orphaned " "because it is no longer primitive", rsc_id); return NULL; } else { parent = uber_parent(rsc); } if (pcmk__is_anonymous_clone(parent)) { if (pcmk__is_bundled(parent)) { rsc = pe__find_bundle_replica(parent->priv->parent, node); } else { char *base = clone_strip(rsc_id); rsc = find_anonymous_clone(scheduler, node, parent, base); free(base); pcmk__assert(rsc != NULL); } } if (rsc && !pcmk__str_eq(rsc_id, rsc->id, pcmk__str_none) && !pcmk__str_eq(rsc_id, rsc->priv->history_id, pcmk__str_none)) { pcmk__str_update(&(rsc->priv->history_id), rsc_id); pcmk__rsc_debug(rsc, "Internally renamed %s on %s to %s%s", rsc_id, pcmk__node_name(node), rsc->id, pcmk_is_set(rsc->flags, pcmk__rsc_removed)? " (ORPHAN)" : ""); } return rsc; } static pcmk_resource_t * process_orphan_resource(const xmlNode *rsc_entry, const pcmk_node_t *node, pcmk_scheduler_t *scheduler) { pcmk_resource_t *rsc = NULL; const char *rsc_id = crm_element_value(rsc_entry, PCMK_XA_ID); crm_debug("Detected orphan resource %s on %s", rsc_id, pcmk__node_name(node)); rsc = create_fake_resource(rsc_id, rsc_entry, scheduler); if (rsc == NULL) { return NULL; } if (!pcmk_is_set(scheduler->flags, pcmk__sched_stop_removed_resources)) { pcmk__clear_rsc_flags(rsc, pcmk__rsc_managed); } else { CRM_CHECK(rsc != NULL, return NULL); pcmk__rsc_trace(rsc, "Added orphan %s", rsc->id); resource_location(rsc, NULL, -PCMK_SCORE_INFINITY, "__orphan_do_not_run__", scheduler); } return rsc; } static void process_rsc_state(pcmk_resource_t *rsc, pcmk_node_t *node, enum pcmk__on_fail on_fail) { pcmk_node_t *tmpnode = NULL; char *reason = NULL; enum pcmk__on_fail save_on_fail = pcmk__on_fail_ignore; pcmk_scheduler_t *scheduler = NULL; bool known_active = false; pcmk__assert(rsc != NULL); scheduler = rsc->priv->scheduler; known_active = (rsc->priv->orig_role > pcmk_role_stopped); pcmk__rsc_trace(rsc, "Resource %s is %s on %s: on_fail=%s", rsc->id, pcmk_role_text(rsc->priv->orig_role), pcmk__node_name(node), pcmk__on_fail_text(on_fail)); /* process current state */ if (rsc->priv->orig_role != pcmk_role_unknown) { pcmk_resource_t *iter = rsc; while (iter) { if (g_hash_table_lookup(iter->priv->probed_nodes, node->priv->id) == NULL) { pcmk_node_t *n = pe__copy_node(node); pcmk__rsc_trace(rsc, "%s (%s in history) known on %s", rsc->id, pcmk__s(rsc->priv->history_id, "the same"), pcmk__node_name(n)); g_hash_table_insert(iter->priv->probed_nodes, (gpointer) n->priv->id, n); } if (pcmk_is_set(iter->flags, pcmk__rsc_unique)) { break; } iter = iter->priv->parent; } } /* If a managed resource is believed to be running, but node is down ... */ if (known_active && !node->details->online && !node->details->maintenance && pcmk_is_set(rsc->flags, pcmk__rsc_managed)) { gboolean should_fence = FALSE; /* If this is a guest node, fence it (regardless of whether fencing is * enabled, because guest node fencing is done by recovery of the * container resource rather than by the fencer). Mark the resource * we're processing as failed. When the guest comes back up, its * operation history in the CIB will be cleared, freeing the affected * resource to run again once we are sure we know its state. */ if (pcmk__is_guest_or_bundle_node(node)) { pcmk__set_rsc_flags(rsc, pcmk__rsc_failed|pcmk__rsc_stop_if_failed); should_fence = TRUE; } else if (pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled)) { if (pcmk__is_remote_node(node) && (node->priv->remote != NULL) && !pcmk_is_set(node->priv->remote->flags, pcmk__rsc_failed)) { /* Setting unseen means that fencing of the remote node will * occur only if the connection resource is not going to start * somewhere. This allows connection resources on a failed * cluster node to move to another node without requiring the * remote nodes to be fenced as well. */ pcmk__clear_node_flags(node, pcmk__node_seen); reason = crm_strdup_printf("%s is active there (fencing will be" " revoked if remote connection can " "be re-established elsewhere)", rsc->id); } should_fence = TRUE; } if (should_fence) { if (reason == NULL) { reason = crm_strdup_printf("%s is thought to be active there", rsc->id); } pe_fence_node(scheduler, node, reason, FALSE); } free(reason); } /* In order to calculate priority_fencing_delay correctly, save the failure information and pass it to native_add_running(). */ save_on_fail = on_fail; if (node->details->unclean) { /* No extra processing needed * Also allows resources to be started again after a node is shot */ on_fail = pcmk__on_fail_ignore; } switch (on_fail) { case pcmk__on_fail_ignore: /* nothing to do */ break; case pcmk__on_fail_demote: pcmk__set_rsc_flags(rsc, pcmk__rsc_failed); demote_action(rsc, node, FALSE); break; case pcmk__on_fail_fence_node: /* treat it as if it is still running * but also mark the node as unclean */ reason = crm_strdup_printf("%s failed there", rsc->id); pe_fence_node(scheduler, node, reason, FALSE); free(reason); break; case pcmk__on_fail_standby_node: pcmk__set_node_flags(node, pcmk__node_standby|pcmk__node_fail_standby); break; case pcmk__on_fail_block: /* is_managed == FALSE will prevent any * actions being sent for the resource */ pcmk__clear_rsc_flags(rsc, pcmk__rsc_managed); pcmk__set_rsc_flags(rsc, pcmk__rsc_blocked); break; case pcmk__on_fail_ban: /* make sure it comes up somewhere else * or not at all */ resource_location(rsc, node, -PCMK_SCORE_INFINITY, "__action_migration_auto__", scheduler); break; case pcmk__on_fail_stop: pe__set_next_role(rsc, pcmk_role_stopped, PCMK_META_ON_FAIL "=" PCMK_VALUE_STOP); break; case pcmk__on_fail_restart: if (known_active) { pcmk__set_rsc_flags(rsc, pcmk__rsc_failed|pcmk__rsc_stop_if_failed); stop_action(rsc, node, FALSE); } break; case pcmk__on_fail_restart_container: pcmk__set_rsc_flags(rsc, pcmk__rsc_failed|pcmk__rsc_stop_if_failed); if ((rsc->priv->launcher != NULL) && pcmk__is_bundled(rsc)) { /* A bundle's remote connection can run on a different node than * the bundle's container. We don't necessarily know where the * container is running yet, so remember it and add a stop * action for it later. */ scheduler->priv->stop_needed = g_list_prepend(scheduler->priv->stop_needed, rsc->priv->launcher); } else if (rsc->priv->launcher != NULL) { stop_action(rsc->priv->launcher, node, FALSE); } else if (known_active) { stop_action(rsc, node, FALSE); } break; case pcmk__on_fail_reset_remote: pcmk__set_rsc_flags(rsc, pcmk__rsc_failed|pcmk__rsc_stop_if_failed); if (pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled)) { tmpnode = NULL; if (pcmk_is_set(rsc->flags, pcmk__rsc_is_remote_connection)) { tmpnode = pcmk_find_node(scheduler, rsc->id); } if (pcmk__is_remote_node(tmpnode) && !pcmk_is_set(tmpnode->priv->flags, pcmk__node_remote_fenced)) { /* The remote connection resource failed in a way that * should result in fencing the remote node. */ pe_fence_node(scheduler, tmpnode, "remote connection is unrecoverable", FALSE); } } /* require the stop action regardless if fencing is occurring or not. */ if (known_active) { stop_action(rsc, node, FALSE); } /* if reconnect delay is in use, prevent the connection from exiting the * "STOPPED" role until the failure is cleared by the delay timeout. */ if (rsc->priv->remote_reconnect_ms > 0U) { pe__set_next_role(rsc, pcmk_role_stopped, "remote reset"); } break; } /* Ensure a remote connection failure forces an unclean Pacemaker Remote * node to be fenced. By marking the node as seen, the failure will result * in a fencing operation regardless if we're going to attempt to reconnect * in this transition. */ if (pcmk_all_flags_set(rsc->flags, pcmk__rsc_failed|pcmk__rsc_is_remote_connection)) { tmpnode = pcmk_find_node(scheduler, rsc->id); if (tmpnode && tmpnode->details->unclean) { pcmk__set_node_flags(tmpnode, pcmk__node_seen); } } if (known_active) { if (pcmk_is_set(rsc->flags, pcmk__rsc_removed)) { if (pcmk_is_set(rsc->flags, pcmk__rsc_managed)) { crm_notice("Removed resource %s is active on %s and will be " "stopped when possible", rsc->id, pcmk__node_name(node)); } else { crm_notice("Removed resource %s must be stopped manually on %s " "because " PCMK_OPT_STOP_ORPHAN_RESOURCES " is set to false", rsc->id, pcmk__node_name(node)); } } native_add_running(rsc, node, scheduler, (save_on_fail != pcmk__on_fail_ignore)); switch (on_fail) { case pcmk__on_fail_ignore: break; case pcmk__on_fail_demote: case pcmk__on_fail_block: pcmk__set_rsc_flags(rsc, pcmk__rsc_failed); break; default: pcmk__set_rsc_flags(rsc, pcmk__rsc_failed|pcmk__rsc_stop_if_failed); break; } } else if ((rsc->priv->history_id != NULL) && (strchr(rsc->priv->history_id, ':') != NULL)) { /* Only do this for older status sections that included instance numbers * Otherwise stopped instances will appear as orphans */ pcmk__rsc_trace(rsc, "Clearing history ID %s for %s (stopped)", rsc->priv->history_id, rsc->id); free(rsc->priv->history_id); rsc->priv->history_id = NULL; } else { GList *possible_matches = pe__resource_actions(rsc, node, PCMK_ACTION_STOP, FALSE); GList *gIter = possible_matches; for (; gIter != NULL; gIter = gIter->next) { pcmk_action_t *stop = (pcmk_action_t *) gIter->data; pcmk__set_action_flags(stop, pcmk__action_optional); } g_list_free(possible_matches); } /* A successful stop after migrate_to on the migration source doesn't make * the partially migrated resource stopped on the migration target. */ if ((rsc->priv->orig_role == pcmk_role_stopped) && (rsc->priv->active_nodes != NULL) && (rsc->priv->partial_migration_target != NULL) && pcmk__same_node(rsc->priv->partial_migration_source, node)) { rsc->priv->orig_role = pcmk_role_started; } } /* create active recurring operations as optional */ static void process_recurring(pcmk_node_t *node, pcmk_resource_t *rsc, int start_index, int stop_index, GList *sorted_op_list, pcmk_scheduler_t *scheduler) { int counter = -1; const char *task = NULL; const char *status = NULL; GList *gIter = sorted_op_list; pcmk__assert(rsc != NULL); pcmk__rsc_trace(rsc, "%s: Start index %d, stop index = %d", rsc->id, start_index, stop_index); for (; gIter != NULL; gIter = gIter->next) { xmlNode *rsc_op = (xmlNode *) gIter->data; guint interval_ms = 0; char *key = NULL; const char *id = pcmk__xe_id(rsc_op); counter++; if (node->details->online == FALSE) { pcmk__rsc_trace(rsc, "Skipping %s on %s: node is offline", rsc->id, pcmk__node_name(node)); break; /* Need to check if there's a monitor for role="Stopped" */ } else if (start_index < stop_index && counter <= stop_index) { pcmk__rsc_trace(rsc, "Skipping %s on %s: resource is not active", id, pcmk__node_name(node)); continue; } else if (counter < start_index) { pcmk__rsc_trace(rsc, "Skipping %s on %s: old %d", id, pcmk__node_name(node), counter); continue; } crm_element_value_ms(rsc_op, PCMK_META_INTERVAL, &interval_ms); if (interval_ms == 0) { pcmk__rsc_trace(rsc, "Skipping %s on %s: non-recurring", id, pcmk__node_name(node)); continue; } status = crm_element_value(rsc_op, PCMK__XA_OP_STATUS); if (pcmk__str_eq(status, "-1", pcmk__str_casei)) { pcmk__rsc_trace(rsc, "Skipping %s on %s: status", id, pcmk__node_name(node)); continue; } task = crm_element_value(rsc_op, PCMK_XA_OPERATION); /* create the action */ key = pcmk__op_key(rsc->id, task, interval_ms); pcmk__rsc_trace(rsc, "Creating %s on %s", key, pcmk__node_name(node)); custom_action(rsc, key, task, node, TRUE, scheduler); } } void calculate_active_ops(const GList *sorted_op_list, int *start_index, int *stop_index) { int counter = -1; int implied_monitor_start = -1; int implied_clone_start = -1; const char *task = NULL; const char *status = NULL; *stop_index = -1; *start_index = -1; for (const GList *iter = sorted_op_list; iter != NULL; iter = iter->next) { const xmlNode *rsc_op = (const xmlNode *) iter->data; counter++; task = crm_element_value(rsc_op, PCMK_XA_OPERATION); status = crm_element_value(rsc_op, PCMK__XA_OP_STATUS); if (pcmk__str_eq(task, PCMK_ACTION_STOP, pcmk__str_casei) && pcmk__str_eq(status, "0", pcmk__str_casei)) { *stop_index = counter; } else if (pcmk__strcase_any_of(task, PCMK_ACTION_START, PCMK_ACTION_MIGRATE_FROM, NULL)) { *start_index = counter; } else if ((implied_monitor_start <= *stop_index) && pcmk__str_eq(task, PCMK_ACTION_MONITOR, pcmk__str_casei)) { const char *rc = crm_element_value(rsc_op, PCMK__XA_RC_CODE); if (pcmk__strcase_any_of(rc, "0", "8", NULL)) { implied_monitor_start = counter; } } else if (pcmk__strcase_any_of(task, PCMK_ACTION_PROMOTE, PCMK_ACTION_DEMOTE, NULL)) { implied_clone_start = counter; } } if (*start_index == -1) { if (implied_clone_start != -1) { *start_index = implied_clone_start; } else if (implied_monitor_start != -1) { *start_index = implied_monitor_start; } } } // If resource history entry has shutdown lock, remember lock node and time static void unpack_shutdown_lock(const xmlNode *rsc_entry, pcmk_resource_t *rsc, const pcmk_node_t *node, pcmk_scheduler_t *scheduler) { time_t lock_time = 0; // When lock started (i.e. node shutdown time) if ((crm_element_value_epoch(rsc_entry, PCMK_OPT_SHUTDOWN_LOCK, &lock_time) == pcmk_ok) && (lock_time != 0)) { if ((scheduler->priv->shutdown_lock_ms > 0U) - && (get_effective_time(scheduler) + && (pcmk__scheduler_epoch_time(scheduler) > (lock_time + pcmk__timeout_ms2s(scheduler->priv->shutdown_lock_ms)))) { pcmk__rsc_info(rsc, "Shutdown lock for %s on %s expired", rsc->id, pcmk__node_name(node)); pe__clear_resource_history(rsc, node); } else { rsc->priv->lock_node = node; rsc->priv->lock_time = lock_time; } } } /*! * \internal * \brief Unpack one \c PCMK__XE_LRM_RESOURCE entry from a node's CIB status * * \param[in,out] node Node whose status is being unpacked * \param[in] rsc_entry \c PCMK__XE_LRM_RESOURCE XML being unpacked * \param[in,out] scheduler Scheduler data * * \return Resource corresponding to the entry, or NULL if no operation history */ static pcmk_resource_t * unpack_lrm_resource(pcmk_node_t *node, const xmlNode *lrm_resource, pcmk_scheduler_t *scheduler) { GList *gIter = NULL; int stop_index = -1; int start_index = -1; enum rsc_role_e req_role = pcmk_role_unknown; const char *rsc_id = pcmk__xe_id(lrm_resource); pcmk_resource_t *rsc = NULL; GList *op_list = NULL; GList *sorted_op_list = NULL; xmlNode *rsc_op = NULL; xmlNode *last_failure = NULL; enum pcmk__on_fail on_fail = pcmk__on_fail_ignore; enum rsc_role_e saved_role = pcmk_role_unknown; if (rsc_id == NULL) { pcmk__config_err("Ignoring invalid " PCMK__XE_LRM_RESOURCE " entry: No " PCMK_XA_ID); crm_log_xml_info(lrm_resource, "missing-id"); return NULL; } crm_trace("Unpacking " PCMK__XE_LRM_RESOURCE " for %s on %s", rsc_id, pcmk__node_name(node)); /* Build a list of individual PCMK__XE_LRM_RSC_OP entries, so we can sort * them */ for (rsc_op = pcmk__xe_first_child(lrm_resource, PCMK__XE_LRM_RSC_OP, NULL, NULL); rsc_op != NULL; rsc_op = pcmk__xe_next(rsc_op, PCMK__XE_LRM_RSC_OP)) { op_list = g_list_prepend(op_list, rsc_op); } if (!pcmk_is_set(scheduler->flags, pcmk__sched_shutdown_lock)) { if (op_list == NULL) { // If there are no operations, there is nothing to do return NULL; } } /* find the resource */ rsc = unpack_find_resource(scheduler, node, rsc_id); if (rsc == NULL) { if (op_list == NULL) { // If there are no operations, there is nothing to do return NULL; } else { rsc = process_orphan_resource(lrm_resource, node, scheduler); } } pcmk__assert(rsc != NULL); // Check whether the resource is "shutdown-locked" to this node if (pcmk_is_set(scheduler->flags, pcmk__sched_shutdown_lock)) { unpack_shutdown_lock(lrm_resource, rsc, node, scheduler); } /* process operations */ saved_role = rsc->priv->orig_role; rsc->priv->orig_role = pcmk_role_unknown; sorted_op_list = g_list_sort(op_list, sort_op_by_callid); for (gIter = sorted_op_list; gIter != NULL; gIter = gIter->next) { xmlNode *rsc_op = (xmlNode *) gIter->data; unpack_rsc_op(rsc, node, rsc_op, &last_failure, &on_fail); } /* create active recurring operations as optional */ calculate_active_ops(sorted_op_list, &start_index, &stop_index); process_recurring(node, rsc, start_index, stop_index, sorted_op_list, scheduler); /* no need to free the contents */ g_list_free(sorted_op_list); process_rsc_state(rsc, node, on_fail); if (get_target_role(rsc, &req_role)) { if ((rsc->priv->next_role == pcmk_role_unknown) || (req_role < rsc->priv->next_role)) { pe__set_next_role(rsc, req_role, PCMK_META_TARGET_ROLE); } else if (req_role > rsc->priv->next_role) { pcmk__rsc_info(rsc, "%s: Not overwriting calculated next role %s" " with requested next role %s", rsc->id, pcmk_role_text(rsc->priv->next_role), pcmk_role_text(req_role)); } } if (saved_role > rsc->priv->orig_role) { rsc->priv->orig_role = saved_role; } return rsc; } static void handle_removed_launched_resources(const xmlNode *lrm_rsc_list, pcmk_scheduler_t *scheduler) { for (const xmlNode *rsc_entry = pcmk__xe_first_child(lrm_rsc_list, PCMK__XE_LRM_RESOURCE, NULL, NULL); rsc_entry != NULL; rsc_entry = pcmk__xe_next(rsc_entry, PCMK__XE_LRM_RESOURCE)) { pcmk_resource_t *rsc; pcmk_resource_t *launcher = NULL; const char *rsc_id; const char *launcher_id = NULL; launcher_id = crm_element_value(rsc_entry, PCMK__META_CONTAINER); rsc_id = crm_element_value(rsc_entry, PCMK_XA_ID); if ((launcher_id == NULL) || (rsc_id == NULL)) { continue; } launcher = pe_find_resource(scheduler->priv->resources, launcher_id); if (launcher == NULL) { continue; } rsc = pe_find_resource(scheduler->priv->resources, rsc_id); if ((rsc == NULL) || (rsc->priv->launcher != NULL) || !pcmk_is_set(rsc->flags, pcmk__rsc_removed_launched)) { continue; } pcmk__rsc_trace(rsc, "Mapped launcher of removed resource %s to %s", rsc->id, launcher_id); rsc->priv->launcher = launcher; launcher->priv->launched = g_list_append(launcher->priv->launched, rsc); } } /*! * \internal * \brief Unpack one node's lrm status section * * \param[in,out] node Node whose status is being unpacked * \param[in] xml CIB node state XML * \param[in,out] scheduler Scheduler data */ static void unpack_node_lrm(pcmk_node_t *node, const xmlNode *xml, pcmk_scheduler_t *scheduler) { bool found_removed_launched_resource = false; // Drill down to PCMK__XE_LRM_RESOURCES section xml = pcmk__xe_first_child(xml, PCMK__XE_LRM, NULL, NULL); if (xml == NULL) { return; } xml = pcmk__xe_first_child(xml, PCMK__XE_LRM_RESOURCES, NULL, NULL); if (xml == NULL) { return; } // Unpack each PCMK__XE_LRM_RESOURCE entry for (const xmlNode *rsc_entry = pcmk__xe_first_child(xml, PCMK__XE_LRM_RESOURCE, NULL, NULL); rsc_entry != NULL; rsc_entry = pcmk__xe_next(rsc_entry, PCMK__XE_LRM_RESOURCE)) { pcmk_resource_t *rsc = unpack_lrm_resource(node, rsc_entry, scheduler); if ((rsc != NULL) && pcmk_is_set(rsc->flags, pcmk__rsc_removed_launched)) { found_removed_launched_resource = true; } } /* Now that all resource state has been unpacked for this node, map any * removed launched resources to their launchers. */ if (found_removed_launched_resource) { handle_removed_launched_resources(xml, scheduler); } } static void set_active(pcmk_resource_t *rsc) { const pcmk_resource_t *top = pe__const_top_resource(rsc, false); if (top && pcmk_is_set(top->flags, pcmk__rsc_promotable)) { rsc->priv->orig_role = pcmk_role_unpromoted; } else { rsc->priv->orig_role = pcmk_role_started; } } static void set_node_score(gpointer key, gpointer value, gpointer user_data) { pcmk_node_t *node = value; int *score = user_data; node->assign->score = *score; } #define XPATH_NODE_STATE "/" PCMK_XE_CIB "/" PCMK_XE_STATUS \ "/" PCMK__XE_NODE_STATE #define SUB_XPATH_LRM_RESOURCE "/" PCMK__XE_LRM \ "/" PCMK__XE_LRM_RESOURCES \ "/" PCMK__XE_LRM_RESOURCE #define SUB_XPATH_LRM_RSC_OP "/" PCMK__XE_LRM_RSC_OP static xmlNode * find_lrm_op(const char *resource, const char *op, const char *node, const char *source, int target_rc, pcmk_scheduler_t *scheduler) { GString *xpath = NULL; xmlNode *xml = NULL; CRM_CHECK((resource != NULL) && (op != NULL) && (node != NULL), return NULL); xpath = g_string_sized_new(256); pcmk__g_strcat(xpath, XPATH_NODE_STATE "[@" PCMK_XA_UNAME "='", node, "']" SUB_XPATH_LRM_RESOURCE "[@" PCMK_XA_ID "='", resource, "']" SUB_XPATH_LRM_RSC_OP "[@" PCMK_XA_OPERATION "='", op, "'", NULL); /* Need to check against transition_magic too? */ if ((source != NULL) && (strcmp(op, PCMK_ACTION_MIGRATE_TO) == 0)) { pcmk__g_strcat(xpath, " and @" PCMK__META_MIGRATE_TARGET "='", source, "']", NULL); } else if ((source != NULL) && (strcmp(op, PCMK_ACTION_MIGRATE_FROM) == 0)) { pcmk__g_strcat(xpath, " and @" PCMK__META_MIGRATE_SOURCE "='", source, "']", NULL); } else { g_string_append_c(xpath, ']'); } xml = get_xpath_object((const char *) xpath->str, scheduler->input, LOG_DEBUG); g_string_free(xpath, TRUE); if (xml && target_rc >= 0) { int rc = PCMK_OCF_UNKNOWN_ERROR; int status = PCMK_EXEC_ERROR; crm_element_value_int(xml, PCMK__XA_RC_CODE, &rc); crm_element_value_int(xml, PCMK__XA_OP_STATUS, &status); if ((rc != target_rc) || (status != PCMK_EXEC_DONE)) { return NULL; } } return xml; } static xmlNode * find_lrm_resource(const char *rsc_id, const char *node_name, pcmk_scheduler_t *scheduler) { GString *xpath = NULL; xmlNode *xml = NULL; CRM_CHECK((rsc_id != NULL) && (node_name != NULL), return NULL); xpath = g_string_sized_new(256); pcmk__g_strcat(xpath, XPATH_NODE_STATE "[@" PCMK_XA_UNAME "='", node_name, "']" SUB_XPATH_LRM_RESOURCE "[@" PCMK_XA_ID "='", rsc_id, "']", NULL); xml = get_xpath_object((const char *) xpath->str, scheduler->input, LOG_DEBUG); g_string_free(xpath, TRUE); return xml; } /*! * \internal * \brief Check whether a resource has no completed action history on a node * * \param[in,out] rsc Resource to check * \param[in] node_name Node to check * * \return true if \p rsc_id is unknown on \p node_name, otherwise false */ static bool unknown_on_node(pcmk_resource_t *rsc, const char *node_name) { bool result = false; xmlXPathObjectPtr search; char *xpath = NULL; xpath = crm_strdup_printf(XPATH_NODE_STATE "[@" PCMK_XA_UNAME "='%s']" SUB_XPATH_LRM_RESOURCE "[@" PCMK_XA_ID "='%s']" SUB_XPATH_LRM_RSC_OP "[@" PCMK__XA_RC_CODE "!='%d']", node_name, rsc->id, PCMK_OCF_UNKNOWN); search = xpath_search(rsc->priv->scheduler->input, xpath); result = (numXpathResults(search) == 0); freeXpathObject(search); free(xpath); return result; } /*! * \internal * \brief Check whether a probe/monitor indicating the resource was not running * on a node happened after some event * * \param[in] rsc_id Resource being checked * \param[in] node_name Node being checked * \param[in] xml_op Event that monitor is being compared to * \param[in,out] scheduler Scheduler data * * \return true if such a monitor happened after event, false otherwise */ static bool monitor_not_running_after(const char *rsc_id, const char *node_name, const xmlNode *xml_op, pcmk_scheduler_t *scheduler) { /* Any probe/monitor operation on the node indicating it was not running * there */ xmlNode *monitor = find_lrm_op(rsc_id, PCMK_ACTION_MONITOR, node_name, NULL, PCMK_OCF_NOT_RUNNING, scheduler); return (monitor != NULL) && (pe__is_newer_op(monitor, xml_op) > 0); } /*! * \internal * \brief Check whether any non-monitor operation on a node happened after some * event * * \param[in] rsc_id Resource being checked * \param[in] node_name Node being checked * \param[in] xml_op Event that non-monitor is being compared to * \param[in,out] scheduler Scheduler data * * \return true if such a operation happened after event, false otherwise */ static bool non_monitor_after(const char *rsc_id, const char *node_name, const xmlNode *xml_op, pcmk_scheduler_t *scheduler) { xmlNode *lrm_resource = NULL; lrm_resource = find_lrm_resource(rsc_id, node_name, scheduler); if (lrm_resource == NULL) { return false; } for (xmlNode *op = pcmk__xe_first_child(lrm_resource, PCMK__XE_LRM_RSC_OP, NULL, NULL); op != NULL; op = pcmk__xe_next(op, PCMK__XE_LRM_RSC_OP)) { const char * task = NULL; if (op == xml_op) { continue; } task = crm_element_value(op, PCMK_XA_OPERATION); if (pcmk__str_any_of(task, PCMK_ACTION_START, PCMK_ACTION_STOP, PCMK_ACTION_MIGRATE_TO, PCMK_ACTION_MIGRATE_FROM, NULL) && pe__is_newer_op(op, xml_op) > 0) { return true; } } return false; } /*! * \internal * \brief Check whether the resource has newer state on a node after a migration * attempt * * \param[in] rsc_id Resource being checked * \param[in] node_name Node being checked * \param[in] migrate_to Any migrate_to event that is being compared to * \param[in] migrate_from Any migrate_from event that is being compared to * \param[in,out] scheduler Scheduler data * * \return true if such a operation happened after event, false otherwise */ static bool newer_state_after_migrate(const char *rsc_id, const char *node_name, const xmlNode *migrate_to, const xmlNode *migrate_from, pcmk_scheduler_t *scheduler) { const xmlNode *xml_op = (migrate_from != NULL)? migrate_from : migrate_to; const char *source = crm_element_value(xml_op, PCMK__META_MIGRATE_SOURCE); /* It's preferred to compare to the migrate event on the same node if * existing, since call ids are more reliable. */ if ((xml_op != migrate_to) && (migrate_to != NULL) && pcmk__str_eq(node_name, source, pcmk__str_casei)) { xml_op = migrate_to; } /* If there's any newer non-monitor operation on the node, or any newer * probe/monitor operation on the node indicating it was not running there, * the migration events potentially no longer matter for the node. */ return non_monitor_after(rsc_id, node_name, xml_op, scheduler) || monitor_not_running_after(rsc_id, node_name, xml_op, scheduler); } /*! * \internal * \brief Parse migration source and target node names from history entry * * \param[in] entry Resource history entry for a migration action * \param[in] source_node If not NULL, source must match this node * \param[in] target_node If not NULL, target must match this node * \param[out] source_name Where to store migration source node name * \param[out] target_name Where to store migration target node name * * \return Standard Pacemaker return code */ static int get_migration_node_names(const xmlNode *entry, const pcmk_node_t *source_node, const pcmk_node_t *target_node, const char **source_name, const char **target_name) { *source_name = crm_element_value(entry, PCMK__META_MIGRATE_SOURCE); *target_name = crm_element_value(entry, PCMK__META_MIGRATE_TARGET); if ((*source_name == NULL) || (*target_name == NULL)) { pcmk__config_err("Ignoring resource history entry %s without " PCMK__META_MIGRATE_SOURCE " and " PCMK__META_MIGRATE_TARGET, pcmk__xe_id(entry)); return pcmk_rc_unpack_error; } if ((source_node != NULL) && !pcmk__str_eq(*source_name, source_node->priv->name, pcmk__str_casei|pcmk__str_null_matches)) { pcmk__config_err("Ignoring resource history entry %s because " PCMK__META_MIGRATE_SOURCE "='%s' does not match %s", pcmk__xe_id(entry), *source_name, pcmk__node_name(source_node)); return pcmk_rc_unpack_error; } if ((target_node != NULL) && !pcmk__str_eq(*target_name, target_node->priv->name, pcmk__str_casei|pcmk__str_null_matches)) { pcmk__config_err("Ignoring resource history entry %s because " PCMK__META_MIGRATE_TARGET "='%s' does not match %s", pcmk__xe_id(entry), *target_name, pcmk__node_name(target_node)); return pcmk_rc_unpack_error; } return pcmk_rc_ok; } /* * \internal * \brief Add a migration source to a resource's list of dangling migrations * * If the migrate_to and migrate_from actions in a live migration both * succeeded, but there is no stop on the source, the migration is considered * "dangling." Add the source to the resource's dangling migration list, which * will be used to schedule a stop on the source without affecting the target. * * \param[in,out] rsc Resource involved in migration * \param[in] node Migration source */ static void add_dangling_migration(pcmk_resource_t *rsc, const pcmk_node_t *node) { pcmk__rsc_trace(rsc, "Dangling migration of %s requires stop on %s", rsc->id, pcmk__node_name(node)); rsc->priv->orig_role = pcmk_role_stopped; rsc->priv->dangling_migration_sources = g_list_prepend(rsc->priv->dangling_migration_sources, (gpointer) node); } /*! * \internal * \brief Update resource role etc. after a successful migrate_to action * * \param[in,out] history Parsed action result history */ static void unpack_migrate_to_success(struct action_history *history) { /* A complete migration sequence is: * 1. migrate_to on source node (which succeeded if we get to this function) * 2. migrate_from on target node * 3. stop on source node * * If no migrate_from has happened, the migration is considered to be * "partial". If the migrate_from succeeded but no stop has happened, the * migration is considered to be "dangling". * * If a successful migrate_to and stop have happened on the source node, we * still need to check for a partial migration, due to scenarios (easier to * produce with batch-limit=1) like: * * - A resource is migrating from node1 to node2, and a migrate_to is * initiated for it on node1. * * - node2 goes into standby mode while the migrate_to is pending, which * aborts the transition. * * - Upon completion of the migrate_to, a new transition schedules a stop * on both nodes and a start on node1. * * - If the new transition is aborted for any reason while the resource is * stopping on node1, the transition after that stop completes will see * the migrate_to and stop on the source, but it's still a partial * migration, and the resource must be stopped on node2 because it is * potentially active there due to the migrate_to. * * We also need to take into account that either node's history may be * cleared at any point in the migration process. */ int from_rc = PCMK_OCF_OK; int from_status = PCMK_EXEC_PENDING; pcmk_node_t *target_node = NULL; xmlNode *migrate_from = NULL; const char *source = NULL; const char *target = NULL; bool source_newer_op = false; bool target_newer_state = false; bool active_on_target = false; pcmk_scheduler_t *scheduler = history->rsc->priv->scheduler; // Get source and target node names from XML if (get_migration_node_names(history->xml, history->node, NULL, &source, &target) != pcmk_rc_ok) { return; } // Check for newer state on the source source_newer_op = non_monitor_after(history->rsc->id, source, history->xml, scheduler); // Check for a migrate_from action from this source on the target migrate_from = find_lrm_op(history->rsc->id, PCMK_ACTION_MIGRATE_FROM, target, source, -1, scheduler); if (migrate_from != NULL) { if (source_newer_op) { /* There's a newer non-monitor operation on the source and a * migrate_from on the target, so this migrate_to is irrelevant to * the resource's state. */ return; } crm_element_value_int(migrate_from, PCMK__XA_RC_CODE, &from_rc); crm_element_value_int(migrate_from, PCMK__XA_OP_STATUS, &from_status); } /* If the resource has newer state on both the source and target after the * migration events, this migrate_to is irrelevant to the resource's state. */ target_newer_state = newer_state_after_migrate(history->rsc->id, target, history->xml, migrate_from, scheduler); if (source_newer_op && target_newer_state) { return; } /* Check for dangling migration (migrate_from succeeded but stop not done). * We know there's no stop because we already returned if the target has a * migrate_from and the source has any newer non-monitor operation. */ if ((from_rc == PCMK_OCF_OK) && (from_status == PCMK_EXEC_DONE)) { add_dangling_migration(history->rsc, history->node); return; } /* Without newer state, this migrate_to implies the resource is active. * (Clones are not allowed to migrate, so role can't be promoted.) */ history->rsc->priv->orig_role = pcmk_role_started; target_node = pcmk_find_node(scheduler, target); active_on_target = !target_newer_state && (target_node != NULL) && target_node->details->online; if (from_status != PCMK_EXEC_PENDING) { // migrate_from failed on target if (active_on_target) { native_add_running(history->rsc, target_node, scheduler, TRUE); } else { // Mark resource as failed, require recovery, and prevent migration pcmk__set_rsc_flags(history->rsc, pcmk__rsc_failed|pcmk__rsc_stop_if_failed); pcmk__clear_rsc_flags(history->rsc, pcmk__rsc_migratable); } return; } // The migrate_from is pending, complete but erased, or to be scheduled /* If there is no history at all for the resource on an online target, then * it was likely cleaned. Just return, and we'll schedule a probe. Once we * have the probe result, it will be reflected in target_newer_state. */ if ((target_node != NULL) && target_node->details->online && unknown_on_node(history->rsc, target)) { return; } if (active_on_target) { pcmk_node_t *source_node = pcmk_find_node(scheduler, source); native_add_running(history->rsc, target_node, scheduler, FALSE); if ((source_node != NULL) && source_node->details->online) { /* This is a partial migration: the migrate_to completed * successfully on the source, but the migrate_from has not * completed. Remember the source and target; if the newly * chosen target remains the same when we schedule actions * later, we may continue with the migration. */ history->rsc->priv->partial_migration_target = target_node; history->rsc->priv->partial_migration_source = source_node; } } else if (!source_newer_op) { // Mark resource as failed, require recovery, and prevent migration pcmk__set_rsc_flags(history->rsc, pcmk__rsc_failed|pcmk__rsc_stop_if_failed); pcmk__clear_rsc_flags(history->rsc, pcmk__rsc_migratable); } } /*! * \internal * \brief Update resource role etc. after a failed migrate_to action * * \param[in,out] history Parsed action result history */ static void unpack_migrate_to_failure(struct action_history *history) { xmlNode *target_migrate_from = NULL; const char *source = NULL; const char *target = NULL; pcmk_scheduler_t *scheduler = history->rsc->priv->scheduler; // Get source and target node names from XML if (get_migration_node_names(history->xml, history->node, NULL, &source, &target) != pcmk_rc_ok) { return; } /* If a migration failed, we have to assume the resource is active. Clones * are not allowed to migrate, so role can't be promoted. */ history->rsc->priv->orig_role = pcmk_role_started; // Check for migrate_from on the target target_migrate_from = find_lrm_op(history->rsc->id, PCMK_ACTION_MIGRATE_FROM, target, source, PCMK_OCF_OK, scheduler); if (/* If the resource state is unknown on the target, it will likely be * probed there. * Don't just consider it running there. We will get back here anyway in * case the probe detects it's running there. */ !unknown_on_node(history->rsc, target) /* If the resource has newer state on the target after the migration * events, this migrate_to no longer matters for the target. */ && !newer_state_after_migrate(history->rsc->id, target, history->xml, target_migrate_from, scheduler)) { /* The resource has no newer state on the target, so assume it's still * active there. * (if it is up). */ pcmk_node_t *target_node = pcmk_find_node(scheduler, target); if (target_node && target_node->details->online) { native_add_running(history->rsc, target_node, scheduler, FALSE); } } else if (!non_monitor_after(history->rsc->id, source, history->xml, scheduler)) { /* We know the resource has newer state on the target, but this * migrate_to still matters for the source as long as there's no newer * non-monitor operation there. */ // Mark node as having dangling migration so we can force a stop later history->rsc->priv->dangling_migration_sources = g_list_prepend(history->rsc->priv->dangling_migration_sources, (gpointer) history->node); } } /*! * \internal * \brief Update resource role etc. after a failed migrate_from action * * \param[in,out] history Parsed action result history */ static void unpack_migrate_from_failure(struct action_history *history) { xmlNode *source_migrate_to = NULL; const char *source = NULL; const char *target = NULL; pcmk_scheduler_t *scheduler = history->rsc->priv->scheduler; // Get source and target node names from XML if (get_migration_node_names(history->xml, NULL, history->node, &source, &target) != pcmk_rc_ok) { return; } /* If a migration failed, we have to assume the resource is active. Clones * are not allowed to migrate, so role can't be promoted. */ history->rsc->priv->orig_role = pcmk_role_started; // Check for a migrate_to on the source source_migrate_to = find_lrm_op(history->rsc->id, PCMK_ACTION_MIGRATE_TO, source, target, PCMK_OCF_OK, scheduler); if (/* If the resource state is unknown on the source, it will likely be * probed there. * Don't just consider it running there. We will get back here anyway in * case the probe detects it's running there. */ !unknown_on_node(history->rsc, source) /* If the resource has newer state on the source after the migration * events, this migrate_from no longer matters for the source. */ && !newer_state_after_migrate(history->rsc->id, source, source_migrate_to, history->xml, scheduler)) { /* The resource has no newer state on the source, so assume it's still * active there (if it is up). */ pcmk_node_t *source_node = pcmk_find_node(scheduler, source); if (source_node && source_node->details->online) { native_add_running(history->rsc, source_node, scheduler, TRUE); } } } /*! * \internal * \brief Add an action to cluster's list of failed actions * * \param[in,out] history Parsed action result history */ static void record_failed_op(struct action_history *history) { const pcmk_scheduler_t *scheduler = history->rsc->priv->scheduler; if (!(history->node->details->online)) { return; } for (const xmlNode *xIter = scheduler->priv->failed->children; xIter != NULL; xIter = xIter->next) { const char *key = pcmk__xe_history_key(xIter); const char *uname = crm_element_value(xIter, PCMK_XA_UNAME); if (pcmk__str_eq(history->key, key, pcmk__str_none) && pcmk__str_eq(uname, history->node->priv->name, pcmk__str_casei)) { crm_trace("Skipping duplicate entry %s on %s", history->key, pcmk__node_name(history->node)); return; } } crm_trace("Adding entry for %s on %s to failed action list", history->key, pcmk__node_name(history->node)); crm_xml_add(history->xml, PCMK_XA_UNAME, history->node->priv->name); crm_xml_add(history->xml, PCMK__XA_RSC_ID, history->rsc->id); pcmk__xml_copy(scheduler->priv->failed, history->xml); } static char * last_change_str(const xmlNode *xml_op) { time_t when; char *result = NULL; if (crm_element_value_epoch(xml_op, PCMK_XA_LAST_RC_CHANGE, &when) == pcmk_ok) { char *when_s = pcmk__epoch2str(&when, 0); const char *p = strchr(when_s, ' '); // Skip day of week to make message shorter if ((p != NULL) && (*(++p) != '\0')) { result = pcmk__str_copy(p); } free(when_s); } if (result == NULL) { result = pcmk__str_copy("unknown_time"); } return result; } /*! * \internal * \brief Ban a resource (or its clone if an anonymous instance) from all nodes * * \param[in,out] rsc Resource to ban */ static void ban_from_all_nodes(pcmk_resource_t *rsc) { int score = -PCMK_SCORE_INFINITY; const pcmk_scheduler_t *scheduler = rsc->priv->scheduler; if (rsc->priv->parent != NULL) { pcmk_resource_t *parent = uber_parent(rsc); if (pcmk__is_anonymous_clone(parent)) { /* For anonymous clones, if an operation with * PCMK_META_ON_FAIL=PCMK_VALUE_STOP fails for any instance, the * entire clone must stop. */ rsc = parent; } } // Ban the resource from all nodes crm_notice("%s will not be started under current conditions", rsc->id); if (rsc->priv->allowed_nodes != NULL) { g_hash_table_destroy(rsc->priv->allowed_nodes); } rsc->priv->allowed_nodes = pe__node_list2table(scheduler->nodes); g_hash_table_foreach(rsc->priv->allowed_nodes, set_node_score, &score); } /*! * \internal * \brief Get configured failure handling and role after failure for an action * * \param[in,out] history Unpacked action history entry * \param[out] on_fail Where to set configured failure handling * \param[out] fail_role Where to set to role after failure */ static void unpack_failure_handling(struct action_history *history, enum pcmk__on_fail *on_fail, enum rsc_role_e *fail_role) { xmlNode *config = pcmk__find_action_config(history->rsc, history->task, history->interval_ms, true); GHashTable *meta = pcmk__unpack_action_meta(history->rsc, history->node, history->task, history->interval_ms, config); const char *on_fail_str = g_hash_table_lookup(meta, PCMK_META_ON_FAIL); *on_fail = pcmk__parse_on_fail(history->rsc, history->task, history->interval_ms, on_fail_str); *fail_role = pcmk__role_after_failure(history->rsc, history->task, *on_fail, meta); g_hash_table_destroy(meta); } /*! * \internal * \brief Update resource role, failure handling, etc., after a failed action * * \param[in,out] history Parsed action result history * \param[in] config_on_fail Action failure handling from configuration * \param[in] fail_role Resource's role after failure of this action * \param[out] last_failure This will be set to the history XML * \param[in,out] on_fail Actual handling of action result */ static void unpack_rsc_op_failure(struct action_history *history, enum pcmk__on_fail config_on_fail, enum rsc_role_e fail_role, xmlNode **last_failure, enum pcmk__on_fail *on_fail) { bool is_probe = false; char *last_change_s = NULL; pcmk_scheduler_t *scheduler = history->rsc->priv->scheduler; *last_failure = history->xml; is_probe = pcmk_xe_is_probe(history->xml); last_change_s = last_change_str(history->xml); if (!pcmk_is_set(scheduler->flags, pcmk__sched_symmetric_cluster) && (history->exit_status == PCMK_OCF_NOT_INSTALLED)) { crm_trace("Unexpected result (%s%s%s) was recorded for " "%s of %s on %s at %s " QB_XS " exit-status=%d id=%s", crm_exit_str(history->exit_status), (pcmk__str_empty(history->exit_reason)? "" : ": "), pcmk__s(history->exit_reason, ""), (is_probe? "probe" : history->task), history->rsc->id, pcmk__node_name(history->node), last_change_s, history->exit_status, history->id); } else { pcmk__sched_warn(scheduler, "Unexpected result (%s%s%s) was recorded for %s of " "%s on %s at %s " QB_XS " exit-status=%d id=%s", crm_exit_str(history->exit_status), (pcmk__str_empty(history->exit_reason)? "" : ": "), pcmk__s(history->exit_reason, ""), (is_probe? "probe" : history->task), history->rsc->id, pcmk__node_name(history->node), last_change_s, history->exit_status, history->id); if (is_probe && (history->exit_status != PCMK_OCF_OK) && (history->exit_status != PCMK_OCF_NOT_RUNNING) && (history->exit_status != PCMK_OCF_RUNNING_PROMOTED)) { /* A failed (not just unexpected) probe result could mean the user * didn't know resources will be probed even where they can't run. */ crm_notice("If it is not possible for %s to run on %s, see " "the " PCMK_XA_RESOURCE_DISCOVERY " option for location " "constraints", history->rsc->id, pcmk__node_name(history->node)); } record_failed_op(history); } free(last_change_s); if (*on_fail < config_on_fail) { pcmk__rsc_trace(history->rsc, "on-fail %s -> %s for %s", pcmk__on_fail_text(*on_fail), pcmk__on_fail_text(config_on_fail), history->key); *on_fail = config_on_fail; } if (strcmp(history->task, PCMK_ACTION_STOP) == 0) { resource_location(history->rsc, history->node, -PCMK_SCORE_INFINITY, "__stop_fail__", scheduler); } else if (strcmp(history->task, PCMK_ACTION_MIGRATE_TO) == 0) { unpack_migrate_to_failure(history); } else if (strcmp(history->task, PCMK_ACTION_MIGRATE_FROM) == 0) { unpack_migrate_from_failure(history); } else if (strcmp(history->task, PCMK_ACTION_PROMOTE) == 0) { history->rsc->priv->orig_role = pcmk_role_promoted; } else if (strcmp(history->task, PCMK_ACTION_DEMOTE) == 0) { if (config_on_fail == pcmk__on_fail_block) { history->rsc->priv->orig_role = pcmk_role_promoted; pe__set_next_role(history->rsc, pcmk_role_stopped, "demote with " PCMK_META_ON_FAIL "=block"); } else if (history->exit_status == PCMK_OCF_NOT_RUNNING) { history->rsc->priv->orig_role = pcmk_role_stopped; } else { /* Staying in the promoted role would put the scheduler and * controller into a loop. Setting the role to unpromoted is not * dangerous because the resource will be stopped as part of * recovery, and any promotion will be ordered after that stop. */ history->rsc->priv->orig_role = pcmk_role_unpromoted; } } if (is_probe && (history->exit_status == PCMK_OCF_NOT_INSTALLED)) { /* leave stopped */ pcmk__rsc_trace(history->rsc, "Leaving %s stopped", history->rsc->id); history->rsc->priv->orig_role = pcmk_role_stopped; } else if (history->rsc->priv->orig_role < pcmk_role_started) { pcmk__rsc_trace(history->rsc, "Setting %s active", history->rsc->id); set_active(history->rsc); } pcmk__rsc_trace(history->rsc, "Resource %s: role=%s unclean=%s on_fail=%s fail_role=%s", history->rsc->id, pcmk_role_text(history->rsc->priv->orig_role), pcmk__btoa(history->node->details->unclean), pcmk__on_fail_text(config_on_fail), pcmk_role_text(fail_role)); if ((fail_role != pcmk_role_started) && (history->rsc->priv->next_role < fail_role)) { pe__set_next_role(history->rsc, fail_role, "failure"); } if (fail_role == pcmk_role_stopped) { ban_from_all_nodes(history->rsc); } } /*! * \internal * \brief Block a resource with a failed action if it cannot be recovered * * If resource action is a failed stop and fencing is not possible, mark the * resource as unmanaged and blocked, since recovery cannot be done. * * \param[in,out] history Parsed action history entry */ static void block_if_unrecoverable(struct action_history *history) { char *last_change_s = NULL; if (strcmp(history->task, PCMK_ACTION_STOP) != 0) { return; // All actions besides stop are always recoverable } if (pe_can_fence(history->node->priv->scheduler, history->node)) { return; // Failed stops are recoverable via fencing } last_change_s = last_change_str(history->xml); pcmk__sched_err(history->node->priv->scheduler, "No further recovery can be attempted for %s " "because %s on %s failed (%s%s%s) at %s " QB_XS " rc=%d id=%s", history->rsc->id, history->task, pcmk__node_name(history->node), crm_exit_str(history->exit_status), (pcmk__str_empty(history->exit_reason)? "" : ": "), pcmk__s(history->exit_reason, ""), last_change_s, history->exit_status, history->id); free(last_change_s); pcmk__clear_rsc_flags(history->rsc, pcmk__rsc_managed); pcmk__set_rsc_flags(history->rsc, pcmk__rsc_blocked); } /*! * \internal * \brief Update action history's execution status and why * * \param[in,out] history Parsed action history entry * \param[out] why Where to store reason for update * \param[in] value New value * \param[in] reason Description of why value was changed */ static inline void remap_because(struct action_history *history, const char **why, int value, const char *reason) { if (history->execution_status != value) { history->execution_status = value; *why = reason; } } /*! * \internal * \brief Remap informational monitor results and operation status * * For the monitor results, certain OCF codes are for providing extended information * to the user about services that aren't yet failed but not entirely healthy either. * These must be treated as the "normal" result by Pacemaker. * * For operation status, the action result can be used to determine an appropriate * status for the purposes of responding to the action. The status provided by the * executor is not directly usable since the executor does not know what was expected. * * \param[in,out] history Parsed action history entry * \param[in,out] on_fail What should be done about the result * \param[in] expired Whether result is expired * * \note If the result is remapped and the node is not shutting down or failed, * the operation will be recorded in the scheduler data's list of failed * operations to highlight it for the user. * * \note This may update the resource's current and next role. */ static void remap_operation(struct action_history *history, enum pcmk__on_fail *on_fail, bool expired) { bool is_probe = false; int orig_exit_status = history->exit_status; int orig_exec_status = history->execution_status; const char *why = NULL; const char *task = history->task; // Remap degraded results to their successful counterparts history->exit_status = pcmk__effective_rc(history->exit_status); if (history->exit_status != orig_exit_status) { why = "degraded result"; if (!expired && (!history->node->details->shutdown || history->node->details->online)) { record_failed_op(history); } } if (!pcmk__is_bundled(history->rsc) && pcmk_xe_mask_probe_failure(history->xml) && ((history->execution_status != PCMK_EXEC_DONE) || (history->exit_status != PCMK_OCF_NOT_RUNNING))) { history->execution_status = PCMK_EXEC_DONE; history->exit_status = PCMK_OCF_NOT_RUNNING; why = "equivalent probe result"; } /* If the executor reported an execution status of anything but done or * error, consider that final. But for done or error, we know better whether * it should be treated as a failure or not, because we know the expected * result. */ switch (history->execution_status) { case PCMK_EXEC_DONE: case PCMK_EXEC_ERROR: break; // These should be treated as node-fatal case PCMK_EXEC_NO_FENCE_DEVICE: case PCMK_EXEC_NO_SECRETS: remap_because(history, &why, PCMK_EXEC_ERROR_HARD, "node-fatal error"); goto remap_done; default: goto remap_done; } is_probe = pcmk_xe_is_probe(history->xml); if (is_probe) { task = "probe"; } if (history->expected_exit_status < 0) { /* Pre-1.0 Pacemaker versions, and Pacemaker 1.1.6 or earlier with * Heartbeat 2.0.7 or earlier as the cluster layer, did not include the * expected exit status in the transition key, which (along with the * similar case of a corrupted transition key in the CIB) will be * reported to this function as -1. Pacemaker 2.0+ does not support * rolling upgrades from those versions or processing of saved CIB files * from those versions, so we do not need to care much about this case. */ remap_because(history, &why, PCMK_EXEC_ERROR, "obsolete history format"); pcmk__config_warn("Expected result not found for %s on %s " "(corrupt or obsolete CIB?)", history->key, pcmk__node_name(history->node)); } else if (history->exit_status == history->expected_exit_status) { remap_because(history, &why, PCMK_EXEC_DONE, "expected result"); } else { remap_because(history, &why, PCMK_EXEC_ERROR, "unexpected result"); pcmk__rsc_debug(history->rsc, "%s on %s: expected %d (%s), got %d (%s%s%s)", history->key, pcmk__node_name(history->node), history->expected_exit_status, crm_exit_str(history->expected_exit_status), history->exit_status, crm_exit_str(history->exit_status), (pcmk__str_empty(history->exit_reason)? "" : ": "), pcmk__s(history->exit_reason, "")); } switch (history->exit_status) { case PCMK_OCF_OK: if (is_probe && (history->expected_exit_status == PCMK_OCF_NOT_RUNNING)) { char *last_change_s = last_change_str(history->xml); remap_because(history, &why, PCMK_EXEC_DONE, "probe"); pcmk__rsc_info(history->rsc, "Probe found %s active on %s at %s", history->rsc->id, pcmk__node_name(history->node), last_change_s); free(last_change_s); } break; case PCMK_OCF_NOT_RUNNING: if (is_probe || (history->expected_exit_status == history->exit_status) || !pcmk_is_set(history->rsc->flags, pcmk__rsc_managed)) { /* For probes, recurring monitors for the Stopped role, and * unmanaged resources, "not running" is not considered a * failure. */ remap_because(history, &why, PCMK_EXEC_DONE, "exit status"); history->rsc->priv->orig_role = pcmk_role_stopped; *on_fail = pcmk__on_fail_ignore; pe__set_next_role(history->rsc, pcmk_role_unknown, "not running"); } break; case PCMK_OCF_RUNNING_PROMOTED: if (is_probe && (history->exit_status != history->expected_exit_status)) { char *last_change_s = last_change_str(history->xml); remap_because(history, &why, PCMK_EXEC_DONE, "probe"); pcmk__rsc_info(history->rsc, "Probe found %s active and promoted on %s at %s", history->rsc->id, pcmk__node_name(history->node), last_change_s); free(last_change_s); } if (!expired || (history->exit_status == history->expected_exit_status)) { history->rsc->priv->orig_role = pcmk_role_promoted; } break; case PCMK_OCF_FAILED_PROMOTED: if (!expired) { history->rsc->priv->orig_role = pcmk_role_promoted; } remap_because(history, &why, PCMK_EXEC_ERROR, "exit status"); break; case PCMK_OCF_NOT_CONFIGURED: remap_because(history, &why, PCMK_EXEC_ERROR_FATAL, "exit status"); break; case PCMK_OCF_UNIMPLEMENT_FEATURE: { guint interval_ms = 0; crm_element_value_ms(history->xml, PCMK_META_INTERVAL, &interval_ms); if (interval_ms == 0) { if (!expired) { block_if_unrecoverable(history); } remap_because(history, &why, PCMK_EXEC_ERROR_HARD, "exit status"); } else { remap_because(history, &why, PCMK_EXEC_NOT_SUPPORTED, "exit status"); } } break; case PCMK_OCF_NOT_INSTALLED: case PCMK_OCF_INVALID_PARAM: case PCMK_OCF_INSUFFICIENT_PRIV: if (!expired) { block_if_unrecoverable(history); } remap_because(history, &why, PCMK_EXEC_ERROR_HARD, "exit status"); break; default: if (history->execution_status == PCMK_EXEC_DONE) { char *last_change_s = last_change_str(history->xml); crm_info("Treating unknown exit status %d from %s of %s " "on %s at %s as failure", history->exit_status, task, history->rsc->id, pcmk__node_name(history->node), last_change_s); remap_because(history, &why, PCMK_EXEC_ERROR, "unknown exit status"); free(last_change_s); } break; } remap_done: if (why != NULL) { pcmk__rsc_trace(history->rsc, "Remapped %s result from [%s: %s] to [%s: %s] " "because of %s", history->key, pcmk_exec_status_str(orig_exec_status), crm_exit_str(orig_exit_status), pcmk_exec_status_str(history->execution_status), crm_exit_str(history->exit_status), why); } } // return TRUE if start or monitor last failure but parameters changed static bool should_clear_for_param_change(const xmlNode *xml_op, const char *task, pcmk_resource_t *rsc, pcmk_node_t *node) { if (pcmk__str_any_of(task, PCMK_ACTION_START, PCMK_ACTION_MONITOR, NULL)) { if (pe__bundle_needs_remote_name(rsc)) { /* We haven't allocated resources yet, so we can't reliably * substitute addr parameters for the REMOTE_CONTAINER_HACK. * When that's needed, defer the check until later. */ - pe__add_param_check(xml_op, rsc, node, pcmk__check_last_failure, - rsc->priv->scheduler); + pcmk__add_param_check(xml_op, rsc, node, pcmk__check_last_failure, + rsc->priv->scheduler); } else { pcmk__op_digest_t *digest_data = NULL; digest_data = rsc_action_digest_cmp(rsc, xml_op, node, rsc->priv->scheduler); switch (digest_data->rc) { case pcmk__digest_unknown: crm_trace("Resource %s history entry %s on %s" " has no digest to compare", rsc->id, pcmk__xe_history_key(xml_op), node->priv->id); break; case pcmk__digest_match: break; default: return TRUE; } } } return FALSE; } // Order action after fencing of remote node, given connection rsc static void order_after_remote_fencing(pcmk_action_t *action, pcmk_resource_t *remote_conn, pcmk_scheduler_t *scheduler) { pcmk_node_t *remote_node = pcmk_find_node(scheduler, remote_conn->id); if (remote_node) { pcmk_action_t *fence = pe_fence_op(remote_node, NULL, TRUE, NULL, FALSE, scheduler); order_actions(fence, action, pcmk__ar_first_implies_then); } } static bool should_ignore_failure_timeout(const pcmk_resource_t *rsc, const char *task, guint interval_ms, bool is_last_failure) { /* Clearing failures of recurring monitors has special concerns. The * executor reports only changes in the monitor result, so if the * monitor is still active and still getting the same failure result, * that will go undetected after the failure is cleared. * * Also, the operation history will have the time when the recurring * monitor result changed to the given code, not the time when the * result last happened. * * @TODO We probably should clear such failures only when the failure * timeout has passed since the last occurrence of the failed result. * However we don't record that information. We could maybe approximate * that by clearing only if there is a more recent successful monitor or * stop result, but we don't even have that information at this point * since we are still unpacking the resource's operation history. * * This is especially important for remote connection resources with a * reconnect interval, so in that case, we skip clearing failures * if the remote node hasn't been fenced. */ if ((rsc->priv->remote_reconnect_ms > 0U) && pcmk_is_set(rsc->priv->scheduler->flags, pcmk__sched_fencing_enabled) && (interval_ms != 0) && pcmk__str_eq(task, PCMK_ACTION_MONITOR, pcmk__str_casei)) { pcmk_node_t *remote_node = pcmk_find_node(rsc->priv->scheduler, rsc->id); if (remote_node && !pcmk_is_set(remote_node->priv->flags, pcmk__node_remote_fenced)) { if (is_last_failure) { crm_info("Waiting to clear monitor failure for remote node %s" " until fencing has occurred", rsc->id); } return TRUE; } } return FALSE; } /*! * \internal * \brief Check operation age and schedule failure clearing when appropriate * * This function has two distinct purposes. The first is to check whether an * operation history entry is expired (i.e. the resource has a failure timeout, * the entry is older than the timeout, and the resource either has no fail * count or its fail count is entirely older than the timeout). The second is to * schedule fail count clearing when appropriate (i.e. the operation is expired * and either the resource has an expired fail count or the operation is a * last_failure for a remote connection resource with a reconnect interval, * or the operation is a last_failure for a start or monitor operation and the * resource's parameters have changed since the operation). * * \param[in,out] history Parsed action result history * * \return true if operation history entry is expired, otherwise false */ static bool check_operation_expiry(struct action_history *history) { bool expired = false; bool is_last_failure = pcmk__ends_with(history->id, "_last_failure_0"); time_t last_run = 0; int unexpired_fail_count = 0; const char *clear_reason = NULL; const guint expiration_sec = pcmk__timeout_ms2s(history->rsc->priv->failure_expiration_ms); pcmk_scheduler_t *scheduler = history->rsc->priv->scheduler; if (history->execution_status == PCMK_EXEC_NOT_INSTALLED) { pcmk__rsc_trace(history->rsc, "Resource history entry %s on %s is not expired: " "Not Installed does not expire", history->id, pcmk__node_name(history->node)); return false; // "Not installed" must always be cleared manually } if ((expiration_sec > 0) && (crm_element_value_epoch(history->xml, PCMK_XA_LAST_RC_CHANGE, &last_run) == 0)) { /* Resource has a PCMK_META_FAILURE_TIMEOUT and history entry has a * timestamp */ - time_t now = get_effective_time(scheduler); + time_t now = pcmk__scheduler_epoch_time(scheduler); time_t last_failure = 0; // Is this particular operation history older than the failure timeout? if ((now >= (last_run + expiration_sec)) && !should_ignore_failure_timeout(history->rsc, history->task, history->interval_ms, is_last_failure)) { expired = true; } // Does the resource as a whole have an unexpired fail count? unexpired_fail_count = pe_get_failcount(history->node, history->rsc, &last_failure, pcmk__fc_effective, history->xml); // Update scheduler recheck time according to *last* failure crm_trace("%s@%lld is %sexpired @%lld with unexpired_failures=%d " "expiration=%s last-failure@%lld", history->id, (long long) last_run, (expired? "" : "not "), (long long) now, unexpired_fail_count, pcmk__readable_interval(expiration_sec * 1000), (long long) last_failure); last_failure += expiration_sec + 1; if (unexpired_fail_count && (now < last_failure)) { - pe__update_recheck_time(last_failure, scheduler, - "fail count expiration"); + pcmk__update_recheck_time(last_failure, scheduler, + "fail count expiration"); } } if (expired) { if (pe_get_failcount(history->node, history->rsc, NULL, pcmk__fc_default, history->xml)) { // There is a fail count ignoring timeout if (unexpired_fail_count == 0) { // There is no fail count considering timeout clear_reason = "it expired"; } else { /* This operation is old, but there is an unexpired fail count. * In a properly functioning cluster, this should only be * possible if this operation is not a failure (otherwise the * fail count should be expired too), so this is really just a * failsafe. */ pcmk__rsc_trace(history->rsc, "Resource history entry %s on %s is not " "expired: Unexpired fail count", history->id, pcmk__node_name(history->node)); expired = false; } } else if (is_last_failure && (history->rsc->priv->remote_reconnect_ms > 0U)) { /* Clear any expired last failure when reconnect interval is set, * even if there is no fail count. */ clear_reason = "reconnect interval is set"; } } if (!expired && is_last_failure && should_clear_for_param_change(history->xml, history->task, history->rsc, history->node)) { clear_reason = "resource parameters have changed"; } if (clear_reason != NULL) { pcmk_action_t *clear_op = NULL; // Schedule clearing of the fail count clear_op = pe__clear_failcount(history->rsc, history->node, clear_reason, scheduler); if (pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled) && (history->rsc->priv->remote_reconnect_ms > 0)) { /* If we're clearing a remote connection due to a reconnect * interval, we want to wait until any scheduled fencing * completes. * * We could limit this to remote_node->details->unclean, but at * this point, that's always true (it won't be reliable until * after unpack_node_history() is done). */ crm_info("Clearing %s failure will wait until any scheduled " "fencing of %s completes", history->task, history->rsc->id); order_after_remote_fencing(clear_op, history->rsc, scheduler); } } if (expired && (history->interval_ms == 0) && pcmk__str_eq(history->task, PCMK_ACTION_MONITOR, pcmk__str_none)) { switch (history->exit_status) { case PCMK_OCF_OK: case PCMK_OCF_NOT_RUNNING: case PCMK_OCF_RUNNING_PROMOTED: case PCMK_OCF_DEGRADED: case PCMK_OCF_DEGRADED_PROMOTED: // Don't expire probes that return these values pcmk__rsc_trace(history->rsc, "Resource history entry %s on %s is not " "expired: Probe result", history->id, pcmk__node_name(history->node)); expired = false; break; } } return expired; } int pe__target_rc_from_xml(const xmlNode *xml_op) { int target_rc = 0; const char *key = crm_element_value(xml_op, PCMK__XA_TRANSITION_KEY); if (key == NULL) { return -1; } decode_transition_key(key, NULL, NULL, NULL, &target_rc); return target_rc; } /*! * \internal * \brief Update a resource's state for an action result * * \param[in,out] history Parsed action history entry * \param[in] exit_status Exit status to base new state on * \param[in] last_failure Resource's last_failure entry, if known * \param[in,out] on_fail Resource's current failure handling */ static void update_resource_state(struct action_history *history, int exit_status, const xmlNode *last_failure, enum pcmk__on_fail *on_fail) { bool clear_past_failure = false; if ((exit_status == PCMK_OCF_NOT_INSTALLED) || (!pcmk__is_bundled(history->rsc) && pcmk_xe_mask_probe_failure(history->xml))) { history->rsc->priv->orig_role = pcmk_role_stopped; } else if (exit_status == PCMK_OCF_NOT_RUNNING) { clear_past_failure = true; } else if (pcmk__str_eq(history->task, PCMK_ACTION_MONITOR, pcmk__str_none)) { if ((last_failure != NULL) && pcmk__str_eq(history->key, pcmk__xe_history_key(last_failure), pcmk__str_none)) { clear_past_failure = true; } if (history->rsc->priv->orig_role < pcmk_role_started) { set_active(history->rsc); } } else if (pcmk__str_eq(history->task, PCMK_ACTION_START, pcmk__str_none)) { history->rsc->priv->orig_role = pcmk_role_started; clear_past_failure = true; } else if (pcmk__str_eq(history->task, PCMK_ACTION_STOP, pcmk__str_none)) { history->rsc->priv->orig_role = pcmk_role_stopped; clear_past_failure = true; } else if (pcmk__str_eq(history->task, PCMK_ACTION_PROMOTE, pcmk__str_none)) { history->rsc->priv->orig_role = pcmk_role_promoted; clear_past_failure = true; } else if (pcmk__str_eq(history->task, PCMK_ACTION_DEMOTE, pcmk__str_none)) { if (*on_fail == pcmk__on_fail_demote) { /* Demote clears an error only if * PCMK_META_ON_FAIL=PCMK_VALUE_DEMOTE */ clear_past_failure = true; } history->rsc->priv->orig_role = pcmk_role_unpromoted; } else if (pcmk__str_eq(history->task, PCMK_ACTION_MIGRATE_FROM, pcmk__str_none)) { history->rsc->priv->orig_role = pcmk_role_started; clear_past_failure = true; } else if (pcmk__str_eq(history->task, PCMK_ACTION_MIGRATE_TO, pcmk__str_none)) { unpack_migrate_to_success(history); } else if (history->rsc->priv->orig_role < pcmk_role_started) { pcmk__rsc_trace(history->rsc, "%s active on %s", history->rsc->id, pcmk__node_name(history->node)); set_active(history->rsc); } if (!clear_past_failure) { return; } switch (*on_fail) { case pcmk__on_fail_stop: case pcmk__on_fail_ban: case pcmk__on_fail_standby_node: case pcmk__on_fail_fence_node: pcmk__rsc_trace(history->rsc, "%s (%s) is not cleared by a completed %s", history->rsc->id, pcmk__on_fail_text(*on_fail), history->task); break; case pcmk__on_fail_block: case pcmk__on_fail_ignore: case pcmk__on_fail_demote: case pcmk__on_fail_restart: case pcmk__on_fail_restart_container: *on_fail = pcmk__on_fail_ignore; pe__set_next_role(history->rsc, pcmk_role_unknown, "clear past failures"); break; case pcmk__on_fail_reset_remote: if (history->rsc->priv->remote_reconnect_ms == 0U) { /* With no reconnect interval, the connection is allowed to * start again after the remote node is fenced and * completely stopped. (With a reconnect interval, we wait * for the failure to be cleared entirely before attempting * to reconnect.) */ *on_fail = pcmk__on_fail_ignore; pe__set_next_role(history->rsc, pcmk_role_unknown, "clear past failures and reset remote"); } break; } } /*! * \internal * \brief Check whether a given history entry matters for resource state * * \param[in] history Parsed action history entry * * \return true if action can affect resource state, otherwise false */ static inline bool can_affect_state(struct action_history *history) { return pcmk__str_any_of(history->task, PCMK_ACTION_MONITOR, PCMK_ACTION_START, PCMK_ACTION_STOP, PCMK_ACTION_PROMOTE, PCMK_ACTION_DEMOTE, PCMK_ACTION_MIGRATE_TO, PCMK_ACTION_MIGRATE_FROM, "asyncmon", NULL); } /*! * \internal * \brief Unpack execution/exit status and exit reason from a history entry * * \param[in,out] history Action history entry to unpack * * \return Standard Pacemaker return code */ static int unpack_action_result(struct action_history *history) { if ((crm_element_value_int(history->xml, PCMK__XA_OP_STATUS, &(history->execution_status)) < 0) || (history->execution_status < PCMK_EXEC_PENDING) || (history->execution_status > PCMK_EXEC_MAX) || (history->execution_status == PCMK_EXEC_CANCELLED)) { pcmk__config_err("Ignoring resource history entry %s for %s on %s " "with invalid " PCMK__XA_OP_STATUS " '%s'", history->id, history->rsc->id, pcmk__node_name(history->node), pcmk__s(crm_element_value(history->xml, PCMK__XA_OP_STATUS), "")); return pcmk_rc_unpack_error; } if ((crm_element_value_int(history->xml, PCMK__XA_RC_CODE, &(history->exit_status)) < 0) || (history->exit_status < 0) || (history->exit_status > CRM_EX_MAX)) { pcmk__config_err("Ignoring resource history entry %s for %s on %s " "with invalid " PCMK__XA_RC_CODE " '%s'", history->id, history->rsc->id, pcmk__node_name(history->node), pcmk__s(crm_element_value(history->xml, PCMK__XA_RC_CODE), "")); return pcmk_rc_unpack_error; } history->exit_reason = crm_element_value(history->xml, PCMK_XA_EXIT_REASON); return pcmk_rc_ok; } /*! * \internal * \brief Process an action history entry whose result expired * * \param[in,out] history Parsed action history entry * \param[in] orig_exit_status Action exit status before remapping * * \return Standard Pacemaker return code (in particular, pcmk_rc_ok means the * entry needs no further processing) */ static int process_expired_result(struct action_history *history, int orig_exit_status) { if (!pcmk__is_bundled(history->rsc) && pcmk_xe_mask_probe_failure(history->xml) && (orig_exit_status != history->expected_exit_status)) { if (history->rsc->priv->orig_role <= pcmk_role_stopped) { history->rsc->priv->orig_role = pcmk_role_unknown; } crm_trace("Ignoring resource history entry %s for probe of %s on %s: " "Masked failure expired", history->id, history->rsc->id, pcmk__node_name(history->node)); return pcmk_rc_ok; } if (history->exit_status == history->expected_exit_status) { return pcmk_rc_undetermined; // Only failures expire } if (history->interval_ms == 0) { crm_notice("Ignoring resource history entry %s for %s of %s on %s: " "Expired failure", history->id, history->task, history->rsc->id, pcmk__node_name(history->node)); return pcmk_rc_ok; } if (history->node->details->online && !history->node->details->unclean) { /* Reschedule the recurring action. schedule_cancel() won't work at * this stage, so as a hacky workaround, forcibly change the restart * digest so pcmk__check_action_config() does what we want later. * * @TODO We should skip this if there is a newer successful monitor. * Also, this causes rescheduling only if the history entry * has a PCMK__XA_OP_DIGEST (which the expire-non-blocked-failure * scheduler regression test doesn't, but that may not be a * realistic scenario in production). */ crm_notice("Rescheduling %s-interval %s of %s on %s " "after failure expired", pcmk__readable_interval(history->interval_ms), history->task, history->rsc->id, pcmk__node_name(history->node)); crm_xml_add(history->xml, PCMK__XA_OP_RESTART_DIGEST, "calculated-failure-timeout"); return pcmk_rc_ok; } return pcmk_rc_undetermined; } /*! * \internal * \brief Process a masked probe failure * * \param[in,out] history Parsed action history entry * \param[in] orig_exit_status Action exit status before remapping * \param[in] last_failure Resource's last_failure entry, if known * \param[in,out] on_fail Resource's current failure handling */ static void mask_probe_failure(struct action_history *history, int orig_exit_status, const xmlNode *last_failure, enum pcmk__on_fail *on_fail) { pcmk_resource_t *ban_rsc = history->rsc; if (!pcmk_is_set(history->rsc->flags, pcmk__rsc_unique)) { ban_rsc = uber_parent(history->rsc); } crm_notice("Treating probe result '%s' for %s on %s as 'not running'", crm_exit_str(orig_exit_status), history->rsc->id, pcmk__node_name(history->node)); update_resource_state(history, history->expected_exit_status, last_failure, on_fail); crm_xml_add(history->xml, PCMK_XA_UNAME, history->node->priv->name); record_failed_op(history); resource_location(ban_rsc, history->node, -PCMK_SCORE_INFINITY, "masked-probe-failure", ban_rsc->priv->scheduler); } /*! * \internal Check whether a given failure is for a given pending action * * \param[in] history Parsed history entry for pending action * \param[in] last_failure Resource's last_failure entry, if known * * \return true if \p last_failure is failure of pending action in \p history, * otherwise false * \note Both \p history and \p last_failure must come from the same * \c PCMK__XE_LRM_RESOURCE block, as node and resource are assumed to be * the same. */ static bool failure_is_newer(const struct action_history *history, const xmlNode *last_failure) { guint failure_interval_ms = 0U; long long failure_change = 0LL; long long this_change = 0LL; if (last_failure == NULL) { return false; // Resource has no last_failure entry } if (!pcmk__str_eq(history->task, crm_element_value(last_failure, PCMK_XA_OPERATION), pcmk__str_none)) { return false; // last_failure is for different action } if ((crm_element_value_ms(last_failure, PCMK_META_INTERVAL, &failure_interval_ms) != pcmk_ok) || (history->interval_ms != failure_interval_ms)) { return false; // last_failure is for action with different interval } if ((pcmk__scan_ll(crm_element_value(history->xml, PCMK_XA_LAST_RC_CHANGE), &this_change, 0LL) != pcmk_rc_ok) || (pcmk__scan_ll(crm_element_value(last_failure, PCMK_XA_LAST_RC_CHANGE), &failure_change, 0LL) != pcmk_rc_ok) || (failure_change < this_change)) { return false; // Failure is not known to be newer } return true; } /*! * \internal * \brief Update a resource's role etc. for a pending action * * \param[in,out] history Parsed history entry for pending action * \param[in] last_failure Resource's last_failure entry, if known */ static void process_pending_action(struct action_history *history, const xmlNode *last_failure) { /* For recurring monitors, a failure is recorded only in RSC_last_failure_0, * and there might be a RSC_monitor_INTERVAL entry with the last successful * or pending result. * * If last_failure contains the failure of the pending recurring monitor * we're processing here, and is newer, the action is no longer pending. * (Pending results have call ID -1, which sorts last, so the last failure * if any should be known.) */ if (failure_is_newer(history, last_failure)) { return; } if (strcmp(history->task, PCMK_ACTION_START) == 0) { pcmk__set_rsc_flags(history->rsc, pcmk__rsc_start_pending); set_active(history->rsc); } else if (strcmp(history->task, PCMK_ACTION_PROMOTE) == 0) { history->rsc->priv->orig_role = pcmk_role_promoted; } else if ((strcmp(history->task, PCMK_ACTION_MIGRATE_TO) == 0) && history->node->details->unclean) { /* A migrate_to action is pending on a unclean source, so force a stop * on the target. */ const char *migrate_target = NULL; pcmk_node_t *target = NULL; migrate_target = crm_element_value(history->xml, PCMK__META_MIGRATE_TARGET); target = pcmk_find_node(history->rsc->priv->scheduler, migrate_target); if (target != NULL) { stop_action(history->rsc, target, FALSE); } } if (history->rsc->priv->pending_action != NULL) { /* There should never be multiple pending actions, but as a failsafe, * just remember the first one processed for display purposes. */ return; } if (pcmk_is_probe(history->task, history->interval_ms)) { /* Pending probes are currently never displayed, even if pending * operations are requested. If we ever want to change that, * enable the below and the corresponding part of * native.c:native_pending_action(). */ #if 0 history->rsc->private->pending_action = strdup("probe"); history->rsc->private->pending_node = history->node; #endif } else { history->rsc->priv->pending_action = strdup(history->task); history->rsc->priv->pending_node = history->node; } } static void unpack_rsc_op(pcmk_resource_t *rsc, pcmk_node_t *node, xmlNode *xml_op, xmlNode **last_failure, enum pcmk__on_fail *on_fail) { int old_rc = 0; bool expired = false; pcmk_resource_t *parent = rsc; enum rsc_role_e fail_role = pcmk_role_unknown; enum pcmk__on_fail failure_strategy = pcmk__on_fail_restart; struct action_history history = { .rsc = rsc, .node = node, .xml = xml_op, .execution_status = PCMK_EXEC_UNKNOWN, }; CRM_CHECK(rsc && node && xml_op, return); history.id = pcmk__xe_id(xml_op); if (history.id == NULL) { pcmk__config_err("Ignoring resource history entry for %s on %s " "without ID", rsc->id, pcmk__node_name(node)); return; } // Task and interval history.task = crm_element_value(xml_op, PCMK_XA_OPERATION); if (history.task == NULL) { pcmk__config_err("Ignoring resource history entry %s for %s on %s " "without " PCMK_XA_OPERATION, history.id, rsc->id, pcmk__node_name(node)); return; } crm_element_value_ms(xml_op, PCMK_META_INTERVAL, &(history.interval_ms)); if (!can_affect_state(&history)) { pcmk__rsc_trace(rsc, "Ignoring resource history entry %s for %s on %s " "with irrelevant action '%s'", history.id, rsc->id, pcmk__node_name(node), history.task); return; } if (unpack_action_result(&history) != pcmk_rc_ok) { return; // Error already logged } history.expected_exit_status = pe__target_rc_from_xml(xml_op); history.key = pcmk__xe_history_key(xml_op); crm_element_value_int(xml_op, PCMK__XA_CALL_ID, &(history.call_id)); pcmk__rsc_trace(rsc, "Unpacking %s (%s call %d on %s): %s (%s)", history.id, history.task, history.call_id, pcmk__node_name(node), pcmk_exec_status_str(history.execution_status), crm_exit_str(history.exit_status)); if (node->details->unclean) { pcmk__rsc_trace(rsc, "%s is running on %s, which is unclean (further action " "depends on value of stop's on-fail attribute)", rsc->id, pcmk__node_name(node)); } expired = check_operation_expiry(&history); old_rc = history.exit_status; remap_operation(&history, on_fail, expired); if (expired && (process_expired_result(&history, old_rc) == pcmk_rc_ok)) { goto done; } if (!pcmk__is_bundled(rsc) && pcmk_xe_mask_probe_failure(xml_op)) { mask_probe_failure(&history, old_rc, *last_failure, on_fail); goto done; } if (!pcmk_is_set(rsc->flags, pcmk__rsc_unique)) { parent = uber_parent(rsc); } switch (history.execution_status) { case PCMK_EXEC_PENDING: process_pending_action(&history, *last_failure); goto done; case PCMK_EXEC_DONE: update_resource_state(&history, history.exit_status, *last_failure, on_fail); goto done; case PCMK_EXEC_NOT_INSTALLED: unpack_failure_handling(&history, &failure_strategy, &fail_role); if (failure_strategy == pcmk__on_fail_ignore) { crm_warn("Cannot ignore failed %s of %s on %s: " "Resource agent doesn't exist " QB_XS " status=%d rc=%d id=%s", history.task, rsc->id, pcmk__node_name(node), history.execution_status, history.exit_status, history.id); /* Also for printing it as "FAILED" by marking it as * pcmk__rsc_failed later */ *on_fail = pcmk__on_fail_ban; } resource_location(parent, node, -PCMK_SCORE_INFINITY, "hard-error", rsc->priv->scheduler); unpack_rsc_op_failure(&history, failure_strategy, fail_role, last_failure, on_fail); goto done; case PCMK_EXEC_NOT_CONNECTED: if (pcmk__is_pacemaker_remote_node(node) && pcmk_is_set(node->priv->remote->flags, pcmk__rsc_managed)) { /* We should never get into a situation where a managed remote * connection resource is considered OK but a resource action * behind the connection gets a "not connected" status. But as a * fail-safe in case a bug or unusual circumstances do lead to * that, ensure the remote connection is considered failed. */ pcmk__set_rsc_flags(node->priv->remote, pcmk__rsc_failed|pcmk__rsc_stop_if_failed); } break; // Not done, do error handling case PCMK_EXEC_ERROR: case PCMK_EXEC_ERROR_HARD: case PCMK_EXEC_ERROR_FATAL: case PCMK_EXEC_TIMEOUT: case PCMK_EXEC_NOT_SUPPORTED: case PCMK_EXEC_INVALID: break; // Not done, do error handling default: // No other value should be possible at this point break; } unpack_failure_handling(&history, &failure_strategy, &fail_role); if ((failure_strategy == pcmk__on_fail_ignore) || ((failure_strategy == pcmk__on_fail_restart_container) && (strcmp(history.task, PCMK_ACTION_STOP) == 0))) { char *last_change_s = last_change_str(xml_op); crm_warn("Pretending failed %s (%s%s%s) of %s on %s at %s succeeded " QB_XS " %s", history.task, crm_exit_str(history.exit_status), (pcmk__str_empty(history.exit_reason)? "" : ": "), pcmk__s(history.exit_reason, ""), rsc->id, pcmk__node_name(node), last_change_s, history.id); free(last_change_s); update_resource_state(&history, history.expected_exit_status, *last_failure, on_fail); crm_xml_add(xml_op, PCMK_XA_UNAME, node->priv->name); pcmk__set_rsc_flags(rsc, pcmk__rsc_ignore_failure); record_failed_op(&history); if ((failure_strategy == pcmk__on_fail_restart_container) && (*on_fail <= pcmk__on_fail_restart)) { *on_fail = failure_strategy; } } else { unpack_rsc_op_failure(&history, failure_strategy, fail_role, last_failure, on_fail); if (history.execution_status == PCMK_EXEC_ERROR_HARD) { uint8_t log_level = LOG_ERR; if (history.exit_status == PCMK_OCF_NOT_INSTALLED) { log_level = LOG_NOTICE; } do_crm_log(log_level, "Preventing %s from restarting on %s because " "of hard failure (%s%s%s) " QB_XS " %s", parent->id, pcmk__node_name(node), crm_exit_str(history.exit_status), (pcmk__str_empty(history.exit_reason)? "" : ": "), pcmk__s(history.exit_reason, ""), history.id); resource_location(parent, node, -PCMK_SCORE_INFINITY, "hard-error", rsc->priv->scheduler); } else if (history.execution_status == PCMK_EXEC_ERROR_FATAL) { pcmk__sched_err(rsc->priv->scheduler, "Preventing %s from restarting anywhere because " "of fatal failure (%s%s%s) " QB_XS " %s", parent->id, crm_exit_str(history.exit_status), (pcmk__str_empty(history.exit_reason)? "" : ": "), pcmk__s(history.exit_reason, ""), history.id); resource_location(parent, NULL, -PCMK_SCORE_INFINITY, "fatal-error", rsc->priv->scheduler); } } done: pcmk__rsc_trace(rsc, "%s role on %s after %s is %s (next %s)", rsc->id, pcmk__node_name(node), history.id, pcmk_role_text(rsc->priv->orig_role), pcmk_role_text(rsc->priv->next_role)); } /*! * \internal * \brief Insert a node attribute with value into a \c GHashTable * * \param[in,out] key Key to insert (either freed or owned by * \p user_data upon return) * \param[in] value Value to insert (owned by \p user_data upon return) * \param[in] user_data \c GHashTable to insert into */ static gboolean insert_attr(gpointer key, gpointer value, gpointer user_data) { GHashTable *table = user_data; g_hash_table_insert(table, key, value); return TRUE; } static void add_node_attrs(const xmlNode *xml_obj, pcmk_node_t *node, bool overwrite, pcmk_scheduler_t *scheduler) { const char *cluster_name = NULL; const char *dc_id = crm_element_value(scheduler->input, PCMK_XA_DC_UUID); const pcmk_rule_input_t rule_input = { .now = scheduler->priv->now, }; pcmk__insert_dup(node->priv->attrs, CRM_ATTR_UNAME, node->priv->name); pcmk__insert_dup(node->priv->attrs, CRM_ATTR_ID, node->priv->id); if ((scheduler->dc_node == NULL) && pcmk__str_eq(node->priv->id, dc_id, pcmk__str_casei)) { scheduler->dc_node = node; pcmk__insert_dup(node->priv->attrs, CRM_ATTR_IS_DC, PCMK_VALUE_TRUE); } else if (!pcmk__same_node(node, scheduler->dc_node)) { pcmk__insert_dup(node->priv->attrs, CRM_ATTR_IS_DC, PCMK_VALUE_FALSE); } cluster_name = g_hash_table_lookup(scheduler->priv->options, PCMK_OPT_CLUSTER_NAME); if (cluster_name) { pcmk__insert_dup(node->priv->attrs, CRM_ATTR_CLUSTER_NAME, cluster_name); } if (overwrite) { /* @TODO Try to reorder some unpacking so that we don't need the * overwrite argument or to unpack into a temporary table */ GHashTable *unpacked = pcmk__strkey_table(free, free); pe__unpack_dataset_nvpairs(xml_obj, PCMK_XE_INSTANCE_ATTRIBUTES, &rule_input, unpacked, NULL, scheduler); g_hash_table_foreach_steal(unpacked, insert_attr, node->priv->attrs); g_hash_table_destroy(unpacked); } else { pe__unpack_dataset_nvpairs(xml_obj, PCMK_XE_INSTANCE_ATTRIBUTES, &rule_input, node->priv->attrs, NULL, scheduler); } pe__unpack_dataset_nvpairs(xml_obj, PCMK_XE_UTILIZATION, &rule_input, node->priv->utilization, NULL, scheduler); if (pcmk__node_attr(node, CRM_ATTR_SITE_NAME, NULL, pcmk__rsc_node_current) == NULL) { const char *site_name = pcmk__node_attr(node, "site-name", NULL, pcmk__rsc_node_current); if (site_name) { pcmk__insert_dup(node->priv->attrs, CRM_ATTR_SITE_NAME, site_name); } else if (cluster_name) { /* Default to cluster-name if unset */ pcmk__insert_dup(node->priv->attrs, CRM_ATTR_SITE_NAME, cluster_name); } } } static GList * extract_operations(const char *node, const char *rsc, xmlNode * rsc_entry, gboolean active_filter) { int counter = -1; int stop_index = -1; int start_index = -1; xmlNode *rsc_op = NULL; GList *gIter = NULL; GList *op_list = NULL; GList *sorted_op_list = NULL; /* extract operations */ op_list = NULL; sorted_op_list = NULL; for (rsc_op = pcmk__xe_first_child(rsc_entry, PCMK__XE_LRM_RSC_OP, NULL, NULL); rsc_op != NULL; rsc_op = pcmk__xe_next(rsc_op, PCMK__XE_LRM_RSC_OP)) { crm_xml_add(rsc_op, PCMK_XA_RESOURCE, rsc); crm_xml_add(rsc_op, PCMK_XA_UNAME, node); op_list = g_list_prepend(op_list, rsc_op); } if (op_list == NULL) { /* if there are no operations, there is nothing to do */ return NULL; } sorted_op_list = g_list_sort(op_list, sort_op_by_callid); /* create active recurring operations as optional */ if (active_filter == FALSE) { return sorted_op_list; } op_list = NULL; calculate_active_ops(sorted_op_list, &start_index, &stop_index); for (gIter = sorted_op_list; gIter != NULL; gIter = gIter->next) { xmlNode *rsc_op = (xmlNode *) gIter->data; counter++; if (start_index < stop_index) { crm_trace("Skipping %s: not active", pcmk__xe_id(rsc_entry)); break; } else if (counter < start_index) { crm_trace("Skipping %s: old", pcmk__xe_id(rsc_op)); continue; } op_list = g_list_append(op_list, rsc_op); } g_list_free(sorted_op_list); return op_list; } GList * find_operations(const char *rsc, const char *node, gboolean active_filter, pcmk_scheduler_t *scheduler) { GList *output = NULL; GList *intermediate = NULL; xmlNode *tmp = NULL; xmlNode *status = pcmk__xe_first_child(scheduler->input, PCMK_XE_STATUS, NULL, NULL); pcmk_node_t *this_node = NULL; xmlNode *node_state = NULL; CRM_CHECK(status != NULL, return NULL); for (node_state = pcmk__xe_first_child(status, PCMK__XE_NODE_STATE, NULL, NULL); node_state != NULL; node_state = pcmk__xe_next(node_state, PCMK__XE_NODE_STATE)) { const char *uname = crm_element_value(node_state, PCMK_XA_UNAME); if (node != NULL && !pcmk__str_eq(uname, node, pcmk__str_casei)) { continue; } this_node = pcmk_find_node(scheduler, uname); if(this_node == NULL) { CRM_LOG_ASSERT(this_node != NULL); continue; } else if (pcmk__is_pacemaker_remote_node(this_node)) { determine_remote_online_status(scheduler, this_node); } else { determine_online_status(node_state, this_node, scheduler); } if (this_node->details->online || pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled)) { /* offline nodes run no resources... * unless stonith is enabled in which case we need to * make sure rsc start events happen after the stonith */ xmlNode *lrm_rsc = NULL; tmp = pcmk__xe_first_child(node_state, PCMK__XE_LRM, NULL, NULL); tmp = pcmk__xe_first_child(tmp, PCMK__XE_LRM_RESOURCES, NULL, NULL); for (lrm_rsc = pcmk__xe_first_child(tmp, PCMK__XE_LRM_RESOURCE, NULL, NULL); lrm_rsc != NULL; lrm_rsc = pcmk__xe_next(lrm_rsc, PCMK__XE_LRM_RESOURCE)) { const char *rsc_id = crm_element_value(lrm_rsc, PCMK_XA_ID); if ((rsc != NULL) && !pcmk__str_eq(rsc_id, rsc, pcmk__str_none)) { continue; } intermediate = extract_operations(uname, rsc_id, lrm_rsc, active_filter); output = g_list_concat(output, intermediate); } } } return output; } diff --git a/lib/pengine/utils.c b/lib/pengine/utils.c index f6c25ced2f..c44e7c00de 100644 --- a/lib/pengine/utils.c +++ b/lib/pengine/utils.c @@ -1,926 +1,890 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include "pe_status_private.h" extern bool pcmk__is_daemon; gboolean ghash_free_str_str(gpointer key, gpointer value, gpointer user_data); /*! * \internal * \brief Check whether we can fence a particular node * * \param[in] scheduler Scheduler data * \param[in] node Name of node to check * * \return true if node can be fenced, false otherwise */ bool pe_can_fence(const pcmk_scheduler_t *scheduler, const pcmk_node_t *node) { if (pcmk__is_guest_or_bundle_node(node)) { /* A guest or bundle node is fenced by stopping its launcher, which is * possible if the launcher's host is either online or fenceable. */ pcmk_resource_t *rsc = node->priv->remote->priv->launcher; for (GList *n = rsc->priv->active_nodes; n != NULL; n = n->next) { pcmk_node_t *launcher_node = n->data; if (!launcher_node->details->online && !pe_can_fence(scheduler, launcher_node)) { return false; } } return true; } else if (!pcmk_is_set(scheduler->flags, pcmk__sched_fencing_enabled)) { return false; /* Turned off */ } else if (!pcmk_is_set(scheduler->flags, pcmk__sched_have_fencing)) { return false; /* No devices */ } else if (pcmk_is_set(scheduler->flags, pcmk__sched_quorate)) { return true; } else if (scheduler->no_quorum_policy == pcmk_no_quorum_ignore) { return true; } else if(node == NULL) { return false; } else if(node->details->online) { crm_notice("We can fence %s without quorum because they're in our membership", pcmk__node_name(node)); return true; } crm_trace("Cannot fence %s", pcmk__node_name(node)); return false; } /*! * \internal * \brief Copy a node object * * \param[in] this_node Node object to copy * * \return Newly allocated shallow copy of this_node * \note This function asserts on errors and is guaranteed to return non-NULL. * The caller is responsible for freeing the result using * pcmk__free_node_copy(). */ pcmk_node_t * pe__copy_node(const pcmk_node_t *this_node) { pcmk_node_t *new_node = NULL; pcmk__assert(this_node != NULL); new_node = pcmk__assert_alloc(1, sizeof(pcmk_node_t)); new_node->assign = pcmk__assert_alloc(1, sizeof(struct pcmk__node_assignment)); new_node->assign->probe_mode = this_node->assign->probe_mode; new_node->assign->score = this_node->assign->score; new_node->assign->count = this_node->assign->count; new_node->details = this_node->details; new_node->priv = this_node->priv; return new_node; } /*! * \internal * \brief Create a hash table of node copies from a list of nodes * * \param[in] list Node list * * \return Hash table equivalent of node list */ GHashTable * pe__node_list2table(const GList *list) { GHashTable *result = NULL; result = pcmk__strkey_table(NULL, pcmk__free_node_copy); for (const GList *gIter = list; gIter != NULL; gIter = gIter->next) { pcmk_node_t *new_node = NULL; new_node = pe__copy_node((const pcmk_node_t *) gIter->data); g_hash_table_insert(result, (gpointer) new_node->priv->id, new_node); } return result; } /*! * \internal * \brief Compare two nodes by name, with numeric portions sorted numerically * * Sort two node names case-insensitively like strcasecmp(), but with any * numeric portions of the name sorted numerically. For example, "node10" will * sort higher than "node9" but lower than "remotenode9". * * \param[in] a First node to compare (can be \c NULL) * \param[in] b Second node to compare (can be \c NULL) * * \retval -1 \c a comes before \c b (or \c a is \c NULL and \c b is not) * \retval 0 \c a and \c b are equal (or both are \c NULL) * \retval 1 \c a comes after \c b (or \c b is \c NULL and \c a is not) */ gint pe__cmp_node_name(gconstpointer a, gconstpointer b) { const pcmk_node_t *node1 = (const pcmk_node_t *) a; const pcmk_node_t *node2 = (const pcmk_node_t *) b; if ((node1 == NULL) && (node2 == NULL)) { return 0; } if (node1 == NULL) { return -1; } if (node2 == NULL) { return 1; } return pcmk__numeric_strcasecmp(node1->priv->name, node2->priv->name); } /*! * \internal * \brief Output node weights to stdout * * \param[in] rsc Use allowed nodes for this resource * \param[in] comment Text description to prefix lines with * \param[in] nodes If rsc is not specified, use these nodes * \param[in,out] scheduler Scheduler data */ static void pe__output_node_weights(const pcmk_resource_t *rsc, const char *comment, GHashTable *nodes, pcmk_scheduler_t *scheduler) { pcmk__output_t *out = scheduler->priv->out; // Sort the nodes so the output is consistent for regression tests GList *list = g_list_sort(g_hash_table_get_values(nodes), pe__cmp_node_name); for (const GList *gIter = list; gIter != NULL; gIter = gIter->next) { const pcmk_node_t *node = (const pcmk_node_t *) gIter->data; out->message(out, "node-weight", rsc, comment, node->priv->name, pcmk_readable_score(node->assign->score)); } g_list_free(list); } /*! * \internal * \brief Log node weights at trace level * * \param[in] file Caller's filename * \param[in] function Caller's function name * \param[in] line Caller's line number * \param[in] rsc If not NULL, include this resource's ID in logs * \param[in] comment Text description to prefix lines with * \param[in] nodes Nodes whose scores should be logged */ static void pe__log_node_weights(const char *file, const char *function, int line, const pcmk_resource_t *rsc, const char *comment, GHashTable *nodes) { GHashTableIter iter; pcmk_node_t *node = NULL; // Don't waste time if we're not tracing at this point pcmk__if_tracing({}, return); g_hash_table_iter_init(&iter, nodes); while (g_hash_table_iter_next(&iter, NULL, (void **) &node)) { if (rsc) { qb_log_from_external_source(function, file, "%s: %s allocation score on %s: %s", LOG_TRACE, line, 0, comment, rsc->id, pcmk__node_name(node), pcmk_readable_score(node->assign->score)); } else { qb_log_from_external_source(function, file, "%s: %s = %s", LOG_TRACE, line, 0, comment, pcmk__node_name(node), pcmk_readable_score(node->assign->score)); } } } /*! * \internal * \brief Log or output node weights * * \param[in] file Caller's filename * \param[in] function Caller's function name * \param[in] line Caller's line number * \param[in] to_log Log if true, otherwise output * \param[in] rsc If not NULL, use this resource's ID in logs, * and show scores recursively for any children * \param[in] comment Text description to prefix lines with * \param[in] nodes Nodes whose scores should be shown * \param[in,out] scheduler Scheduler data */ void pe__show_node_scores_as(const char *file, const char *function, int line, bool to_log, const pcmk_resource_t *rsc, const char *comment, GHashTable *nodes, pcmk_scheduler_t *scheduler) { if ((rsc != NULL) && pcmk_is_set(rsc->flags, pcmk__rsc_removed)) { // Don't show allocation scores for orphans return; } if (nodes == NULL) { // Nothing to show return; } if (to_log) { pe__log_node_weights(file, function, line, rsc, comment, nodes); } else { pe__output_node_weights(rsc, comment, nodes, scheduler); } if (rsc == NULL) { return; } // If this resource has children, repeat recursively for each for (GList *gIter = rsc->priv->children; gIter != NULL; gIter = gIter->next) { pcmk_resource_t *child = (pcmk_resource_t *) gIter->data; pe__show_node_scores_as(file, function, line, to_log, child, comment, child->priv->allowed_nodes, scheduler); } } /*! * \internal * \brief Compare two resources by priority * * \param[in] a First resource to compare (can be \c NULL) * \param[in] b Second resource to compare (can be \c NULL) * * \retval -1 a's priority > b's priority (or \c b is \c NULL and \c a is not) * \retval 0 a's priority == b's priority (or both \c a and \c b are \c NULL) * \retval 1 a's priority < b's priority (or \c a is \c NULL and \c b is not) */ gint pe__cmp_rsc_priority(gconstpointer a, gconstpointer b) { const pcmk_resource_t *resource1 = (const pcmk_resource_t *)a; const pcmk_resource_t *resource2 = (const pcmk_resource_t *)b; if (a == NULL && b == NULL) { return 0; } if (a == NULL) { return 1; } if (b == NULL) { return -1; } if (resource1->priv->priority > resource2->priv->priority) { return -1; } if (resource1->priv->priority < resource2->priv->priority) { return 1; } return 0; } static void resource_node_score(pcmk_resource_t *rsc, const pcmk_node_t *node, int score, const char *tag) { pcmk_node_t *match = NULL; if ((pcmk_is_set(rsc->flags, pcmk__rsc_exclusive_probes) || (node->assign->probe_mode == pcmk__probe_never)) && pcmk__str_eq(tag, "symmetric_default", pcmk__str_casei)) { /* This string comparision may be fragile, but exclusive resources and * exclusive nodes should not have the symmetric_default constraint * applied to them. */ return; } else { for (GList *gIter = rsc->priv->children; gIter != NULL; gIter = gIter->next) { pcmk_resource_t *child_rsc = (pcmk_resource_t *) gIter->data; resource_node_score(child_rsc, node, score, tag); } } match = g_hash_table_lookup(rsc->priv->allowed_nodes, node->priv->id); if (match == NULL) { match = pe__copy_node(node); g_hash_table_insert(rsc->priv->allowed_nodes, (gpointer) match->priv->id, match); } match->assign->score = pcmk__add_scores(match->assign->score, score); pcmk__rsc_trace(rsc, "Enabling %s preference (%s) for %s on %s (now %s)", tag, pcmk_readable_score(score), rsc->id, pcmk__node_name(node), pcmk_readable_score(match->assign->score)); } void resource_location(pcmk_resource_t *rsc, const pcmk_node_t *node, int score, const char *tag, pcmk_scheduler_t *scheduler) { if (node != NULL) { resource_node_score(rsc, node, score, tag); } else if (scheduler != NULL) { GList *gIter = scheduler->nodes; for (; gIter != NULL; gIter = gIter->next) { pcmk_node_t *node_iter = (pcmk_node_t *) gIter->data; resource_node_score(rsc, node_iter, score, tag); } } else { GHashTableIter iter; pcmk_node_t *node_iter = NULL; g_hash_table_iter_init(&iter, rsc->priv->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (void **)&node_iter)) { resource_node_score(rsc, node_iter, score, tag); } } if ((node == NULL) && (score == -PCMK_SCORE_INFINITY) && (rsc->priv->assigned_node != NULL)) { // @TODO Should this be more like pcmk__unassign_resource()? crm_info("Unassigning %s from %s", rsc->id, pcmk__node_name(rsc->priv->assigned_node)); pcmk__free_node_copy(rsc->priv->assigned_node); rsc->priv->assigned_node = NULL; } } -time_t -get_effective_time(pcmk_scheduler_t *scheduler) -{ - if(scheduler) { - if (scheduler->priv->now == NULL) { - crm_trace("Recording a new 'now'"); - scheduler->priv->now = crm_time_new(NULL); - } - return crm_time_get_seconds_since_epoch(scheduler->priv->now); - } - - crm_trace("Defaulting to 'now'"); - return time(NULL); -} - gboolean get_target_role(const pcmk_resource_t *rsc, enum rsc_role_e *role) { enum rsc_role_e local_role = pcmk_role_unknown; const char *value = g_hash_table_lookup(rsc->priv->meta, PCMK_META_TARGET_ROLE); CRM_CHECK(role != NULL, return FALSE); if (pcmk__str_eq(value, PCMK_ROLE_STARTED, pcmk__str_null_matches|pcmk__str_casei)) { return FALSE; } if (pcmk__str_eq(PCMK_VALUE_DEFAULT, value, pcmk__str_casei)) { // @COMPAT Deprecated since 2.1.8 pcmk__config_warn("Support for setting " PCMK_META_TARGET_ROLE " to the explicit value '" PCMK_VALUE_DEFAULT "' is deprecated and will be removed in a " "future release (just leave it unset)"); return FALSE; } local_role = pcmk_parse_role(value); if (local_role == pcmk_role_unknown) { pcmk__config_err("Ignoring '" PCMK_META_TARGET_ROLE "' for %s " "because '%s' is not valid", rsc->id, value); return FALSE; } else if (local_role > pcmk_role_started) { if (pcmk_is_set(pe__const_top_resource(rsc, false)->flags, pcmk__rsc_promotable)) { if (local_role > pcmk_role_unpromoted) { /* This is what we'd do anyway, just leave the default to avoid messing up the placement algorithm */ return FALSE; } } else { pcmk__config_err("Ignoring '" PCMK_META_TARGET_ROLE "' for %s " "because '%s' only makes sense for promotable " "clones", rsc->id, value); return FALSE; } } *role = local_role; return TRUE; } gboolean order_actions(pcmk_action_t *first, pcmk_action_t *then, uint32_t flags) { GList *gIter = NULL; pcmk__related_action_t *wrapper = NULL; GList *list = NULL; if (flags == pcmk__ar_none) { return FALSE; } if ((first == NULL) || (then == NULL)) { return FALSE; } crm_trace("Creating action wrappers for ordering: %s then %s", first->uuid, then->uuid); /* Ensure we never create a dependency on ourselves... it's happened */ pcmk__assert(first != then); /* Filter dups, otherwise update_action_states() has too much work to do */ gIter = first->actions_after; for (; gIter != NULL; gIter = gIter->next) { pcmk__related_action_t *after = gIter->data; if ((after->action == then) && pcmk_any_flags_set(after->flags, flags)) { return FALSE; } } wrapper = pcmk__assert_alloc(1, sizeof(pcmk__related_action_t)); wrapper->action = then; wrapper->flags = flags; list = first->actions_after; list = g_list_prepend(list, wrapper); first->actions_after = list; wrapper = pcmk__assert_alloc(1, sizeof(pcmk__related_action_t)); wrapper->action = first; wrapper->flags = flags; list = then->actions_before; list = g_list_prepend(list, wrapper); then->actions_before = list; return TRUE; } void destroy_ticket(gpointer data) { pcmk__ticket_t *ticket = data; if (ticket->state) { g_hash_table_destroy(ticket->state); } free(ticket->id); free(ticket); } pcmk__ticket_t * ticket_new(const char *ticket_id, pcmk_scheduler_t *scheduler) { pcmk__ticket_t *ticket = NULL; if (pcmk__str_empty(ticket_id)) { return NULL; } if (scheduler->priv->ticket_constraints == NULL) { scheduler->priv->ticket_constraints = pcmk__strkey_table(free, destroy_ticket); } ticket = g_hash_table_lookup(scheduler->priv->ticket_constraints, ticket_id); if (ticket == NULL) { ticket = calloc(1, sizeof(pcmk__ticket_t)); if (ticket == NULL) { pcmk__sched_err(scheduler, "Cannot allocate ticket '%s'", ticket_id); return NULL; } crm_trace("Creating ticket entry for %s", ticket_id); ticket->id = strdup(ticket_id); ticket->last_granted = -1; ticket->state = pcmk__strkey_table(free, free); g_hash_table_insert(scheduler->priv->ticket_constraints, pcmk__str_copy(ticket->id), ticket); } return ticket; } const char * rsc_printable_id(const pcmk_resource_t *rsc) { if (pcmk_is_set(rsc->flags, pcmk__rsc_unique)) { return rsc->id; } return pcmk__xe_id(rsc->priv->xml); } void pe__clear_resource_flags_recursive(pcmk_resource_t *rsc, uint64_t flags) { pcmk__clear_rsc_flags(rsc, flags); for (GList *gIter = rsc->priv->children; gIter != NULL; gIter = gIter->next) { pe__clear_resource_flags_recursive((pcmk_resource_t *) gIter->data, flags); } } void pe__clear_resource_flags_on_all(pcmk_scheduler_t *scheduler, uint64_t flag) { for (GList *lpc = scheduler->priv->resources; lpc != NULL; lpc = lpc->next) { pcmk_resource_t *r = (pcmk_resource_t *) lpc->data; pe__clear_resource_flags_recursive(r, flag); } } void pe__set_resource_flags_recursive(pcmk_resource_t *rsc, uint64_t flags) { pcmk__set_rsc_flags(rsc, flags); for (GList *gIter = rsc->priv->children; gIter != NULL; gIter = gIter->next) { pe__set_resource_flags_recursive((pcmk_resource_t *) gIter->data, flags); } } void trigger_unfencing(pcmk_resource_t *rsc, pcmk_node_t *node, const char *reason, pcmk_action_t *dependency, pcmk_scheduler_t *scheduler) { if (!pcmk_is_set(scheduler->flags, pcmk__sched_enable_unfencing)) { /* No resources require it */ return; } else if ((rsc != NULL) && !pcmk_is_set(rsc->flags, pcmk__rsc_fence_device)) { /* Wasn't a stonith device */ return; } else if(node && node->details->online && node->details->unclean == FALSE && node->details->shutdown == FALSE) { pcmk_action_t *unfence = pe_fence_op(node, PCMK_ACTION_ON, FALSE, reason, FALSE, scheduler); if(dependency) { order_actions(unfence, dependency, pcmk__ar_ordered); } } else if(rsc) { GHashTableIter iter; g_hash_table_iter_init(&iter, rsc->priv->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (void **)&node)) { if(node->details->online && node->details->unclean == FALSE && node->details->shutdown == FALSE) { trigger_unfencing(rsc, node, reason, dependency, scheduler); } } } } /*! * \internal * \brief Check whether shutdown has been requested for a node * * \param[in] node Node to check * * \return TRUE if node has shutdown attribute set and nonzero, FALSE otherwise * \note This differs from simply using node->details->shutdown in that it can * be used before that has been determined (and in fact to determine it), * and it can also be used to distinguish requested shutdown from implicit * shutdown of remote nodes by virtue of their connection stopping. */ bool pe__shutdown_requested(const pcmk_node_t *node) { const char *shutdown = pcmk__node_attr(node, PCMK__NODE_ATTR_SHUTDOWN, NULL, pcmk__rsc_node_current); return !pcmk__str_eq(shutdown, "0", pcmk__str_null_matches); } -/*! - * \internal - * \brief Update "recheck by" time in scheduler data - * - * \param[in] recheck Epoch time when recheck should happen - * \param[in,out] scheduler Scheduler data - * \param[in] reason What time is being updated for (for logs) - */ -void -pe__update_recheck_time(time_t recheck, pcmk_scheduler_t *scheduler, - const char *reason) -{ - if ((recheck > get_effective_time(scheduler)) - && ((scheduler->priv->recheck_by == 0) - || (scheduler->priv->recheck_by > recheck))) { - scheduler->priv->recheck_by = recheck; - crm_debug("Updated next scheduler recheck to %s for %s", - pcmk__trim(ctime(&recheck)), reason); - } -} - /*! * \internal * \brief Extract nvpair blocks contained by a CIB XML element into a hash table * * \param[in] xml_obj XML element containing blocks of nvpair elements * \param[in] set_name If not NULL, only use blocks of this element * \param[in] rule_input Values used to evaluate rule criteria * (node_attrs member must be NULL if \p set_name * is PCMK_XE_META_ATTRIBUTES) * \param[out] hash Where to store extracted name/value pairs * \param[in] always_first If not NULL, process block with this ID first * \param[in,out] scheduler Scheduler data containing \p xml_obj */ void pe__unpack_dataset_nvpairs(const xmlNode *xml_obj, const char *set_name, const pcmk_rule_input_t *rule_input, GHashTable *hash, const char *always_first, pcmk_scheduler_t *scheduler) { crm_time_t *next_change = NULL; CRM_CHECK((set_name != NULL) && (rule_input != NULL) && (hash != NULL) && (scheduler != NULL), return); // Node attribute expressions are not allowed for meta-attributes CRM_CHECK((rule_input->node_attrs == NULL) || (strcmp(set_name, PCMK_XE_META_ATTRIBUTES) != 0), return); if (xml_obj == NULL) { return; } next_change = crm_time_new_undefined(); pcmk_unpack_nvpair_blocks(xml_obj, set_name, always_first, rule_input, hash, next_change); if (crm_time_is_defined(next_change)) { time_t recheck = (time_t) crm_time_get_seconds_since_epoch(next_change); - pe__update_recheck_time(recheck, scheduler, "rule evaluation"); + pcmk__update_recheck_time(recheck, scheduler, "rule evaluation"); } crm_time_free(next_change); } bool pe__resource_is_disabled(const pcmk_resource_t *rsc) { const char *target_role = NULL; CRM_CHECK(rsc != NULL, return false); target_role = g_hash_table_lookup(rsc->priv->meta, PCMK_META_TARGET_ROLE); if (target_role) { // If invalid, we've already logged an error when unpacking enum rsc_role_e target_role_e = pcmk_parse_role(target_role); if ((target_role_e == pcmk_role_stopped) || ((target_role_e == pcmk_role_unpromoted) && pcmk_is_set(pe__const_top_resource(rsc, false)->flags, pcmk__rsc_promotable))) { return true; } } return false; } /*! * \internal * \brief Check whether a resource is running only on given node * * \param[in] rsc Resource to check * \param[in] node Node to check * * \return true if \p rsc is running only on \p node, otherwise false */ bool pe__rsc_running_on_only(const pcmk_resource_t *rsc, const pcmk_node_t *node) { return (rsc != NULL) && pcmk__list_of_1(rsc->priv->active_nodes) && pcmk__same_node((const pcmk_node_t *) rsc->priv->active_nodes->data, node); } bool pe__rsc_running_on_any(pcmk_resource_t *rsc, GList *node_list) { if (rsc != NULL) { for (GList *ele = rsc->priv->active_nodes; ele; ele = ele->next) { pcmk_node_t *node = (pcmk_node_t *) ele->data; if (pcmk__str_in_list(node->priv->name, node_list, pcmk__str_star_matches|pcmk__str_casei)) { return true; } } } return false; } bool pcmk__rsc_filtered_by_node(pcmk_resource_t *rsc, GList *only_node) { return rsc->priv->fns->active(rsc, FALSE) && !pe__rsc_running_on_any(rsc, only_node); } GList * pe__filter_rsc_list(GList *rscs, GList *filter) { GList *retval = NULL; for (GList *gIter = rscs; gIter; gIter = gIter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) gIter->data; /* I think the second condition is safe here for all callers of this * function. If not, it needs to move into pe__node_text. */ if (pcmk__str_in_list(rsc_printable_id(rsc), filter, pcmk__str_star_matches) || ((rsc->priv->parent != NULL) && pcmk__str_in_list(rsc_printable_id(rsc->priv->parent), filter, pcmk__str_star_matches))) { retval = g_list_prepend(retval, rsc); } } return retval; } GList * pe__build_node_name_list(pcmk_scheduler_t *scheduler, const char *s) { GList *nodes = NULL; if (pcmk__str_eq(s, "*", pcmk__str_null_matches)) { /* Nothing was given so return a list of all node names. Or, '*' was * given. This would normally fall into the pe__unames_with_tag branch * where it will return an empty list. Catch it here instead. */ nodes = g_list_prepend(nodes, strdup("*")); } else { pcmk_node_t *node = pcmk_find_node(scheduler, s); if (node) { /* The given string was a valid uname for a node. Return a * singleton list containing just that uname. */ nodes = g_list_prepend(nodes, strdup(s)); } else { /* The given string was not a valid uname. It's either a tag or * it's a typo or something. In the first case, we'll return a * list of all the unames of the nodes with the given tag. In the * second case, we'll return a NULL pointer and nothing will * get displayed. */ nodes = pe__unames_with_tag(scheduler, s); } } return nodes; } GList * pe__build_rsc_list(pcmk_scheduler_t *scheduler, const char *s) { GList *resources = NULL; if (pcmk__str_eq(s, "*", pcmk__str_null_matches)) { resources = g_list_prepend(resources, strdup("*")); } else { const uint32_t flags = pcmk_rsc_match_history|pcmk_rsc_match_basename; pcmk_resource_t *rsc = pe_find_resource_with_flags(scheduler->priv->resources, s, flags); if (rsc) { /* A colon in the name we were given means we're being asked to filter * on a specific instance of a cloned resource. Put that exact string * into the filter list. Otherwise, use the printable ID of whatever * resource was found that matches what was asked for. */ if (strstr(s, ":") != NULL) { resources = g_list_prepend(resources, strdup(rsc->id)); } else { resources = g_list_prepend(resources, strdup(rsc_printable_id(rsc))); } } else { /* The given string was not a valid resource name. It's a tag or a * typo or something. See pe__build_node_name_list() for more * detail. */ resources = pe__rscs_with_tag(scheduler, s); } } return resources; } xmlNode * pe__failed_probe_for_rsc(const pcmk_resource_t *rsc, const char *name) { const pcmk_resource_t *parent = pe__const_top_resource(rsc, false); const char *rsc_id = rsc->id; const pcmk_scheduler_t *scheduler = rsc->priv->scheduler; if (pcmk__is_clone(parent)) { rsc_id = pe__clone_child_id(parent); } for (xmlNode *xml_op = pcmk__xe_first_child(scheduler->priv->failed, NULL, NULL, NULL); xml_op != NULL; xml_op = pcmk__xe_next(xml_op, NULL)) { const char *value = NULL; char *op_id = NULL; /* This resource operation is not a failed probe. */ if (!pcmk_xe_mask_probe_failure(xml_op)) { continue; } /* This resource operation was not run on the given node. Note that if name is * NULL, this will always succeed. */ value = crm_element_value(xml_op, PCMK__META_ON_NODE); if (value == NULL || !pcmk__str_eq(value, name, pcmk__str_casei|pcmk__str_null_matches)) { continue; } if (!parse_op_key(pcmk__xe_history_key(xml_op), &op_id, NULL, NULL)) { continue; // This history entry is missing an operation key } /* This resource operation's ID does not match the rsc_id we are looking for. */ if (!pcmk__str_eq(op_id, rsc_id, pcmk__str_none)) { free(op_id); continue; } free(op_id); return xml_op; } return NULL; } diff --git a/tools/crm_mon.c b/tools/crm_mon.c index 149ff7737e..5f104d6192 100644 --- a/tools/crm_mon.c +++ b/tools/crm_mon.c @@ -1,2188 +1,2188 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include // pcmk__ends_with_ext() #include #include #include #include #include #include #include #include #include #include #include #include #include #include // stonith__* #include "crm_mon.h" #define SUMMARY "Provides a summary of cluster's current state.\n\n" \ "Outputs varying levels of detail in a number of different formats." /* * Definitions indicating which items to print */ static uint32_t show; static uint32_t show_opts = pcmk_show_pending; /* * Definitions indicating how to output */ static mon_output_format_t output_format = mon_output_unset; /* other globals */ static GIOChannel *io_channel = NULL; static GMainLoop *mainloop = NULL; static guint reconnect_timer = 0; static mainloop_timer_t *refresh_timer = NULL; static enum pcmk_pacemakerd_state pcmkd_state = pcmk_pacemakerd_state_invalid; static cib_t *cib = NULL; static stonith_t *st = NULL; static xmlNode *current_cib = NULL; static GError *error = NULL; static pcmk__common_args_t *args = NULL; static pcmk__output_t *out = NULL; static GOptionContext *context = NULL; static gchar **processed_args = NULL; static time_t last_refresh = 0; volatile crm_trigger_t *refresh_trigger = NULL; static pcmk_scheduler_t *scheduler = NULL; static enum pcmk__fence_history fence_history = pcmk__fence_history_none; int interactive_fence_level = 0; static pcmk__supported_format_t formats[] = { #if PCMK__ENABLE_CURSES CRM_MON_SUPPORTED_FORMAT_CURSES, #endif PCMK__SUPPORTED_FORMAT_HTML, PCMK__SUPPORTED_FORMAT_NONE, PCMK__SUPPORTED_FORMAT_TEXT, PCMK__SUPPORTED_FORMAT_XML, { NULL, NULL, NULL } }; PCMK__OUTPUT_ARGS("crm-mon-disconnected", "const char *", "enum pcmk_pacemakerd_state") static int crm_mon_disconnected_default(pcmk__output_t *out, va_list args) { return pcmk_rc_no_output; } PCMK__OUTPUT_ARGS("crm-mon-disconnected", "const char *", "enum pcmk_pacemakerd_state") static int crm_mon_disconnected_html(pcmk__output_t *out, va_list args) { const char *desc = va_arg(args, const char *); enum pcmk_pacemakerd_state state = (enum pcmk_pacemakerd_state) va_arg(args, int); if (out->dest != stdout) { out->reset(out); } pcmk__output_create_xml_text_node(out, PCMK__XE_SPAN, "Not connected to CIB"); if (desc != NULL) { pcmk__output_create_xml_text_node(out, PCMK__XE_SPAN, ": "); pcmk__output_create_xml_text_node(out, PCMK__XE_SPAN, desc); } if (state != pcmk_pacemakerd_state_invalid) { const char *state_s = pcmk__pcmkd_state_enum2friendly(state); pcmk__output_create_xml_text_node(out, PCMK__XE_SPAN, " ("); pcmk__output_create_xml_text_node(out, PCMK__XE_SPAN, state_s); pcmk__output_create_xml_text_node(out, PCMK__XE_SPAN, ")"); } out->finish(out, CRM_EX_DISCONNECT, true, NULL); return pcmk_rc_ok; } PCMK__OUTPUT_ARGS("crm-mon-disconnected", "const char *", "enum pcmk_pacemakerd_state") static int crm_mon_disconnected_text(pcmk__output_t *out, va_list args) { const char *desc = va_arg(args, const char *); enum pcmk_pacemakerd_state state = (enum pcmk_pacemakerd_state) va_arg(args, int); int rc = pcmk_rc_ok; if (out->dest != stdout) { out->reset(out); } if (state != pcmk_pacemakerd_state_invalid) { rc = out->info(out, "Not connected to CIB%s%s (%s)", (desc != NULL)? ": " : "", pcmk__s(desc, ""), pcmk__pcmkd_state_enum2friendly(state)); } else { rc = out->info(out, "Not connected to CIB%s%s", (desc != NULL)? ": " : "", pcmk__s(desc, "")); } out->finish(out, CRM_EX_DISCONNECT, true, NULL); return rc; } PCMK__OUTPUT_ARGS("crm-mon-disconnected", "const char *", "enum pcmk_pacemakerd_state") static int crm_mon_disconnected_xml(pcmk__output_t *out, va_list args) { const char *desc = va_arg(args, const char *); enum pcmk_pacemakerd_state state = (enum pcmk_pacemakerd_state) va_arg(args, int); const char *state_s = NULL; if (out->dest != stdout) { out->reset(out); } if (state != pcmk_pacemakerd_state_invalid) { state_s = pcmk_pacemakerd_api_daemon_state_enum2text(state); } pcmk__output_create_xml_node(out, PCMK_XE_CRM_MON_DISCONNECTED, PCMK_XA_DESCRIPTION, desc, PCMK_XA_PACEMAKERD_STATE, state_s, NULL); out->finish(out, CRM_EX_DISCONNECT, true, NULL); return pcmk_rc_ok; } static pcmk__message_entry_t fmt_functions[] = { { "crm-mon-disconnected", "default", crm_mon_disconnected_default }, { "crm-mon-disconnected", "html", crm_mon_disconnected_html }, { "crm-mon-disconnected", "text", crm_mon_disconnected_text }, { "crm-mon-disconnected", "xml", crm_mon_disconnected_xml }, { NULL, NULL, NULL }, }; #define RECONNECT_MSECS 5000 struct { guint reconnect_ms; enum mon_exec_mode exec_mode; gboolean fence_connect; gboolean print_pending; gboolean show_bans; gboolean watch_fencing; char *pid_file; char *external_agent; char *external_recipient; char *neg_location_prefix; char *only_node; char *only_rsc; GSList *user_includes_excludes; GSList *includes_excludes; } options = { .reconnect_ms = RECONNECT_MSECS, .exec_mode = mon_exec_unset, .fence_connect = TRUE, }; static crm_exit_t clean_up(crm_exit_t exit_code); static void crm_diff_update(const char *event, xmlNode * msg); static void clean_up_on_connection_failure(int rc); static int mon_refresh_display(gpointer user_data); static int setup_cib_connection(void); static int setup_fencer_connection(void); static int setup_api_connections(void); static void mon_st_callback_event(stonith_t * st, stonith_event_t * e); static void mon_st_callback_display(stonith_t * st, stonith_event_t * e); static void refresh_after_event(gboolean data_updated, gboolean enforce); static uint32_t all_includes(mon_output_format_t fmt) { if ((fmt == mon_output_plain) || (fmt == mon_output_console)) { return ~pcmk_section_options; } else { return pcmk_section_all; } } static uint32_t default_includes(mon_output_format_t fmt) { switch (fmt) { case mon_output_plain: case mon_output_console: case mon_output_html: return pcmk_section_summary |pcmk_section_nodes |pcmk_section_resources |pcmk_section_failures; case mon_output_xml: return all_includes(fmt); default: return 0; } } struct { const char *name; uint32_t bit; } sections[] = { { "attributes", pcmk_section_attributes }, { "bans", pcmk_section_bans }, { "counts", pcmk_section_counts }, { "dc", pcmk_section_dc }, { "failcounts", pcmk_section_failcounts }, { "failures", pcmk_section_failures }, { PCMK_VALUE_FENCING, pcmk_section_fencing_all }, { "fencing-failed", pcmk_section_fence_failed }, { "fencing-pending", pcmk_section_fence_pending }, { "fencing-succeeded", pcmk_section_fence_worked }, { "maint-mode", pcmk_section_maint_mode }, { "nodes", pcmk_section_nodes }, { "operations", pcmk_section_operations }, { "options", pcmk_section_options }, { "resources", pcmk_section_resources }, { "stack", pcmk_section_stack }, { "summary", pcmk_section_summary }, { "tickets", pcmk_section_tickets }, { "times", pcmk_section_times }, { NULL } }; static uint32_t find_section_bit(const char *name) { for (int i = 0; sections[i].name != NULL; i++) { if (pcmk__str_eq(sections[i].name, name, pcmk__str_casei)) { return sections[i].bit; } } return 0; } static gboolean apply_exclude(const gchar *excludes, GError **error) { char **parts = NULL; gboolean result = TRUE; parts = g_strsplit(excludes, ",", 0); for (char **s = parts; *s != NULL; s++) { uint32_t bit = find_section_bit(*s); if (pcmk__str_eq(*s, "all", pcmk__str_none)) { show = 0; } else if (pcmk__str_eq(*s, PCMK_VALUE_NONE, pcmk__str_none)) { show = all_includes(output_format); } else if (bit != 0) { show &= ~bit; } else { g_set_error(error, PCMK__EXITC_ERROR, CRM_EX_USAGE, "--exclude options: all, attributes, bans, counts, dc, " "failcounts, failures, fencing, fencing-failed, " "fencing-pending, fencing-succeeded, maint-mode, nodes, " PCMK_VALUE_NONE ", operations, options, resources, " "stack, summary, tickets, times"); result = FALSE; break; } } g_strfreev(parts); return result; } static gboolean apply_include(const gchar *includes, GError **error) { char **parts = NULL; gboolean result = TRUE; parts = g_strsplit(includes, ",", 0); for (char **s = parts; *s != NULL; s++) { uint32_t bit = find_section_bit(*s); if (pcmk__str_eq(*s, "all", pcmk__str_none)) { show = all_includes(output_format); } else if (pcmk__starts_with(*s, "bans")) { show |= pcmk_section_bans; if (options.neg_location_prefix != NULL) { free(options.neg_location_prefix); options.neg_location_prefix = NULL; } if (strlen(*s) > 4 && (*s)[4] == ':') { options.neg_location_prefix = strdup(*s+5); } } else if (pcmk__str_any_of(*s, PCMK_VALUE_DEFAULT, "defaults", NULL)) { show |= default_includes(output_format); } else if (pcmk__str_eq(*s, PCMK_VALUE_NONE, pcmk__str_none)) { show = 0; } else if (bit != 0) { show |= bit; } else { g_set_error(error, PCMK__EXITC_ERROR, CRM_EX_USAGE, "--include options: all, attributes, bans[:PREFIX], counts, dc, " PCMK_VALUE_DEFAULT ", failcounts, failures, fencing, " "fencing-failed, fencing-pending, fencing-succeeded, " "maint-mode, nodes, " PCMK_VALUE_NONE ", operations, " "options, resources, stack, summary, tickets, times"); result = FALSE; break; } } g_strfreev(parts); return result; } static gboolean apply_include_exclude(GSList *lst, GError **error) { gboolean rc = TRUE; GSList *node = lst; while (node != NULL) { char *s = node->data; if (pcmk__starts_with(s, "--include=")) { rc = apply_include(s+10, error); } else if (pcmk__starts_with(s, "-I=")) { rc = apply_include(s+3, error); } else if (pcmk__starts_with(s, "--exclude=")) { rc = apply_exclude(s+10, error); } else if (pcmk__starts_with(s, "-U=")) { rc = apply_exclude(s+3, error); } if (rc != TRUE) { break; } node = node->next; } return rc; } static gboolean user_include_exclude_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { char *s = crm_strdup_printf("%s=%s", option_name, optarg); options.user_includes_excludes = g_slist_append(options.user_includes_excludes, s); return TRUE; } static gboolean include_exclude_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { char *s = crm_strdup_printf("%s=%s", option_name, optarg); options.includes_excludes = g_slist_append(options.includes_excludes, s); return TRUE; } static gboolean as_xml_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { pcmk__str_update(&args->output_ty, "xml"); output_format = mon_output_legacy_xml; return TRUE; } static gboolean fence_history_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { if (optarg == NULL) { interactive_fence_level = 2; } else { pcmk__scan_min_int(optarg, &interactive_fence_level, 0); } switch (interactive_fence_level) { case 3: options.fence_connect = TRUE; fence_history = pcmk__fence_history_full; return include_exclude_cb("--include", PCMK_VALUE_FENCING, data, err); case 2: options.fence_connect = TRUE; fence_history = pcmk__fence_history_full; return include_exclude_cb("--include", PCMK_VALUE_FENCING, data, err); case 1: options.fence_connect = TRUE; fence_history = pcmk__fence_history_full; return include_exclude_cb("--include", "fencing-failed,fencing-pending", data, err); case 0: options.fence_connect = FALSE; fence_history = pcmk__fence_history_none; return include_exclude_cb("--exclude", PCMK_VALUE_FENCING, data, err); default: g_set_error(err, PCMK__EXITC_ERROR, CRM_EX_INVALID_PARAM, "Fence history must be 0-3"); return FALSE; } } static gboolean group_by_node_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { show_opts |= pcmk_show_rscs_by_node; return TRUE; } static gboolean hide_headers_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { return user_include_exclude_cb("--exclude", "summary", data, err); } static gboolean inactive_resources_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { show_opts |= pcmk_show_inactive_rscs; return TRUE; } static gboolean print_brief_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { show_opts |= pcmk_show_brief; return TRUE; } static gboolean print_detail_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { show_opts |= pcmk_show_details; return TRUE; } static gboolean print_description_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { show_opts |= pcmk_show_description; return TRUE; } static gboolean print_timing_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { show_opts |= pcmk_show_timing; return user_include_exclude_cb("--include", "operations", data, err); } static gboolean reconnect_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { int rc = crm_get_msec(optarg); if (rc == -1) { g_set_error(err, PCMK__EXITC_ERROR, CRM_EX_INVALID_PARAM, "Invalid value for -i: %s", optarg); return FALSE; } else { pcmk_parse_interval_spec(optarg, &options.reconnect_ms); if (options.exec_mode != mon_exec_daemonized) { // Reconnect interval applies to daemonized too, so don't override options.exec_mode = mon_exec_update; } } return TRUE; } /*! * \internal * \brief Enable one-shot mode * * \param[in] option_name Name of option being parsed (ignored) * \param[in] optarg Value to be parsed (ignored) * \param[in] data User data (ignored) * \param[out] err Where to store error (ignored) */ static gboolean one_shot_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { options.exec_mode = mon_exec_one_shot; return TRUE; } /*! * \internal * \brief Enable daemonized mode * * \param[in] option_name Name of option being parsed (ignored) * \param[in] optarg Value to be parsed (ignored) * \param[in] data User data (ignored) * \param[out] err Where to store error (ignored) */ static gboolean daemonize_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { options.exec_mode = mon_exec_daemonized; return TRUE; } static gboolean show_attributes_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { return user_include_exclude_cb("--include", "attributes", data, err); } static gboolean show_bans_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { if (optarg != NULL) { char *s = crm_strdup_printf("bans:%s", optarg); gboolean rc = user_include_exclude_cb("--include", s, data, err); free(s); return rc; } else { return user_include_exclude_cb("--include", "bans", data, err); } } static gboolean show_failcounts_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { return user_include_exclude_cb("--include", "failcounts", data, err); } static gboolean show_operations_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { return user_include_exclude_cb("--include", "failcounts,operations", data, err); } static gboolean show_tickets_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { return user_include_exclude_cb("--include", "tickets", data, err); } static gboolean use_cib_file_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { setenv("CIB_file", optarg, 1); options.exec_mode = mon_exec_one_shot; return TRUE; } #define INDENT " " /* *INDENT-OFF* */ static GOptionEntry addl_entries[] = { { "interval", 'i', 0, G_OPTION_ARG_CALLBACK, reconnect_cb, "Update frequency (default is 5 seconds)", "TIMESPEC" }, { "one-shot", '1', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, one_shot_cb, "Display the cluster status once and exit", NULL }, { "daemonize", 'd', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, daemonize_cb, "Run in the background as a daemon.\n" INDENT "Requires at least one of --output-to and --external-agent.", NULL }, { "pid-file", 'p', 0, G_OPTION_ARG_FILENAME, &options.pid_file, "(Advanced) Daemon pid file location", "FILE" }, { "external-agent", 'E', 0, G_OPTION_ARG_FILENAME, &options.external_agent, "A program to run when resource operations take place", "FILE" }, { "external-recipient", 'e', 0, G_OPTION_ARG_STRING, &options.external_recipient, "A recipient for your program (assuming you want the program to send something to someone).", "RCPT" }, { "watch-fencing", 'W', 0, G_OPTION_ARG_NONE, &options.watch_fencing, "Listen for fencing events. For use with --external-agent.", NULL }, { "xml-file", 'x', G_OPTION_FLAG_HIDDEN, G_OPTION_ARG_CALLBACK, use_cib_file_cb, NULL, NULL }, { NULL } }; static GOptionEntry display_entries[] = { { "include", 'I', 0, G_OPTION_ARG_CALLBACK, user_include_exclude_cb, "A list of sections to include in the output.\n" INDENT "See `Output Control` help for more information.", "SECTION(s)" }, { "exclude", 'U', 0, G_OPTION_ARG_CALLBACK, user_include_exclude_cb, "A list of sections to exclude from the output.\n" INDENT "See `Output Control` help for more information.", "SECTION(s)" }, { "node", 0, 0, G_OPTION_ARG_STRING, &options.only_node, "When displaying information about nodes, show only what's related to the given\n" INDENT "node, or to all nodes tagged with the given tag", "NODE" }, { "resource", 0, 0, G_OPTION_ARG_STRING, &options.only_rsc, "When displaying information about resources, show only what's related to the given\n" INDENT "resource, or to all resources tagged with the given tag", "RSC" }, { "group-by-node", 'n', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, group_by_node_cb, "Group resources by node", NULL }, { "inactive", 'r', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, inactive_resources_cb, "Display inactive resources", NULL }, { "failcounts", 'f', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, show_failcounts_cb, "Display resource fail counts", NULL }, { "operations", 'o', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, show_operations_cb, "Display resource operation history", NULL }, { "timing-details", 't', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, print_timing_cb, "Display resource operation history with timing details", NULL }, { "tickets", 'c', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, show_tickets_cb, "Display cluster tickets", NULL }, { "fence-history", 'm', G_OPTION_FLAG_OPTIONAL_ARG, G_OPTION_ARG_CALLBACK, fence_history_cb, "Show fence history:\n" INDENT "0=off, 1=failures and pending (default without option),\n" INDENT "2=add successes (default without value for option),\n" INDENT "3=show full history without reduction to most recent of each flavor", "LEVEL" }, { "neg-locations", 'L', G_OPTION_FLAG_OPTIONAL_ARG, G_OPTION_ARG_CALLBACK, show_bans_cb, "Display negative location constraints [optionally filtered by id prefix]", NULL }, { "show-node-attributes", 'A', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, show_attributes_cb, "Display node attributes", NULL }, { "hide-headers", 'D', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, hide_headers_cb, "Hide all headers", NULL }, { "show-detail", 'R', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, print_detail_cb, "Show more details (node IDs, individual clone instances)", NULL }, { "show-description", 0, G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, print_description_cb, "Show resource descriptions", NULL }, { "brief", 'b', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, print_brief_cb, "Brief output", NULL }, { "pending", 'j', G_OPTION_FLAG_HIDDEN, G_OPTION_ARG_NONE, &options.print_pending, "Display pending state if '" PCMK_META_RECORD_PENDING "' is enabled", NULL }, { NULL } }; static GOptionEntry deprecated_entries[] = { /* @COMPAT resource-agents <4.15.0 uses --as-xml, so removing this option * must wait until we no longer support building on any platforms that ship * the older agents. */ { "as-xml", 'X', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, as_xml_cb, "Write cluster status as XML to stdout. This will enable one-shot mode.\n" INDENT "Use --output-as=xml instead.", NULL }, { NULL } }; /* *INDENT-ON* */ /* Reconnect to the CIB and fencing agent after reconnect_ms has passed. This sounds * like it would be more broadly useful, but only ever happens after a disconnect via * mon_cib_connection_destroy. */ static gboolean reconnect_after_timeout(gpointer data) { #if PCMK__ENABLE_CURSES if (output_format == mon_output_console) { clear(); refresh(); } #endif out->transient(out, "Reconnecting..."); if (setup_api_connections() == pcmk_rc_ok) { // Trigger redrawing the screen (needs reconnect_timer == 0) reconnect_timer = 0; refresh_after_event(FALSE, TRUE); return G_SOURCE_REMOVE; } out->message(out, "crm-mon-disconnected", "Latest connection attempt failed", pcmkd_state); reconnect_timer = pcmk__create_timer(options.reconnect_ms, reconnect_after_timeout, NULL); return G_SOURCE_REMOVE; } /* Called from various places when we are disconnected from the CIB or from the * fencing agent. If the CIB connection is still valid, this function will also * attempt to sign off and reconnect. */ static void mon_cib_connection_destroy(gpointer user_data) { const char *msg = "Connection to the cluster lost"; pcmkd_state = pcmk_pacemakerd_state_invalid; /* No crm-mon-disconnected message for console; a working implementation * is not currently worth the effort */ out->transient(out, "%s", msg); out->message(out, "crm-mon-disconnected", msg, pcmkd_state); if (refresh_timer != NULL) { /* we'll trigger a refresh after reconnect */ mainloop_timer_stop(refresh_timer); } if (reconnect_timer) { /* we'll trigger a new reconnect-timeout at the end */ g_source_remove(reconnect_timer); reconnect_timer = 0; } /* the client API won't properly reconnect notifications if they are still * in the table - so remove them */ if (st != NULL) { if (st->state != stonith_disconnected) { st->cmds->disconnect(st); } st->cmds->remove_notification(st, NULL); } if (cib) { cib->cmds->signoff(cib); reconnect_timer = pcmk__create_timer(options.reconnect_ms, reconnect_after_timeout, NULL); } } /* Signal handler installed into the mainloop for normal program shutdown */ static void mon_shutdown(int nsig) { clean_up(CRM_EX_OK); } #if PCMK__ENABLE_CURSES static volatile sighandler_t ncurses_winch_handler; /* Signal handler installed the regular way (not into the main loop) for when * the screen is resized. Commonly, this happens when running in an xterm and * the user changes its size. */ static void mon_winresize(int nsig) { static int not_done; int lines = 0, cols = 0; if (!not_done++) { if (ncurses_winch_handler) /* the original ncurses WINCH signal handler does the * magic of retrieving the new window size; * otherwise, we'd have to use ioctl or tgetent */ (*ncurses_winch_handler) (SIGWINCH); getmaxyx(stdscr, lines, cols); resizeterm(lines, cols); /* Alert the mainloop code we'd like the refresh_trigger to run next * time the mainloop gets around to checking. */ mainloop_set_trigger((crm_trigger_t *) refresh_trigger); } not_done--; } #endif static int setup_fencer_connection(void) { int rc = pcmk_ok; if (options.fence_connect && st == NULL) { st = stonith_api_new(); } if (!options.fence_connect || st == NULL || st->state != stonith_disconnected) { return rc; } rc = st->cmds->connect(st, crm_system_name, NULL); if (rc == pcmk_ok) { crm_trace("Setting up stonith callbacks"); if (options.watch_fencing) { st->cmds->register_notification(st, PCMK__VALUE_ST_NOTIFY_DISCONNECT, mon_st_callback_event); st->cmds->register_notification(st, PCMK__VALUE_ST_NOTIFY_FENCE, mon_st_callback_event); } else { st->cmds->register_notification(st, PCMK__VALUE_ST_NOTIFY_DISCONNECT, mon_st_callback_display); st->cmds->register_notification(st, PCMK__VALUE_ST_NOTIFY_HISTORY, mon_st_callback_display); } } else { stonith_api_delete(st); st = NULL; } return rc; } static int setup_cib_connection(void) { int rc = pcmk_rc_ok; CRM_CHECK(cib != NULL, return EINVAL); if (cib->state != cib_disconnected) { // Already connected with notifications registered for CIB updates return rc; } rc = cib__signon_query(out, &cib, ¤t_cib); if (rc == pcmk_rc_ok) { rc = pcmk_legacy2rc(cib->cmds->set_connection_dnotify(cib, mon_cib_connection_destroy)); if (rc == EPROTONOSUPPORT) { out->err(out, "CIB client does not support connection loss " "notifications; crm_mon will be unable to reconnect after " "connection loss"); rc = pcmk_rc_ok; } if (rc == pcmk_rc_ok) { cib->cmds->del_notify_callback(cib, PCMK__VALUE_CIB_DIFF_NOTIFY, crm_diff_update); rc = cib->cmds->add_notify_callback(cib, PCMK__VALUE_CIB_DIFF_NOTIFY, crm_diff_update); rc = pcmk_legacy2rc(rc); } if (rc != pcmk_rc_ok) { if (rc == EPROTONOSUPPORT) { out->err(out, "CIB client does not support CIB diff " "notifications"); } else { out->err(out, "CIB diff notification setup failed"); } out->err(out, "Cannot monitor CIB changes; exiting"); cib__clean_up_connection(&cib); stonith_api_delete(st); st = NULL; } } return rc; } /* This is used to set up the fencing options after the interactive UI has been stared. * fence_history_cb can't be used because it builds up a list of includes/excludes that * then have to be processed with apply_include_exclude and that could affect other * things. */ static void set_fencing_options(int level) { switch (level) { case 3: options.fence_connect = TRUE; fence_history = pcmk__fence_history_full; show |= pcmk_section_fencing_all; break; case 2: options.fence_connect = TRUE; fence_history = pcmk__fence_history_full; show |= pcmk_section_fencing_all; break; case 1: options.fence_connect = TRUE; fence_history = pcmk__fence_history_full; show |= pcmk_section_fence_failed | pcmk_section_fence_pending; break; default: interactive_fence_level = 0; options.fence_connect = FALSE; fence_history = pcmk__fence_history_none; show &= ~pcmk_section_fencing_all; break; } } static int setup_api_connections(void) { int rc = pcmk_rc_ok; CRM_CHECK(cib != NULL, return EINVAL); if (cib->state != cib_disconnected) { return rc; } if (cib->variant == cib_native) { rc = pcmk__pacemakerd_status(out, crm_system_name, options.reconnect_ms / 2, false, &pcmkd_state); if (rc != pcmk_rc_ok) { return rc; } switch (pcmkd_state) { case pcmk_pacemakerd_state_running: case pcmk_pacemakerd_state_remote: case pcmk_pacemakerd_state_shutting_down: /* Fencer and CIB may still be available while shutting down or * running on a Pacemaker Remote node */ break; default: // Fencer and CIB are definitely unavailable return ENOTCONN; } setup_fencer_connection(); } rc = setup_cib_connection(); return rc; } #if PCMK__ENABLE_CURSES static const char * get_option_desc(char c) { const char *desc = "No help available"; for (GOptionEntry *entry = display_entries; entry != NULL; entry++) { if (entry->short_name == c) { desc = entry->description; break; } } return desc; } #define print_option_help(out, option, condition) \ curses_formatted_printf(out, "%c %c: \t%s\n", ((condition)? '*': ' '), option, get_option_desc(option)); /* This function is called from the main loop when there is something to be read * on stdin, like an interactive user's keystroke. All it does is read the keystroke, * set flags (or show the page showing which keystrokes are valid), and redraw the * screen. It does not do anything with connections to the CIB or fencing agent * agent what would happen in mon_refresh_display. */ static gboolean detect_user_input(GIOChannel *channel, GIOCondition condition, gpointer user_data) { int c; gboolean config_mode = FALSE; gboolean rc = G_SOURCE_CONTINUE; /* If the attached pty device (pseudo-terminal) has been closed/deleted, * the condition (G_IO_IN | G_IO_ERR | G_IO_HUP) occurs. * Exit with an error, otherwise the process would persist in the * background and significantly raise the CPU usage. */ if ((condition & G_IO_ERR) && (condition & G_IO_HUP)) { rc = G_SOURCE_REMOVE; clean_up(CRM_EX_IOERR); } /* The connection/fd has been closed. Refresh the screen and remove this * event source hence ignore stdin. */ if (condition & (G_IO_HUP | G_IO_NVAL)) { rc = G_SOURCE_REMOVE; } if ((condition & G_IO_IN) == 0) { return rc; } while (1) { /* Get user input */ c = getchar(); switch (c) { case 'm': interactive_fence_level++; if (interactive_fence_level > 3) { interactive_fence_level = 0; } set_fencing_options(interactive_fence_level); break; case 'c': show ^= pcmk_section_tickets; break; case 'f': show ^= pcmk_section_failcounts; break; case 'n': show_opts ^= pcmk_show_rscs_by_node; break; case 'o': show ^= pcmk_section_operations; if (!pcmk_is_set(show, pcmk_section_operations)) { show_opts &= ~pcmk_show_timing; } break; case 'r': show_opts ^= pcmk_show_inactive_rscs; break; case 'R': show_opts ^= pcmk_show_details; break; case 't': show_opts ^= pcmk_show_timing; if (pcmk_is_set(show_opts, pcmk_show_timing)) { show |= pcmk_section_operations; } break; case 'A': show ^= pcmk_section_attributes; break; case 'L': show ^= pcmk_section_bans; break; case 'D': /* If any header is shown, clear them all, otherwise set them all */ if (pcmk_any_flags_set(show, pcmk_section_summary)) { show &= ~pcmk_section_summary; } else { show |= pcmk_section_summary; } /* Regardless, we don't show options in console mode. */ show &= ~pcmk_section_options; break; case 'b': show_opts ^= pcmk_show_brief; break; case 'j': show_opts ^= pcmk_show_pending; break; case '?': config_mode = TRUE; break; default: /* All other keys just redraw the screen. */ goto refresh; } if (!config_mode) goto refresh; clear(); refresh(); curses_formatted_printf(out, "%s", "Display option change mode\n"); print_option_help(out, 'c', pcmk_is_set(show, pcmk_section_tickets)); print_option_help(out, 'f', pcmk_is_set(show, pcmk_section_failcounts)); print_option_help(out, 'n', pcmk_is_set(show_opts, pcmk_show_rscs_by_node)); print_option_help(out, 'o', pcmk_is_set(show, pcmk_section_operations)); print_option_help(out, 'r', pcmk_is_set(show_opts, pcmk_show_inactive_rscs)); print_option_help(out, 't', pcmk_is_set(show_opts, pcmk_show_timing)); print_option_help(out, 'A', pcmk_is_set(show, pcmk_section_attributes)); print_option_help(out, 'L', pcmk_is_set(show, pcmk_section_bans)); print_option_help(out, 'D', !pcmk_is_set(show, pcmk_section_summary)); print_option_help(out, 'R', pcmk_any_flags_set(show_opts, pcmk_show_details)); print_option_help(out, 'b', pcmk_is_set(show_opts, pcmk_show_brief)); print_option_help(out, 'j', pcmk_is_set(show_opts, pcmk_show_pending)); curses_formatted_printf(out, "%d m: \t%s\n", interactive_fence_level, get_option_desc('m')); curses_formatted_printf(out, "%s", "\nToggle fields via field letter, type any other key to return\n"); } refresh: refresh_after_event(FALSE, TRUE); return rc; } #endif // PCMK__ENABLE_CURSES // Basically crm_signal_handler(SIGCHLD, SIG_IGN) plus the SA_NOCLDWAIT flag static void avoid_zombies(void) { struct sigaction sa; memset(&sa, 0, sizeof(struct sigaction)); if (sigemptyset(&sa.sa_mask) < 0) { crm_warn("Cannot avoid zombies: %s", pcmk_rc_str(errno)); return; } sa.sa_handler = SIG_IGN; sa.sa_flags = SA_RESTART|SA_NOCLDWAIT; if (sigaction(SIGCHLD, &sa, NULL) < 0) { crm_warn("Cannot avoid zombies: %s", pcmk_rc_str(errno)); } } static GOptionContext * build_arg_context(pcmk__common_args_t *args, GOptionGroup **group) { GOptionContext *context = NULL; GOptionEntry extra_prog_entries[] = { { "quiet", 'Q', 0, G_OPTION_ARG_NONE, &(args->quiet), "Be less descriptive in output.", NULL }, { NULL } }; #if PCMK__ENABLE_CURSES const char *fmts = "console (default), html, text, xml, none"; #else const char *fmts = "text (default), html, xml, none"; #endif // PCMK__ENABLE_CURSES const char *desc = NULL; desc = "Notes:\n\n" "Time Specification:\n\n" "The TIMESPEC in any command line option can be specified in many\n" "different formats. It can be an integer number of seconds, a\n" "number plus units (us/usec/ms/msec/s/sec/m/min/h/hr), or an ISO\n" "8601 period specification.\n\n" "Output Control:\n\n" "By default, a particular set of sections are written to the\n" "output destination. The default varies based on the output\n" "format: XML includes all sections by default, while other output\n" "formats include less. This set can be modified with the --include\n" "and --exclude command line options. Each option may be passed\n" "multiple times, and each can specify a comma-separated list of\n" "sections. The options are applied to the default set, in order\n" "from left to right as they are passed on the command line. For a\n" "list of valid sections, pass --include=list or --exclude=list.\n\n" "Interactive Use:\n\n" #if PCMK__ENABLE_CURSES "When run interactively, crm_mon can be told to hide and show\n" "various sections of output. To see a help screen explaining the\n" "options, press '?'. Any key stroke aside from those listed will\n" "cause the screen to refresh.\n\n" #else "The local installation of Pacemaker was built without support for\n" "interactive (console) mode. A curses library must be available at\n" "build time to support interactive mode.\n\n" #endif // PCMK__ENABLE_CURSES "Examples:\n\n" #if PCMK__ENABLE_CURSES "Display the cluster status on the console with updates as they\n" "occur:\n\n" "\tcrm_mon\n\n" #endif // PCMK__ENABLE_CURSES "Display the cluster status once and exit:\n\n" "\tcrm_mon -1\n\n" "Display the cluster status, group resources by node, and include\n" "inactive resources in the list:\n\n" "\tcrm_mon --group-by-node --inactive\n\n" "Start crm_mon as a background daemon and have it write the\n" "cluster status to an HTML file:\n\n" "\tcrm_mon --daemonize --output-as html " "--output-to /path/to/docroot/filename.html\n\n" "Display the cluster status as XML:\n\n" "\tcrm_mon --output-as xml\n\n"; context = pcmk__build_arg_context(args, fmts, group, NULL); pcmk__add_main_args(context, extra_prog_entries); g_option_context_set_description(context, desc); pcmk__add_arg_group(context, "display", "Display Options:", "Show display options", display_entries); pcmk__add_arg_group(context, "additional", "Additional Options:", "Show additional options", addl_entries); pcmk__add_arg_group(context, "deprecated", "Deprecated Options:", "Show deprecated options", deprecated_entries); return context; } /*! * \internal * \brief Set output format based on \c --output-as arguments and mode arguments * * When the deprecated \c --as-xml argument is parsed, a callback function sets * \c output_format. Otherwise, this function does the same based on the current * \c --output-as arguments and the \c --one-shot and \c --daemonize arguments. * * \param[in,out] args Command line arguments */ static void reconcile_output_format(pcmk__common_args_t *args) { if (output_format != mon_output_unset) { /* The deprecated --as-xml argument was used, and we're finished. Note * that this means the deprecated argument takes precedence. */ return; } if (pcmk__str_eq(args->output_ty, PCMK_VALUE_NONE, pcmk__str_none)) { output_format = mon_output_none; } else if (pcmk__str_eq(args->output_ty, "html", pcmk__str_none)) { output_format = mon_output_html; umask(S_IWGRP | S_IWOTH); // World-readable HTML } else if (pcmk__str_eq(args->output_ty, "xml", pcmk__str_none)) { output_format = mon_output_xml; #if PCMK__ENABLE_CURSES } else if (pcmk__str_eq(args->output_ty, "console", pcmk__str_null_matches)) { /* Console is the default format if no conflicting options are given. * * Use text output instead if one of the following conditions is met: * * We've requested daemonized or one-shot mode (console output is * incompatible with modes other than mon_exec_update) * * We requested the version, which is effectively one-shot * * We specified a non-stdout output destination (console mode is * compatible only with stdout) */ if ((options.exec_mode == mon_exec_daemonized) || (options.exec_mode == mon_exec_one_shot) || args->version || !pcmk__str_eq(args->output_dest, "-", pcmk__str_null_matches)) { pcmk__str_update(&args->output_ty, "text"); output_format = mon_output_plain; } else { pcmk__str_update(&args->output_ty, "console"); output_format = mon_output_console; crm_enable_stderr(FALSE); } #endif // PCMK__ENABLE_CURSES } else if (pcmk__str_eq(args->output_ty, "text", pcmk__str_null_matches)) { /* Text output was explicitly requested, or it's the default because * curses is not enabled */ pcmk__str_update(&args->output_ty, "text"); output_format = mon_output_plain; } // Otherwise, invalid format. Let pcmk__output_new() throw an error. } /*! * \internal * \brief Set execution mode to the output format's default if appropriate * * \param[in,out] args Command line arguments */ static void set_default_exec_mode(const pcmk__common_args_t *args) { if (output_format == mon_output_console) { /* Update is the only valid mode for console, but set here instead of * reconcile_output_format() for isolation and consistency */ options.exec_mode = mon_exec_update; } else if (options.exec_mode == mon_exec_unset) { // Default to one-shot mode for all other formats options.exec_mode = mon_exec_one_shot; } else if ((options.exec_mode == mon_exec_update) && pcmk__str_eq(args->output_dest, "-", pcmk__str_null_matches)) { // If not using console format, update mode cannot be used with stdout options.exec_mode = mon_exec_one_shot; } } static void clean_up_on_connection_failure(int rc) { if (rc == ENOTCONN) { if (pcmkd_state == pcmk_pacemakerd_state_remote) { g_set_error(&error, PCMK__EXITC_ERROR, CRM_EX_ERROR, "Error: remote-node not connected to cluster"); } else { g_set_error(&error, PCMK__EXITC_ERROR, CRM_EX_ERROR, "Error: cluster is not available on this node"); } } else { g_set_error(&error, PCMK__EXITC_ERROR, CRM_EX_ERROR, "Connection to cluster failed: %s", pcmk_rc_str(rc)); } clean_up(pcmk_rc2exitc(rc)); } static void one_shot(void) { int rc = pcmk__status(out, cib, fence_history, show, show_opts, options.only_node, options.only_rsc, options.neg_location_prefix, 0); if (rc == pcmk_rc_ok) { clean_up(pcmk_rc2exitc(rc)); } else { clean_up_on_connection_failure(rc); } } static void exit_on_invalid_cib(void) { if (cib != NULL) { return; } // Shouldn't really be possible g_set_error(&error, PCMK__EXITC_ERROR, CRM_EX_ERROR, "Invalid CIB source"); clean_up(CRM_EX_ERROR); } int main(int argc, char **argv) { int rc = pcmk_rc_ok; GOptionGroup *output_group = NULL; args = pcmk__new_common_args(SUMMARY); context = build_arg_context(args, &output_group); pcmk__register_formats(output_group, formats); options.pid_file = strdup("/tmp/ClusterMon.pid"); pcmk__cli_init_logging("crm_mon", 0); // Avoid needing to wait for subprocesses forked for -E/--external-agent avoid_zombies(); processed_args = pcmk__cmdline_preproc(argv, "eimpxEILU"); fence_history_cb("--fence-history", "1", NULL, NULL); /* Set an HTML title regardless of what format we will eventually use. * Doing this here means the user can give their own title on the command * line. */ if (!pcmk__force_args(context, &error, "%s --html-title \"Cluster Status\"", g_get_prgname())) { return clean_up(CRM_EX_USAGE); } if (!g_option_context_parse_strv(context, &processed_args, &error)) { return clean_up(CRM_EX_USAGE); } for (int i = 0; i < args->verbosity; i++) { crm_bump_log_level(argc, argv); } if (!args->version) { if (args->quiet) { include_exclude_cb("--exclude", "times", NULL, NULL); } if (options.watch_fencing) { fence_history_cb("--fence-history", "0", NULL, NULL); options.fence_connect = TRUE; } /* create the cib-object early to be able to do further * decisions based on the cib-source */ cib = cib_new(); exit_on_invalid_cib(); switch (cib->variant) { case cib_native: // Everything (fencer, CIB, pcmkd status) should be available break; case cib_file: // Live fence history is not meaningful fence_history_cb("--fence-history", "0", NULL, NULL); /* Notifications are unsupported; nothing to monitor * @COMPAT: Let setup_cib_connection() handle this by exiting? */ options.exec_mode = mon_exec_one_shot; break; case cib_remote: // We won't receive any fencing updates fence_history_cb("--fence-history", "0", NULL, NULL); break; default: /* something is odd */ exit_on_invalid_cib(); break; } if ((options.exec_mode == mon_exec_daemonized) && !options.external_agent && pcmk__str_eq(args->output_dest, "-", pcmk__str_null_matches)) { g_set_error(&error, PCMK__EXITC_ERROR, CRM_EX_USAGE, "--daemonize requires at least one of --output-to " "(with value not set to '-') and --external-agent"); return clean_up(CRM_EX_USAGE); } } reconcile_output_format(args); set_default_exec_mode(args); rc = pcmk__output_new(&out, args->output_ty, args->output_dest, argv); if (rc != pcmk_rc_ok) { g_set_error(&error, PCMK__EXITC_ERROR, CRM_EX_ERROR, "Error creating output format %s: %s", args->output_ty, pcmk_rc_str(rc)); return clean_up(CRM_EX_ERROR); } if (output_format == mon_output_legacy_xml) { output_format = mon_output_xml; pcmk__output_set_legacy_xml(out); } /* output_format MUST NOT BE CHANGED AFTER THIS POINT. */ /* If we had a valid format for pcmk__output_new(), output_format should be * set by now. */ pcmk__assert(output_format != mon_output_unset); if (output_format == mon_output_plain) { pcmk__output_text_set_fancy(out, true); } if (options.exec_mode == mon_exec_daemonized) { if (!options.external_agent && (output_format == mon_output_none)) { g_set_error(&error, PCMK__EXITC_ERROR, CRM_EX_USAGE, "--daemonize requires --external-agent if used with " "--output-as=none"); return clean_up(CRM_EX_USAGE); } crm_enable_stderr(FALSE); cib_delete(cib); cib = NULL; pcmk__daemonize(crm_system_name, options.pid_file); cib = cib_new(); exit_on_invalid_cib(); } show = default_includes(output_format); /* Apply --include/--exclude flags we used internally. There's no error reporting * here because this would be a programming error. */ apply_include_exclude(options.includes_excludes, &error); /* And now apply any --include/--exclude flags the user gave on the command line. * These are done in a separate pass from the internal ones because we want to * make sure whatever the user specifies overrides whatever we do. */ if (!apply_include_exclude(options.user_includes_excludes, &error)) { return clean_up(CRM_EX_USAGE); } /* Sync up the initial value of interactive_fence_level with whatever was set with * --include/--exclude= options. */ if (pcmk_all_flags_set(show, pcmk_section_fencing_all)) { interactive_fence_level = 3; } else if (pcmk_is_set(show, pcmk_section_fence_worked)) { interactive_fence_level = 2; } else if (pcmk_any_flags_set(show, pcmk_section_fence_failed | pcmk_section_fence_pending)) { interactive_fence_level = 1; } else { interactive_fence_level = 0; } pcmk__register_lib_messages(out); crm_mon_register_messages(out); pe__register_messages(out); stonith__register_messages(out); // Messages internal to this file, nothing curses-specific pcmk__register_messages(out, fmt_functions); if (args->version) { out->version(out, false); return clean_up(CRM_EX_OK); } if (output_format == mon_output_xml) { show_opts |= pcmk_show_inactive_rscs | pcmk_show_timing; } if ((output_format == mon_output_html) && (out->dest != stdout)) { char *content = pcmk__itoa(pcmk__timeout_ms2s(options.reconnect_ms)); pcmk__html_add_header(PCMK__XE_META, PCMK__XA_HTTP_EQUIV, PCMK__VALUE_REFRESH, PCMK__XA_CONTENT, content, NULL); free(content); } crm_info("Starting %s", crm_system_name); cib__set_output(cib, out); if (options.exec_mode == mon_exec_one_shot) { one_shot(); } - scheduler = pe_new_working_set(); + scheduler = pcmk_new_scheduler(); pcmk__mem_assert(scheduler); scheduler->priv->out = out; if ((cib->variant == cib_native) && pcmk_is_set(show, pcmk_section_times)) { // Currently used only in the times section pcmk__query_node_name(out, 0, &(scheduler->priv->local_node_name), 0); } out->message(out, "crm-mon-disconnected", "Waiting for initial connection", pcmkd_state); do { out->transient(out, "Connecting to cluster..."); rc = setup_api_connections(); if (rc != pcmk_rc_ok) { if ((rc == ENOTCONN) || (rc == ECONNREFUSED)) { out->transient(out, "Connection failed. Retrying in %s...", pcmk__readable_interval(options.reconnect_ms)); } // Give some time to view all output even if we won't retry pcmk__sleep_ms(options.reconnect_ms); #if PCMK__ENABLE_CURSES if (output_format == mon_output_console) { clear(); refresh(); } #endif } } while ((rc == ENOTCONN) || (rc == ECONNREFUSED)); if (rc != pcmk_rc_ok) { clean_up_on_connection_failure(rc); } set_fencing_options(interactive_fence_level); mon_refresh_display(NULL); mainloop = g_main_loop_new(NULL, FALSE); mainloop_add_signal(SIGTERM, mon_shutdown); mainloop_add_signal(SIGINT, mon_shutdown); #if PCMK__ENABLE_CURSES if (output_format == mon_output_console) { ncurses_winch_handler = crm_signal_handler(SIGWINCH, mon_winresize); if (ncurses_winch_handler == SIG_DFL || ncurses_winch_handler == SIG_IGN || ncurses_winch_handler == SIG_ERR) ncurses_winch_handler = NULL; io_channel = g_io_channel_unix_new(STDIN_FILENO); g_io_add_watch(io_channel, (G_IO_IN | G_IO_ERR | G_IO_HUP | G_IO_NVAL), detect_user_input, NULL); } #endif /* When refresh_trigger->trigger is set to TRUE, call mon_refresh_display. In * this file, that is anywhere mainloop_set_trigger is called. */ refresh_trigger = mainloop_add_trigger(G_PRIORITY_LOW, mon_refresh_display, NULL); g_main_loop_run(mainloop); g_main_loop_unref(mainloop); crm_info("Exiting %s", crm_system_name); return clean_up(CRM_EX_OK); } static int send_custom_trap(const char *node, const char *rsc, const char *task, int target_rc, int rc, int status, const char *desc) { pid_t pid; /*setenv needs chars, these are ints */ char *rc_s = pcmk__itoa(rc); char *status_s = pcmk__itoa(status); char *target_rc_s = pcmk__itoa(target_rc); crm_debug("Sending external notification to '%s' via '%s'", options.external_recipient, options.external_agent); if(rsc) { setenv("CRM_notify_rsc", rsc, 1); } if (options.external_recipient) { setenv("CRM_notify_recipient", options.external_recipient, 1); } setenv("CRM_notify_node", node, 1); setenv("CRM_notify_task", task, 1); setenv("CRM_notify_desc", desc, 1); setenv("CRM_notify_rc", rc_s, 1); setenv("CRM_notify_target_rc", target_rc_s, 1); setenv("CRM_notify_status", status_s, 1); pid = fork(); if (pid == -1) { out->err(out, "notification fork() failed: %s", strerror(errno)); } if (pid == 0) { /* crm_debug("notification: I am the child. Executing the nofitication program."); */ execl(options.external_agent, options.external_agent, NULL); crm_exit(CRM_EX_ERROR); } crm_trace("Finished running custom notification program '%s'.", options.external_agent); free(target_rc_s); free(status_s); free(rc_s); return 0; } static int handle_rsc_op(xmlNode *xml, void *userdata) { const char *node_id = (const char *) userdata; int rc = -1; int status = -1; int target_rc = -1; gboolean notify = TRUE; char *rsc = NULL; char *task = NULL; const char *desc = NULL; const char *magic = NULL; const char *id = NULL; const char *node = NULL; xmlNode *n = xml; xmlNode * rsc_op = xml; if(strcmp((const char*)xml->name, PCMK__XE_LRM_RSC_OP) != 0) { pcmk__xe_foreach_child(xml, NULL, handle_rsc_op, (void *) node_id); return pcmk_rc_ok; } id = pcmk__xe_history_key(rsc_op); magic = crm_element_value(rsc_op, PCMK__XA_TRANSITION_MAGIC); if (magic == NULL) { /* non-change */ return pcmk_rc_ok; } if (!decode_transition_magic(magic, NULL, NULL, NULL, &status, &rc, &target_rc)) { crm_err("Invalid event %s detected for %s", magic, id); return pcmk_rc_ok; } if (parse_op_key(id, &rsc, &task, NULL) == FALSE) { crm_err("Invalid event detected for %s", id); goto bail; } node = crm_element_value(rsc_op, PCMK__META_ON_NODE); while ((n != NULL) && !pcmk__xe_is(n, PCMK__XE_NODE_STATE)) { n = n->parent; } if(node == NULL && n) { node = crm_element_value(n, PCMK_XA_UNAME); } if (node == NULL && n) { node = pcmk__xe_id(n); } if (node == NULL) { node = node_id; } if (node == NULL) { crm_err("No node detected for event %s (%s)", magic, id); goto bail; } /* look up where we expected it to be? */ desc = pcmk_rc_str(pcmk_rc_ok); if ((status == PCMK_EXEC_DONE) && (target_rc == rc)) { crm_notice("%s of %s on %s completed: %s", task, rsc, node, desc); if (rc == PCMK_OCF_NOT_RUNNING) { notify = FALSE; } } else if (status == PCMK_EXEC_DONE) { desc = crm_exit_str(rc); crm_warn("%s of %s on %s failed: %s", task, rsc, node, desc); } else { desc = pcmk_exec_status_str(status); crm_warn("%s of %s on %s failed: %s", task, rsc, node, desc); } if (notify && options.external_agent) { send_custom_trap(node, rsc, task, target_rc, rc, status, desc); } bail: free(rsc); free(task); return pcmk_rc_ok; } /* This function is just a wrapper around mainloop_set_trigger so that it can be * called from a mainloop directly. It's simply another way of ensuring the screen * gets redrawn. */ static gboolean mon_trigger_refresh(gpointer user_data) { mainloop_set_trigger((crm_trigger_t *) refresh_trigger); return FALSE; } static int handle_op_for_node(xmlNode *xml, void *userdata) { const char *node = crm_element_value(xml, PCMK_XA_UNAME); if (node == NULL) { node = pcmk__xe_id(xml); } handle_rsc_op(xml, (void *) node); return pcmk_rc_ok; } static int crm_diff_update_element(xmlNode *change, void *userdata) { const char *name = NULL; const char *op = crm_element_value(change, PCMK_XA_OPERATION); const char *xpath = crm_element_value(change, PCMK_XA_PATH); xmlNode *match = NULL; const char *node = NULL; if (op == NULL) { return pcmk_rc_ok; } else if (strcmp(op, PCMK_VALUE_CREATE) == 0) { match = change->children; } else if (pcmk__str_any_of(op, PCMK_VALUE_MOVE, PCMK_VALUE_DELETE, NULL)) { return pcmk_rc_ok; } else if (strcmp(op, PCMK_VALUE_MODIFY) == 0) { match = pcmk__xe_first_child(change, PCMK_XE_CHANGE_RESULT, NULL, NULL); if(match) { match = match->children; } } if(match) { name = (const char *)match->name; } crm_trace("Handling %s operation for %s %p, %s", op, xpath, match, name); if(xpath == NULL) { /* Version field, ignore */ } else if(name == NULL) { crm_debug("No result for %s operation to %s", op, xpath); pcmk__assert(pcmk__str_any_of(op, PCMK_VALUE_MOVE, PCMK_VALUE_DELETE, NULL)); } else if (strcmp(name, PCMK_XE_CIB) == 0) { pcmk__xe_foreach_child(pcmk__xe_first_child(match, PCMK_XE_STATUS, NULL, NULL), NULL, handle_op_for_node, NULL); } else if (strcmp(name, PCMK_XE_STATUS) == 0) { pcmk__xe_foreach_child(match, NULL, handle_op_for_node, NULL); } else if (strcmp(name, PCMK__XE_NODE_STATE) == 0) { node = crm_element_value(match, PCMK_XA_UNAME); if (node == NULL) { node = pcmk__xe_id(match); } handle_rsc_op(match, (void *) node); } else if (strcmp(name, PCMK__XE_LRM) == 0) { node = pcmk__xe_id(match); handle_rsc_op(match, (void *) node); } else if (strcmp(name, PCMK__XE_LRM_RESOURCES) == 0) { char *local_node = pcmk__xpath_node_id(xpath, PCMK__XE_LRM); handle_rsc_op(match, local_node); free(local_node); } else if (strcmp(name, PCMK__XE_LRM_RESOURCE) == 0) { char *local_node = pcmk__xpath_node_id(xpath, PCMK__XE_LRM); handle_rsc_op(match, local_node); free(local_node); } else if (strcmp(name, PCMK__XE_LRM_RSC_OP) == 0) { char *local_node = pcmk__xpath_node_id(xpath, PCMK__XE_LRM); handle_rsc_op(match, local_node); free(local_node); } else { crm_trace("Ignoring %s operation for %s %p, %s", op, xpath, match, name); } return pcmk_rc_ok; } static void crm_diff_update(const char *event, xmlNode * msg) { int rc = -1; static bool stale = FALSE; gboolean cib_updated = FALSE; xmlNode *wrapper = pcmk__xe_first_child(msg, PCMK__XE_CIB_UPDATE_RESULT, NULL, NULL); xmlNode *diff = pcmk__xe_first_child(wrapper, NULL, NULL, NULL); out->progress(out, false); if (current_cib != NULL) { rc = xml_apply_patchset(current_cib, diff, TRUE); switch (rc) { case -pcmk_err_diff_resync: case -pcmk_err_diff_failed: crm_notice("[%s] Patch aborted: %s (%d)", event, pcmk_strerror(rc), rc); pcmk__xml_free(current_cib); current_cib = NULL; break; case pcmk_ok: cib_updated = TRUE; break; default: crm_notice("[%s] ABORTED: %s (%d)", event, pcmk_strerror(rc), rc); pcmk__xml_free(current_cib); current_cib = NULL; } } if (current_cib == NULL) { crm_trace("Re-requesting the full cib"); cib->cmds->query(cib, NULL, ¤t_cib, cib_sync_call); } if (options.external_agent) { int format = 0; crm_element_value_int(diff, PCMK_XA_FORMAT, &format); if (format == 2) { xmlNode *wrapper = pcmk__xe_first_child(msg, PCMK__XE_CIB_UPDATE_RESULT, NULL, NULL); xmlNode *diff = pcmk__xe_first_child(wrapper, NULL, NULL, NULL); pcmk__xe_foreach_child(diff, NULL, crm_diff_update_element, NULL); } else { crm_err("Unknown patch format: %d", format); } } if (current_cib == NULL) { if(!stale) { out->info(out, "--- Stale data ---"); } stale = TRUE; return; } stale = FALSE; refresh_after_event(cib_updated, FALSE); } static int mon_refresh_display(gpointer user_data) { int rc = pcmk_rc_ok; last_refresh = time(NULL); if (output_format == mon_output_none) { return G_SOURCE_REMOVE; } if (fence_history == pcmk__fence_history_full && !pcmk_all_flags_set(show, pcmk_section_fencing_all) && output_format != mon_output_xml) { fence_history = pcmk__fence_history_reduced; } // Get an up-to-date pacemakerd status for the cluster summary if (cib->variant == cib_native) { pcmk__pacemakerd_status(out, crm_system_name, options.reconnect_ms / 2, false, &pcmkd_state); } if (out->dest != stdout) { out->reset(out); } rc = pcmk__output_cluster_status(scheduler, st, cib, current_cib, pcmkd_state, fence_history, show, show_opts, options.only_node,options.only_rsc, options.neg_location_prefix); if (rc == pcmk_rc_schema_validation) { clean_up(CRM_EX_CONFIG); return G_SOURCE_REMOVE; } if (out->dest != stdout) { out->finish(out, CRM_EX_OK, true, NULL); } return G_SOURCE_CONTINUE; } /* This function is called for fencing events (see setup_fencer_connection() for * which ones) when --watch-fencing is used on the command line */ static void mon_st_callback_event(stonith_t * st, stonith_event_t * e) { if (st->state == stonith_disconnected) { /* disconnect cib as well and have everything reconnect */ mon_cib_connection_destroy(NULL); } else if (options.external_agent) { char *desc = stonith__event_description(e); send_custom_trap(e->target, NULL, e->operation, pcmk_ok, e->result, 0, desc); free(desc); } } /* Cause the screen to be redrawn (via mainloop_set_trigger) when various conditions are met: * * - If the last update occurred more than reconnect_ms ago (defaults to 5s, but * can be changed via the -i command line option), or * - After every 10 CIB updates, or * - If it's been 2s since the last update * * This function sounds like it would be more broadly useful, but it is only called when a * fencing event is received or a CIB diff occurrs. */ static void refresh_after_event(gboolean data_updated, gboolean enforce) { static int updates = 0; time_t now = time(NULL); if (data_updated) { updates++; } if(refresh_timer == NULL) { refresh_timer = mainloop_timer_add("refresh", 2000, FALSE, mon_trigger_refresh, NULL); } if (reconnect_timer > 0) { /* we will receive a refresh request after successful reconnect */ mainloop_timer_stop(refresh_timer); return; } /* as we're not handling initial failure of fencer-connection as * fatal give it a retry here * not getting here if cib-reconnection is already on the way */ setup_fencer_connection(); if (enforce || ((now - last_refresh) > pcmk__timeout_ms2s(options.reconnect_ms)) || updates >= 10) { mainloop_set_trigger((crm_trigger_t *) refresh_trigger); mainloop_timer_stop(refresh_timer); updates = 0; } else { mainloop_timer_start(refresh_timer); } } /* This function is called for fencing events (see setup_fencer_connection() for * which ones) when --watch-fencing is NOT used on the command line */ static void mon_st_callback_display(stonith_t * st, stonith_event_t * e) { if (st->state == stonith_disconnected) { /* disconnect cib as well and have everything reconnect */ mon_cib_connection_destroy(NULL); } else { out->progress(out, false); refresh_after_event(TRUE, FALSE); } } /* * De-init ncurses, disconnect from the CIB manager, disconnect fencing, * deallocate memory and show usage-message if requested. * * We don't actually return, but nominally returning crm_exit_t allows a usage * like "return clean_up(exit_code);" which helps static analysis understand the * code flow. */ static crm_exit_t clean_up(crm_exit_t exit_code) { /* Quitting crm_mon is much more complicated than it ought to be. */ /* (1) Close connections, free things, etc. */ if (io_channel != NULL) { g_io_channel_shutdown(io_channel, TRUE, NULL); } cib__clean_up_connection(&cib); stonith_api_delete(st); free(options.neg_location_prefix); free(options.only_node); free(options.only_rsc); free(options.pid_file); g_slist_free_full(options.includes_excludes, free); g_strfreev(processed_args); - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); /* (2) If this is abnormal termination and we're in curses mode, shut down * curses first. Any messages displayed to the screen before curses is shut * down will be lost because doing the shut down will also restore the * screen to whatever it looked like before crm_mon was started. */ if (((error != NULL) || (exit_code == CRM_EX_USAGE)) && (output_format == mon_output_console) && (out != NULL)) { out->finish(out, exit_code, false, NULL); pcmk__output_free(out); out = NULL; } /* (3) If this is a command line usage related failure, print the usage * message. */ if (exit_code == CRM_EX_USAGE && (output_format == mon_output_console || output_format == mon_output_plain)) { char *help = g_option_context_get_help(context, TRUE, NULL); fprintf(stderr, "%s", help); g_free(help); } pcmk__free_arg_context(context); /* (4) If this is any kind of error, print the error out and exit. Make * sure to handle situations both before and after formatted output is * set up. We want errors to appear formatted if at all possible. */ if (error != NULL) { if (out != NULL) { out->err(out, "%s: %s", g_get_prgname(), error->message); out->finish(out, exit_code, true, NULL); pcmk__output_free(out); } else { fprintf(stderr, "%s: %s\n", g_get_prgname(), error->message); } g_clear_error(&error); crm_exit(exit_code); } /* (5) Print formatted output to the screen if we made it far enough in * crm_mon to be able to do so. */ if (out != NULL) { if (options.exec_mode != mon_exec_daemonized) { out->finish(out, exit_code, true, NULL); } pcmk__output_free(out); pcmk__unregister_formats(); } crm_exit(exit_code); } diff --git a/tools/crm_resource.c b/tools/crm_resource.c index 5f177799e2..0e425ddbf1 100644 --- a/tools/crm_resource.c +++ b/tools/crm_resource.c @@ -1,2094 +1,2094 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include // uint32_t #include #include #include #include #include #include #include #include #include #include #include #include #define SUMMARY "crm_resource - perform tasks related to Pacemaker cluster resources" enum rsc_command { cmd_none = 0, // No command option given (yet) cmd_ban, cmd_cleanup, cmd_clear, cmd_colocations, cmd_cts, cmd_delete, cmd_delete_param, cmd_digests, cmd_execute_agent, cmd_fail, cmd_get_param, cmd_list_active_ops, cmd_list_agents, cmd_list_all_ops, cmd_list_alternatives, cmd_list_instances, cmd_list_options, cmd_list_providers, cmd_list_resources, cmd_list_standards, cmd_locate, cmd_metadata, cmd_move, cmd_query_xml, cmd_query_xml_raw, cmd_refresh, cmd_restart, cmd_set_param, cmd_wait, cmd_why, }; struct { enum rsc_command rsc_cmd; // crm_resource command to perform // Command-line option values gchar *rsc_id; // Value of --resource gchar *rsc_type; // Value of --resource-type gboolean all; // --all was given gboolean force; // --force was given gboolean clear_expired; // --expired was given gboolean recursive; // --recursive was given gboolean promoted_role_only; // --promoted was given gchar *host_uname; // Value of --node gchar *interval_spec; // Value of --interval gchar *move_lifetime; // Value of --lifetime gchar *operation; // Value of --operation enum pcmk__opt_flags opt_list; // Parsed from --list-options const char *attr_set_type; // Instance, meta, utilization, or element attribute gchar *prop_id; // --nvpair (attribute XML ID) char *prop_name; // Attribute name gchar *prop_set; // --set-name (attribute block XML ID) gchar *prop_value; // --parameter-value (attribute value) guint timeout_ms; // Parsed from --timeout value char *agent_spec; // Standard and/or provider and/or agent int check_level; // Optional value of --validate or --force-check // Resource configuration specified via command-line arguments bool cmdline_config; // Resource configuration was via arguments char *v_agent; // Value of --agent char *v_class; // Value of --class char *v_provider; // Value of --provider GHashTable *cmdline_params; // Resource parameters specified // Positional command-line arguments gchar **remainder; // Positional arguments as given GHashTable *override_params; // Resource parameter values that override config } options = { .attr_set_type = PCMK_XE_INSTANCE_ATTRIBUTES, .check_level = -1, .rsc_cmd = cmd_list_resources, // List all resources if no command given }; gboolean attr_set_type_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error); gboolean cmdline_config_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error); gboolean option_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error); gboolean timeout_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error); static crm_exit_t exit_code = CRM_EX_OK; static pcmk__output_t *out = NULL; static pcmk__common_args_t *args = NULL; // Things that should be cleaned up on exit static GError *error = NULL; static GMainLoop *mainloop = NULL; static cib_t *cib_conn = NULL; static pcmk_ipc_api_t *controld_api = NULL; static pcmk_scheduler_t *scheduler = NULL; #define MESSAGE_TIMEOUT_S 60 #define INDENT " " static pcmk__supported_format_t formats[] = { PCMK__SUPPORTED_FORMAT_NONE, PCMK__SUPPORTED_FORMAT_TEXT, PCMK__SUPPORTED_FORMAT_XML, { NULL, NULL, NULL } }; // Clean up and exit static crm_exit_t bye(crm_exit_t ec) { pcmk__output_and_clear_error(&error, out); if (out != NULL) { out->finish(out, ec, true, NULL); pcmk__output_free(out); } pcmk__unregister_formats(); if (cib_conn != NULL) { cib_t *save_cib_conn = cib_conn; cib_conn = NULL; // Ensure we can't free this twice cib__clean_up_connection(&save_cib_conn); } if (controld_api != NULL) { pcmk_ipc_api_t *save_controld_api = controld_api; controld_api = NULL; // Ensure we can't free this twice pcmk_free_ipc_api(save_controld_api); } if (mainloop != NULL) { g_main_loop_unref(mainloop); mainloop = NULL; } - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); scheduler = NULL; crm_exit(ec); return ec; } static void quit_main_loop(crm_exit_t ec) { exit_code = ec; if (mainloop != NULL) { GMainLoop *mloop = mainloop; mainloop = NULL; // Don't re-enter this block pcmk_quit_main_loop(mloop, 10); g_main_loop_unref(mloop); } } static gboolean resource_ipc_timeout(gpointer data) { // Start with newline because "Waiting for ..." message doesn't have one if (error != NULL) { g_clear_error(&error); } g_set_error(&error, PCMK__EXITC_ERROR, CRM_EX_TIMEOUT, _("Aborting because no messages received in %d seconds"), MESSAGE_TIMEOUT_S); quit_main_loop(CRM_EX_TIMEOUT); return FALSE; } static void controller_event_callback(pcmk_ipc_api_t *api, enum pcmk_ipc_event event_type, crm_exit_t status, void *event_data, void *user_data) { switch (event_type) { case pcmk_ipc_event_disconnect: if (exit_code == CRM_EX_DISCONNECT) { // Unexpected crm_info("Connection to controller was terminated"); } quit_main_loop(exit_code); break; case pcmk_ipc_event_reply: if (status != CRM_EX_OK) { out->err(out, "Error: bad reply from controller: %s", crm_exit_str(status)); pcmk_disconnect_ipc(api); quit_main_loop(status); } else { if ((pcmk_controld_api_replies_expected(api) == 0) && mainloop && g_main_loop_is_running(mainloop)) { out->info(out, "... got reply (done)"); crm_debug("Got all the replies we expected"); pcmk_disconnect_ipc(api); quit_main_loop(CRM_EX_OK); } else { out->info(out, "... got reply"); } } break; default: break; } } static void start_mainloop(pcmk_ipc_api_t *capi) { unsigned int count = pcmk_controld_api_replies_expected(capi); if (count > 0) { out->info(out, "Waiting for %u %s from the controller", count, pcmk__plural_alt(count, "reply", "replies")); exit_code = CRM_EX_DISCONNECT; // For unexpected disconnects mainloop = g_main_loop_new(NULL, FALSE); pcmk__create_timer(MESSAGE_TIMEOUT_S * 1000, resource_ipc_timeout, NULL); g_main_loop_run(mainloop); } } static int compare_id(gconstpointer a, gconstpointer b) { return strcmp((const char *)a, (const char *)b); } static GList * build_constraint_list(xmlNode *root) { GList *retval = NULL; xmlNode *cib_constraints = NULL; xmlXPathObjectPtr xpathObj = NULL; int ndx = 0; cib_constraints = pcmk_find_cib_element(root, PCMK_XE_CONSTRAINTS); xpathObj = xpath_search(cib_constraints, "//" PCMK_XE_RSC_LOCATION); for (ndx = 0; ndx < numXpathResults(xpathObj); ndx++) { xmlNode *match = getXpathResult(xpathObj, ndx); retval = g_list_insert_sorted(retval, (gpointer) pcmk__xe_id(match), compare_id); } freeXpathObject(xpathObj); return retval; } static gboolean validate_opt_list(const gchar *optarg) { if (pcmk__str_eq(optarg, PCMK_VALUE_FENCING, pcmk__str_none)) { options.opt_list = pcmk__opt_fencing; } else if (pcmk__str_eq(optarg, PCMK__VALUE_PRIMITIVE, pcmk__str_none)) { options.opt_list = pcmk__opt_primitive; } else { return FALSE; } return TRUE; } /*! * \internal * \brief Process options that set the command * * Nothing else should set \c options.rsc_cmd. * * \param[in] option_name Name of the option being parsed * \param[in] optarg Value to be parsed * \param[in] data Ignored * \param[out] error Where to store recoverable error, if any * * \return \c TRUE if the option was successfully parsed, or \c FALSE if an * error occurred, in which case \p *error is set */ static gboolean command_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { // Sorted by enum rsc_command name if (pcmk__str_any_of(option_name, "-B", "--ban", NULL)) { options.rsc_cmd = cmd_ban; } else if (pcmk__str_any_of(option_name, "-C", "--cleanup", NULL)) { options.rsc_cmd = cmd_cleanup; } else if (pcmk__str_any_of(option_name, "-U", "--clear", NULL)) { options.rsc_cmd = cmd_clear; } else if (pcmk__str_any_of(option_name, "-a", "--constraints", NULL)) { options.rsc_cmd = cmd_colocations; } else if (pcmk__str_any_of(option_name, "-A", "--stack", NULL)) { options.rsc_cmd = cmd_colocations; options.recursive = TRUE; } else if (pcmk__str_any_of(option_name, "-c", "--list-cts", NULL)) { options.rsc_cmd = cmd_cts; } else if (pcmk__str_any_of(option_name, "-D", "--delete", NULL)) { options.rsc_cmd = cmd_delete; } else if (pcmk__str_any_of(option_name, "-d", "--delete-parameter", NULL)) { options.rsc_cmd = cmd_delete_param; pcmk__str_update(&options.prop_name, optarg); } else if (pcmk__str_eq(option_name, "--digests", pcmk__str_none)) { options.rsc_cmd = cmd_digests; if (options.override_params == NULL) { options.override_params = pcmk__strkey_table(free, free); } } else if (pcmk__str_any_of(option_name, "--force-demote", "--force-promote", "--force-start", "--force-stop", "--force-check", "--validate", NULL)) { options.rsc_cmd = cmd_execute_agent; g_free(options.operation); options.operation = g_strdup(option_name + 2); // skip "--" if (options.override_params == NULL) { options.override_params = pcmk__strkey_table(free, free); } if (optarg != NULL) { if (pcmk__scan_min_int(optarg, &options.check_level, 0) != pcmk_rc_ok) { g_set_error(error, G_OPTION_ERROR, CRM_EX_INVALID_PARAM, _("Invalid check level setting: %s"), optarg); return FALSE; } } } else if (pcmk__str_any_of(option_name, "-F", "--fail", NULL)) { options.rsc_cmd = cmd_fail; } else if (pcmk__str_any_of(option_name, "-g", "--get-parameter", NULL)) { options.rsc_cmd = cmd_get_param; pcmk__str_update(&options.prop_name, optarg); } else if (pcmk__str_any_of(option_name, "-O", "--list-operations", NULL)) { options.rsc_cmd = cmd_list_active_ops; } else if (pcmk__str_eq(option_name, "--list-agents", pcmk__str_none)) { options.rsc_cmd = cmd_list_agents; pcmk__str_update(&options.agent_spec, optarg); } else if (pcmk__str_any_of(option_name, "-o", "--list-all-operations", NULL)) { options.rsc_cmd = cmd_list_all_ops; } else if (pcmk__str_eq(option_name, "--list-ocf-alternatives", pcmk__str_none)) { options.rsc_cmd = cmd_list_alternatives; pcmk__str_update(&options.agent_spec, optarg); } else if (pcmk__str_eq(option_name, "--list-options", pcmk__str_none)) { options.rsc_cmd = cmd_list_options; return validate_opt_list(optarg); } else if (pcmk__str_any_of(option_name, "-l", "--list-raw", NULL)) { options.rsc_cmd = cmd_list_instances; } else if (pcmk__str_eq(option_name, "--list-ocf-providers", pcmk__str_none)) { options.rsc_cmd = cmd_list_providers; pcmk__str_update(&options.agent_spec, optarg); } else if (pcmk__str_any_of(option_name, "-L", "--list", NULL)) { options.rsc_cmd = cmd_list_resources; } else if (pcmk__str_eq(option_name, "--list-standards", pcmk__str_none)) { options.rsc_cmd = cmd_list_standards; } else if (pcmk__str_any_of(option_name, "-W", "--locate", NULL)) { options.rsc_cmd = cmd_locate; } else if (pcmk__str_eq(option_name, "--show-metadata", pcmk__str_none)) { options.rsc_cmd = cmd_metadata; pcmk__str_update(&options.agent_spec, optarg); } else if (pcmk__str_any_of(option_name, "-M", "--move", NULL)) { options.rsc_cmd = cmd_move; } else if (pcmk__str_any_of(option_name, "-q", "--query-xml", NULL)) { options.rsc_cmd = cmd_query_xml; } else if (pcmk__str_any_of(option_name, "-w", "--query-xml-raw", NULL)) { options.rsc_cmd = cmd_query_xml_raw; } else if (pcmk__str_any_of(option_name, "-R", "--refresh", NULL)) { options.rsc_cmd = cmd_refresh; } else if (pcmk__str_eq(option_name, "--restart", pcmk__str_none)) { options.rsc_cmd = cmd_restart; } else if (pcmk__str_any_of(option_name, "-p", "--set-parameter", NULL)) { options.rsc_cmd = cmd_set_param; pcmk__str_update(&options.prop_name, optarg); } else if (pcmk__str_eq(option_name, "--wait", pcmk__str_none)) { options.rsc_cmd = cmd_wait; } else if (pcmk__str_any_of(option_name, "-Y", "--why", NULL)) { options.rsc_cmd = cmd_why; } return TRUE; } /* short option letters still available: eEJkKXyYZ */ static GOptionEntry query_entries[] = { { "list", 'L', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "List all cluster resources with status", NULL }, { "list-raw", 'l', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "List IDs of all instantiated resources (individual members\n" INDENT "rather than groups etc.)", NULL }, { "list-cts", 'c', G_OPTION_FLAG_HIDDEN|G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, NULL, NULL }, { "list-operations", 'O', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "List active resource operations, optionally filtered by\n" INDENT "--resource and/or --node", NULL }, { "list-all-operations", 'o', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "List all resource operations, optionally filtered by\n" INDENT "--resource and/or --node", NULL }, { "list-options", 0, G_OPTION_FLAG_NONE, G_OPTION_ARG_CALLBACK, command_cb, "List all available options of the given type.\n" INDENT "Allowed values:\n" INDENT PCMK__VALUE_PRIMITIVE " (primitive resource meta-attributes),\n" INDENT PCMK_VALUE_FENCING " (parameters common to all fencing resources)", "TYPE" }, { "list-standards", 0, G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "List supported standards", NULL }, { "list-ocf-providers", 0, G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "List all available OCF providers", NULL }, { "list-agents", 0, G_OPTION_FLAG_NONE, G_OPTION_ARG_CALLBACK, command_cb, "List all agents available for the named standard and/or provider", "STD:PROV" }, { "list-ocf-alternatives", 0, G_OPTION_FLAG_NONE, G_OPTION_ARG_CALLBACK, command_cb, "List all available providers for the named OCF agent", "AGENT" }, { "show-metadata", 0, G_OPTION_FLAG_NONE, G_OPTION_ARG_CALLBACK, command_cb, "Show the metadata for the named class:provider:agent", "SPEC" }, { "query-xml", 'q', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "Show XML configuration of resource (after any template expansion)", NULL }, { "query-xml-raw", 'w', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "Show XML configuration of resource (before any template expansion)", NULL }, { "get-parameter", 'g', G_OPTION_FLAG_NONE, G_OPTION_ARG_CALLBACK, command_cb, "Display named parameter for resource (use instance attribute\n" INDENT "unless --element, --meta, or --utilization is specified)", "PARAM" }, { "locate", 'W', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "Show node(s) currently running resource", NULL }, { "constraints", 'a', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "Display the location and colocation constraints that apply to a\n" INDENT "resource, and if --recursive is specified, to the resources\n" INDENT "directly or indirectly involved in those colocations.\n" INDENT "If the named resource is part of a group, or a clone or\n" INDENT "bundle instance, constraints for the collective resource\n" INDENT "will be shown unless --force is given.", NULL }, { "stack", 'A', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "Equivalent to --constraints --recursive", NULL }, { "why", 'Y', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "Show why resources are not running, optionally filtered by\n" INDENT "--resource and/or --node", NULL }, { NULL } }; static GOptionEntry command_entries[] = { { "validate", 0, G_OPTION_FLAG_OPTIONAL_ARG, G_OPTION_ARG_CALLBACK, command_cb, "Validate resource configuration by calling agent's validate-all\n" INDENT "action. The configuration may be specified either by giving an\n" INDENT "existing resource name with -r, or by specifying --class,\n" INDENT "--agent, and --provider arguments, along with any number of\n" INDENT "--option arguments. An optional LEVEL argument can be given\n" INDENT "to control the level of checking performed.", "LEVEL" }, { "cleanup", 'C', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "If resource has any past failures, clear its history and fail\n" INDENT "count. Optionally filtered by --resource, --node, --operation\n" INDENT "and --interval (otherwise all). --operation and --interval\n" INDENT "apply to fail counts, but entire history is always clear, to\n" INDENT "allow current state to be rechecked. If the named resource is\n" INDENT "part of a group, or one numbered instance of a clone or bundled\n" INDENT "resource, the clean-up applies to the whole collective resource\n" INDENT "unless --force is given.", NULL }, { "refresh", 'R', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "Delete resource's history (including failures) so its current state\n" INDENT "is rechecked. Optionally filtered by --resource and --node\n" INDENT "(otherwise all). If the named resource is part of a group, or one\n" INDENT "numbered instance of a clone or bundled resource, the refresh\n" INDENT "applies to the whole collective resource unless --force is given.", NULL }, { "set-parameter", 'p', G_OPTION_FLAG_NONE, G_OPTION_ARG_CALLBACK, command_cb, "Set named parameter for resource (requires -v). Use instance\n" INDENT "attribute unless --element, --meta, or --utilization is " "specified.", "PARAM" }, { "delete-parameter", 'd', G_OPTION_FLAG_NONE, G_OPTION_ARG_CALLBACK, command_cb, "Delete named parameter for resource. Use instance attribute\n" INDENT "unless --element, --meta or, --utilization is specified.", "PARAM" }, { NULL } }; static GOptionEntry location_entries[] = { { "move", 'M', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "Create a constraint to move resource. If --node is specified,\n" INDENT "the constraint will be to move to that node, otherwise it\n" INDENT "will be to ban the current node. Unless --force is specified\n" INDENT "this will return an error if the resource is already running\n" INDENT "on the specified node. If --force is specified, this will\n" INDENT "always ban the current node.\n" INDENT "Optional: --lifetime, --promoted. NOTE: This may prevent the\n" INDENT "resource from running on its previous location until the\n" INDENT "implicit constraint expires or is removed with --clear.", NULL }, { "ban", 'B', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "Create a constraint to keep resource off a node.\n" INDENT "Optional: --node, --lifetime, --promoted.\n" INDENT "NOTE: This will prevent the resource from running on the\n" INDENT "affected node until the implicit constraint expires or is\n" INDENT "removed with --clear. If --node is not specified, it defaults\n" INDENT "to the node currently running the resource for primitives\n" INDENT "and groups, or the promoted instance of promotable clones with\n" INDENT PCMK_META_PROMOTED_MAX "=1 (all other situations result in an\n" INDENT "error as there is no sane default).", NULL }, { "clear", 'U', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "Remove all constraints created by the --ban and/or --move\n" INDENT "commands. Requires: --resource. Optional: --node, --promoted,\n" INDENT "--expired. If --node is not specified, all constraints created\n" INDENT "by --ban and --move will be removed for the named resource. If\n" INDENT "--node and --force are specified, any constraint created by\n" INDENT "--move will be cleared, even if it is not for the specified\n" INDENT "node. If --expired is specified, only those constraints whose\n" INDENT "lifetimes have expired will be removed.", NULL }, { "expired", 'e', G_OPTION_FLAG_NONE, G_OPTION_ARG_NONE, &options.clear_expired, "Modifies the --clear argument to remove constraints with\n" INDENT "expired lifetimes.", NULL }, { "lifetime", 'u', G_OPTION_FLAG_NONE, G_OPTION_ARG_STRING, &options.move_lifetime, "Lifespan (as ISO 8601 duration) of created constraints (with\n" INDENT "-B, -M) see https://en.wikipedia.org/wiki/ISO_8601#Durations)", "TIMESPEC" }, { "promoted", 0, G_OPTION_FLAG_NONE, G_OPTION_ARG_NONE, &options.promoted_role_only, "Limit scope of command to promoted role (with -B, -M, -U). For\n" INDENT "-B and -M, previously promoted instances may remain\n" INDENT "active in the unpromoted role.", NULL }, // Deprecated since 2.1.0 { "master", 0, G_OPTION_FLAG_NONE, G_OPTION_ARG_NONE, &options.promoted_role_only, "Deprecated: Use --promoted instead", NULL }, { NULL } }; static GOptionEntry advanced_entries[] = { { "delete", 'D', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "(Advanced) Delete a resource from the CIB. Required: -t", NULL }, { "fail", 'F', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "(Advanced) Tell the cluster this resource has failed", NULL }, { "restart", 0, G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "(Advanced) Tell the cluster to restart this resource and\n" INDENT "anything that depends on it", NULL }, { "wait", 0, G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "(Advanced) Wait until the cluster settles into a stable state", NULL }, { "digests", 0, G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "(Advanced) Show parameter hashes that Pacemaker uses to detect\n" INDENT "configuration changes (only accurate if there is resource\n" INDENT "history on the specified node). Required: --resource, --node.\n" INDENT "Optional: any NAME=VALUE parameters will be used to override\n" INDENT "the configuration (to see what the hash would be with those\n" INDENT "changes).", NULL }, { "force-demote", 0, G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "(Advanced) Bypass the cluster and demote a resource on the local\n" INDENT "node. Unless --force is specified, this will refuse to do so if\n" INDENT "the cluster believes the resource is a clone instance already\n" INDENT "running on the local node.", NULL }, { "force-stop", 0, G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "(Advanced) Bypass the cluster and stop a resource on the local node", NULL }, { "force-start", 0, G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "(Advanced) Bypass the cluster and start a resource on the local\n" INDENT "node. Unless --force is specified, this will refuse to do so if\n" INDENT "the cluster believes the resource is a clone instance already\n" INDENT "running on the local node.", NULL }, { "force-promote", 0, G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "(Advanced) Bypass the cluster and promote a resource on the local\n" INDENT "node. Unless --force is specified, this will refuse to do so if\n" INDENT "the cluster believes the resource is a clone instance already\n" INDENT "running on the local node.", NULL }, { "force-check", 0, G_OPTION_FLAG_OPTIONAL_ARG, G_OPTION_ARG_CALLBACK, command_cb, "(Advanced) Bypass the cluster and check the state of a resource on\n" INDENT "the local node. An optional LEVEL argument can be given\n" INDENT "to control the level of checking performed.", "LEVEL" }, { NULL } }; static GOptionEntry addl_entries[] = { { "node", 'N', G_OPTION_FLAG_NONE, G_OPTION_ARG_STRING, &options.host_uname, "Node name", "NAME" }, { "recursive", 0, G_OPTION_FLAG_NONE, G_OPTION_ARG_NONE, &options.recursive, "Follow colocation chains when using --set-parameter or --constraints", NULL }, { "resource-type", 't', G_OPTION_FLAG_NONE, G_OPTION_ARG_STRING, &options.rsc_type, "Resource XML element (primitive, group, etc.) (with -D)", "ELEMENT" }, { "parameter-value", 'v', G_OPTION_FLAG_NONE, G_OPTION_ARG_STRING, &options.prop_value, "Value to use with -p", "PARAM" }, { "meta", 'm', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, attr_set_type_cb, "Use resource meta-attribute instead of instance attribute\n" INDENT "(with -p, -g, -d)", NULL }, { "utilization", 'z', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, attr_set_type_cb, "Use resource utilization attribute instead of instance attribute\n" INDENT "(with -p, -g, -d)", NULL }, { "element", 0, G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, attr_set_type_cb, "Use resource element attribute instead of instance attribute\n" INDENT "(with -p, -g, -d)", NULL }, { "operation", 'n', G_OPTION_FLAG_NONE, G_OPTION_ARG_STRING, &options.operation, "Operation to clear instead of all (with -C -r)", "OPERATION" }, { "interval", 'I', G_OPTION_FLAG_NONE, G_OPTION_ARG_STRING, &options.interval_spec, "Interval of operation to clear (default 0) (with -C -r -n)", "N" }, { "class", 0, G_OPTION_FLAG_NONE, G_OPTION_ARG_CALLBACK, cmdline_config_cb, "The standard the resource agent conforms to (for example, ocf).\n" INDENT "Use with --agent, --provider, --option, and --validate.", "CLASS" }, { "agent", 0, G_OPTION_FLAG_NONE, G_OPTION_ARG_CALLBACK, cmdline_config_cb, "The agent to use (for example, IPaddr). Use with --class,\n" INDENT "--provider, --option, and --validate.", "AGENT" }, { "provider", 0, G_OPTION_FLAG_NONE, G_OPTION_ARG_CALLBACK, cmdline_config_cb, "The vendor that supplies the resource agent (for example,\n" INDENT "heartbeat). Use with --class, --agent, --option, and --validate.", "PROVIDER" }, { "option", 0, G_OPTION_FLAG_NONE, G_OPTION_ARG_CALLBACK, option_cb, "Specify a device configuration parameter as NAME=VALUE (may be\n" INDENT "specified multiple times). Use with --validate and without the\n" INDENT "-r option.", "PARAM" }, { "set-name", 's', G_OPTION_FLAG_NONE, G_OPTION_ARG_STRING, &options.prop_set, "(Advanced) XML ID of attributes element to use (with -p, -d)", "ID" }, { "nvpair", 'i', G_OPTION_FLAG_NONE, G_OPTION_ARG_STRING, &options.prop_id, "(Advanced) XML ID of nvpair element to use (with -p, -d)", "ID" }, { "timeout", 'T', G_OPTION_FLAG_NONE, G_OPTION_ARG_CALLBACK, timeout_cb, "(Advanced) Abort if command does not finish in this time (with\n" INDENT "--restart, --wait, --force-*)", "N" }, { "all", 0, G_OPTION_FLAG_NONE, G_OPTION_ARG_NONE, &options.all, "List all options, including advanced and deprecated (with\n" INDENT "--list-options)", NULL }, { "force", 'f', G_OPTION_FLAG_NONE, G_OPTION_ARG_NONE, &options.force, "Force the action to be performed. See help for individual commands for\n" INDENT "additional behavior.", NULL }, // @COMPAT Used in resource-agents prior to v4.2.0 { "host-uname", 'H', G_OPTION_FLAG_HIDDEN, G_OPTION_ARG_STRING, &options.host_uname, NULL, "HOST" }, { NULL } }; gboolean attr_set_type_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { if (pcmk__str_any_of(option_name, "-m", "--meta", NULL)) { options.attr_set_type = PCMK_XE_META_ATTRIBUTES; } else if (pcmk__str_any_of(option_name, "-z", "--utilization", NULL)) { options.attr_set_type = PCMK_XE_UTILIZATION; } else if (pcmk__str_eq(option_name, "--element", pcmk__str_none)) { options.attr_set_type = ATTR_SET_ELEMENT; } return TRUE; } gboolean cmdline_config_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { options.cmdline_config = true; if (pcmk__str_eq(option_name, "--class", pcmk__str_none)) { pcmk__str_update(&options.v_class, optarg); } else if (pcmk__str_eq(option_name, "--provider", pcmk__str_none)) { pcmk__str_update(&options.v_provider, optarg); } else { // --agent pcmk__str_update(&options.v_agent, optarg); } return TRUE; } gboolean option_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { char *name = NULL; char *value = NULL; if (pcmk__scan_nvpair(optarg, &name, &value) != 2) { return FALSE; } if (options.cmdline_params == NULL) { options.cmdline_params = pcmk__strkey_table(free, free); } g_hash_table_replace(options.cmdline_params, name, value); return TRUE; } gboolean timeout_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { long long timeout_ms = crm_get_msec(optarg); if (timeout_ms < 0) { return FALSE; } options.timeout_ms = (guint) QB_MIN(timeout_ms, UINT_MAX); return TRUE; } static int ban_or_move(pcmk__output_t *out, pcmk_resource_t *rsc, const char *move_lifetime) { int rc = pcmk_rc_ok; pcmk_node_t *current = NULL; unsigned int nactive = 0; CRM_CHECK(rsc != NULL, return EINVAL); current = pe__find_active_requires(rsc, &nactive); if (nactive == 1) { rc = cli_resource_ban(out, options.rsc_id, current->priv->name, move_lifetime, cib_conn, options.promoted_role_only, PCMK_ROLE_PROMOTED); } else if (pcmk_is_set(rsc->flags, pcmk__rsc_promotable)) { int count = 0; GList *iter = NULL; current = NULL; for (iter = rsc->priv->children; iter != NULL; iter = iter->next) { pcmk_resource_t *child = (pcmk_resource_t *)iter->data; enum rsc_role_e child_role = child->priv->fns->state(child, TRUE); if (child_role == pcmk_role_promoted) { count++; current = pcmk__current_node(child); } } if(count == 1 && current) { rc = cli_resource_ban(out, options.rsc_id, current->priv->name, move_lifetime, cib_conn, options.promoted_role_only, PCMK_ROLE_PROMOTED); } else { rc = EINVAL; g_set_error(&error, PCMK__EXITC_ERROR, CRM_EX_USAGE, _("Resource '%s' not moved: active in %d locations (promoted in %d).\n" "To prevent '%s' from running on a specific location, " "specify a node." "To prevent '%s' from being promoted at a specific " "location, specify a node and the --promoted option."), options.rsc_id, nactive, count, options.rsc_id, options.rsc_id); } } else { rc = EINVAL; g_set_error(&error, PCMK__EXITC_ERROR, CRM_EX_USAGE, _("Resource '%s' not moved: active in %d locations.\n" "To prevent '%s' from running on a specific location, " "specify a node."), options.rsc_id, nactive, options.rsc_id); } return rc; } static void cleanup(pcmk__output_t *out, pcmk_resource_t *rsc, pcmk_node_t *node) { int rc = pcmk_rc_ok; if (options.force == FALSE) { rsc = uber_parent(rsc); } crm_debug("Erasing failures of %s (%s requested) on %s", rsc->id, options.rsc_id, (options.host_uname? options.host_uname: "all nodes")); rc = cli_resource_delete(controld_api, options.host_uname, rsc, options.operation, options.interval_spec, TRUE, scheduler, options.force); if ((rc == pcmk_rc_ok) && !out->is_quiet(out)) { // Show any reasons why resource might stay stopped cli_resource_check(out, rsc, node); } if (rc == pcmk_rc_ok) { start_mainloop(controld_api); } } static int clear_constraints(pcmk__output_t *out) { GList *before = NULL; GList *after = NULL; GList *remaining = NULL; GList *ele = NULL; pcmk_node_t *dest = NULL; int rc = pcmk_rc_ok; if (!out->is_quiet(out)) { before = build_constraint_list(scheduler->input); } if (options.clear_expired) { rc = cli_resource_clear_all_expired(scheduler->input, cib_conn, options.rsc_id, options.host_uname, options.promoted_role_only); } else if (options.host_uname) { dest = pcmk_find_node(scheduler, options.host_uname); if (dest == NULL) { rc = pcmk_rc_node_unknown; if (!out->is_quiet(out)) { g_list_free(before); } return rc; } rc = cli_resource_clear(options.rsc_id, dest->priv->name, NULL, cib_conn, true, options.force); } else { rc = cli_resource_clear(options.rsc_id, NULL, scheduler->nodes, cib_conn, true, options.force); } if (!out->is_quiet(out)) { xmlNode *cib_xml = NULL; rc = cib_conn->cmds->query(cib_conn, NULL, &cib_xml, cib_sync_call); rc = pcmk_legacy2rc(rc); if (rc != pcmk_rc_ok) { g_set_error(&error, PCMK__RC_ERROR, rc, _("Could not get modified CIB: %s\n"), pcmk_rc_str(rc)); g_list_free(before); pcmk__xml_free(cib_xml); return rc; } scheduler->input = cib_xml; cluster_status(scheduler); after = build_constraint_list(scheduler->input); remaining = pcmk__subtract_lists(before, after, (GCompareFunc) strcmp); for (ele = remaining; ele != NULL; ele = ele->next) { out->info(out, "Removing constraint: %s", (char *) ele->data); } g_list_free(before); g_list_free(after); g_list_free(remaining); } return rc; } static int initialize_scheduler_data(xmlNode **cib_xml_orig) { int rc = pcmk_rc_ok; - scheduler = pe_new_working_set(); + scheduler = pcmk_new_scheduler(); if (scheduler == NULL) { return ENOMEM; } pcmk__set_scheduler_flags(scheduler, pcmk__sched_no_counts); scheduler->priv->out = out; rc = update_scheduler_input(out, scheduler, cib_conn, cib_xml_orig); if (rc != pcmk_rc_ok) { return rc; } cluster_status(scheduler); return pcmk_rc_ok; } static void list_options(void) { switch (options.opt_list) { case pcmk__opt_fencing: exit_code = pcmk_rc2exitc(pcmk__list_fencing_params(out, options.all)); break; case pcmk__opt_primitive: exit_code = pcmk_rc2exitc(pcmk__list_primitive_meta(out, options.all)); break; default: exit_code = CRM_EX_SOFTWARE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "BUG: Invalid option list type"); break; } } static int refresh(pcmk__output_t *out) { int rc = pcmk_rc_ok; const char *router_node = options.host_uname; int attr_options = pcmk__node_attr_none; if (options.host_uname) { pcmk_node_t *node = pcmk_find_node(scheduler, options.host_uname); if (pcmk__is_pacemaker_remote_node(node)) { node = pcmk__current_node(node->priv->remote); if (node == NULL) { rc = ENXIO; g_set_error(&error, PCMK__RC_ERROR, rc, _("No cluster connection to Pacemaker Remote node %s detected"), options.host_uname); return rc; } router_node = node->priv->name; attr_options |= pcmk__node_attr_remote; } } if (controld_api == NULL) { out->info(out, "Dry run: skipping clean-up of %s due to CIB_file", options.host_uname? options.host_uname : "all nodes"); rc = pcmk_rc_ok; return rc; } crm_debug("Re-checking the state of all resources on %s", options.host_uname?options.host_uname:"all nodes"); rc = pcmk__attrd_api_clear_failures(NULL, options.host_uname, NULL, NULL, NULL, NULL, attr_options); if (pcmk_controld_api_reprobe(controld_api, options.host_uname, router_node) == pcmk_rc_ok) { start_mainloop(controld_api); } return rc; } static void refresh_resource(pcmk__output_t *out, pcmk_resource_t *rsc, pcmk_node_t *node) { int rc = pcmk_rc_ok; if (options.force == FALSE) { rsc = uber_parent(rsc); } crm_debug("Re-checking the state of %s (%s requested) on %s", rsc->id, options.rsc_id, (options.host_uname? options.host_uname: "all nodes")); rc = cli_resource_delete(controld_api, options.host_uname, rsc, NULL, 0, FALSE, scheduler, options.force); if ((rc == pcmk_rc_ok) && !out->is_quiet(out)) { // Show any reasons why resource might stay stopped cli_resource_check(out, rsc, node); } if (rc == pcmk_rc_ok) { start_mainloop(controld_api); } } static int show_metadata(pcmk__output_t *out, const char *agent_spec) { int rc = pcmk_rc_ok; char *standard = NULL; char *provider = NULL; char *type = NULL; char *metadata = NULL; lrmd_t *lrmd_conn = NULL; rc = lrmd__new(&lrmd_conn, NULL, NULL, 0); if (rc != pcmk_rc_ok) { g_set_error(&error, PCMK__RC_ERROR, rc, _("Could not create executor connection")); lrmd_api_delete(lrmd_conn); return rc; } rc = crm_parse_agent_spec(agent_spec, &standard, &provider, &type); rc = pcmk_legacy2rc(rc); if (rc == pcmk_rc_ok) { rc = lrmd_conn->cmds->get_metadata(lrmd_conn, standard, provider, type, &metadata, 0); rc = pcmk_legacy2rc(rc); if (metadata) { out->output_xml(out, PCMK_XE_METADATA, metadata); free(metadata); } else { /* We were given a validly formatted spec, but it doesn't necessarily * match up with anything that exists. Use ENXIO as the return code * here because that maps to an exit code of CRM_EX_NOSUCH, which * probably is the most common reason to get here. */ rc = ENXIO; g_set_error(&error, PCMK__RC_ERROR, rc, _("Metadata query for %s failed: %s"), agent_spec, pcmk_rc_str(rc)); } } else { rc = ENXIO; g_set_error(&error, PCMK__RC_ERROR, rc, _("'%s' is not a valid agent specification"), agent_spec); } lrmd_api_delete(lrmd_conn); return rc; } static void validate_cmdline_config(void) { // Cannot use both --resource and command-line resource configuration if (options.rsc_id != NULL) { g_set_error(&error, PCMK__EXITC_ERROR, CRM_EX_USAGE, _("--resource cannot be used with --class, --agent, and --provider")); // Not all commands support command-line resource configuration } else if (options.rsc_cmd != cmd_execute_agent) { g_set_error(&error, PCMK__EXITC_ERROR, CRM_EX_USAGE, _("--class, --agent, and --provider can only be used with " "--validate and --force-*")); // Not all of --class, --agent, and --provider need to be given. Not all // classes support the concept of a provider. Check that what we were given // is valid. } else if (pcmk__str_eq(options.v_class, "stonith", pcmk__str_none)) { if (options.v_provider != NULL) { g_set_error(&error, PCMK__EXITC_ERROR, CRM_EX_USAGE, _("stonith does not support providers")); } else if (stonith_agent_exists(options.v_agent, 0) == FALSE) { g_set_error(&error, PCMK__EXITC_ERROR, CRM_EX_USAGE, _("%s is not a known stonith agent"), options.v_agent ? options.v_agent : ""); } } else if (resources_agent_exists(options.v_class, options.v_provider, options.v_agent) == FALSE) { g_set_error(&error, PCMK__EXITC_ERROR, CRM_EX_USAGE, _("%s:%s:%s is not a known resource"), options.v_class ? options.v_class : "", options.v_provider ? options.v_provider : "", options.v_agent ? options.v_agent : ""); } if ((error == NULL) && (options.cmdline_params == NULL)) { options.cmdline_params = pcmk__strkey_table(free, free); } } /*! * \internal * \brief Get the enum pe_find flags for a given command * * \return enum pe_find flag group appropriate for \c options.rsc_cmd. */ static uint32_t get_find_flags(void) { switch (options.rsc_cmd) { case cmd_ban: case cmd_cleanup: case cmd_clear: case cmd_colocations: case cmd_digests: case cmd_execute_agent: case cmd_locate: case cmd_move: case cmd_refresh: case cmd_restart: case cmd_why: return pcmk_rsc_match_history|pcmk_rsc_match_anon_basename; case cmd_delete_param: case cmd_get_param: case cmd_query_xml_raw: case cmd_query_xml: case cmd_set_param: return pcmk_rsc_match_history|pcmk_rsc_match_basename; default: return 0; } } /*! * \internal * \brief Check whether a node argument is required * * \return \c true if a \c --node argument is required, or \c false otherwise */ static bool is_node_required(void) { switch (options.rsc_cmd) { case cmd_digests: case cmd_fail: return true; default: return false; } } /*! * \internal * \brief Check whether a resource argument is required * * \return \c true if a \c --resource argument is required, or \c false * otherwise */ static bool is_resource_required(void) { if (options.cmdline_config) { return false; } switch (options.rsc_cmd) { case cmd_clear: return !options.clear_expired; case cmd_cleanup: case cmd_cts: case cmd_list_active_ops: case cmd_list_agents: case cmd_list_all_ops: case cmd_list_alternatives: case cmd_list_instances: case cmd_list_options: case cmd_list_providers: case cmd_list_resources: case cmd_list_standards: case cmd_metadata: case cmd_refresh: case cmd_wait: case cmd_why: return false; default: return true; } } /*! * \internal * \brief Check whether a CIB connection is required * * \return \c true if a CIB connection is required, or \c false otherwise */ static bool is_cib_required(void) { if (options.cmdline_config) { return false; } switch (options.rsc_cmd) { case cmd_list_agents: case cmd_list_alternatives: case cmd_list_options: case cmd_list_providers: case cmd_list_standards: case cmd_metadata: return false; default: return true; } } /*! * \internal * \brief Check whether a controller IPC connection is required * * \return \c true if a controller connection is required, or \c false otherwise */ static bool is_controller_required(void) { switch (options.rsc_cmd) { case cmd_cleanup: case cmd_refresh: return getenv("CIB_file") == NULL; case cmd_fail: return true; default: return false; } } /*! * \internal * \brief Check whether a scheduler IPC connection is required * * \return \c true if a scheduler connection is required, or \c false otherwise */ static bool is_scheduler_required(void) { if (options.cmdline_config) { return false; } switch (options.rsc_cmd) { case cmd_delete: case cmd_list_agents: case cmd_list_alternatives: case cmd_list_options: case cmd_list_providers: case cmd_list_standards: case cmd_metadata: case cmd_wait: return false; default: return true; } } /*! * \internal * \brief Check whether the chosen command accepts clone instances * * \return \c true if \p options.rsc_cmd accepts or ignores clone instances, or * \c false otherwise */ static bool accept_clone_instance(void) { switch (options.rsc_cmd) { case cmd_ban: case cmd_clear: case cmd_delete: case cmd_move: case cmd_restart: return false; default: return true; } } static GOptionContext * build_arg_context(pcmk__common_args_t *args, GOptionGroup **group) { GOptionContext *context = NULL; GOptionEntry extra_prog_entries[] = { { "quiet", 'Q', G_OPTION_FLAG_NONE, G_OPTION_ARG_NONE, &(args->quiet), "Be less descriptive in output.", NULL }, { "resource", 'r', G_OPTION_FLAG_NONE, G_OPTION_ARG_STRING, &options.rsc_id, "Resource ID", "ID" }, { G_OPTION_REMAINING, 0, G_OPTION_FLAG_NONE, G_OPTION_ARG_STRING_ARRAY, &options.remainder, NULL, NULL }, { NULL } }; const char *description = "Examples:\n\n" "List the available OCF agents:\n\n" "\t# crm_resource --list-agents ocf\n\n" "List the available OCF agents from the linux-ha project:\n\n" "\t# crm_resource --list-agents ocf:heartbeat\n\n" "Move 'myResource' to a specific node:\n\n" "\t# crm_resource --resource myResource --move --node altNode\n\n" "Allow (but not force) 'myResource' to move back to its original " "location:\n\n" "\t# crm_resource --resource myResource --clear\n\n" "Stop 'myResource' (and anything that depends on it):\n\n" "\t# crm_resource --resource myResource --set-parameter " PCMK_META_TARGET_ROLE "--meta --parameter-value Stopped\n\n" "Tell the cluster not to manage 'myResource' (the cluster will not " "attempt to start or stop the\n" "resource under any circumstances; useful when performing maintenance " "tasks on a resource):\n\n" "\t# crm_resource --resource myResource --set-parameter " PCMK_META_IS_MANAGED "--meta --parameter-value false\n\n" "Erase the operation history of 'myResource' on 'aNode' (the cluster " "will 'forget' the existing\n" "resource state, including any errors, and attempt to recover the" "resource; useful when a resource\n" "had failed permanently and has been repaired by an administrator):\n\n" "\t# crm_resource --resource myResource --cleanup --node aNode\n\n"; context = pcmk__build_arg_context(args, "text (default), xml", group, NULL); g_option_context_set_description(context, description); /* Add the -Q option, which cannot be part of the globally supported options * because some tools use that flag for something else. */ pcmk__add_main_args(context, extra_prog_entries); pcmk__add_arg_group(context, "queries", "Queries:", "Show query help", query_entries); pcmk__add_arg_group(context, "commands", "Commands:", "Show command help", command_entries); pcmk__add_arg_group(context, "locations", "Locations:", "Show location help", location_entries); pcmk__add_arg_group(context, "advanced", "Advanced:", "Show advanced option help", advanced_entries); pcmk__add_arg_group(context, "additional", "Additional Options:", "Show additional options", addl_entries); return context; } int main(int argc, char **argv) { xmlNode *cib_xml_orig = NULL; pcmk_resource_t *rsc = NULL; pcmk_node_t *node = NULL; uint32_t find_flags = 0; int rc = pcmk_rc_ok; GOptionGroup *output_group = NULL; gchar **processed_args = NULL; GOptionContext *context = NULL; /* * Parse command line arguments */ args = pcmk__new_common_args(SUMMARY); processed_args = pcmk__cmdline_preproc(argv, "GHINSTdginpstuvx"); context = build_arg_context(args, &output_group); pcmk__register_formats(output_group, formats); if (!g_option_context_parse_strv(context, &processed_args, &error)) { exit_code = CRM_EX_USAGE; goto done; } pcmk__cli_init_logging("crm_resource", args->verbosity); rc = pcmk__output_new(&out, args->output_ty, args->output_dest, argv); if (rc != pcmk_rc_ok) { exit_code = CRM_EX_ERROR; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, _("Error creating output format %s: %s"), args->output_ty, pcmk_rc_str(rc)); goto done; } pe__register_messages(out); crm_resource_register_messages(out); lrmd__register_messages(out); pcmk__register_lib_messages(out); out->quiet = args->quiet; crm_log_args(argc, argv); /* * Validate option combinations */ // --expired without --clear/-U doesn't make sense if (options.clear_expired && (options.rsc_cmd != cmd_clear)) { exit_code = CRM_EX_USAGE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, _("--expired requires --clear or -U")); goto done; } if ((options.remainder != NULL) && (options.override_params != NULL)) { // Commands that use positional arguments will create override_params for (gchar **s = options.remainder; *s; s++) { char *name = pcmk__assert_alloc(1, strlen(*s)); char *value = pcmk__assert_alloc(1, strlen(*s)); int rc = sscanf(*s, "%[^=]=%s", name, value); if (rc == 2) { g_hash_table_replace(options.override_params, name, value); } else { exit_code = CRM_EX_USAGE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, _("Error parsing '%s' as a name=value pair"), argv[optind]); free(value); free(name); goto done; } } } else if (options.remainder != NULL) { gchar **strv = NULL; gchar *msg = NULL; int i = 1; int len = 0; for (gchar **s = options.remainder; *s; s++) { len++; } pcmk__assert(len > 0); /* Add 1 for the strv[0] string below, and add another 1 for the NULL * at the end of the array so g_strjoinv knows when to stop. */ strv = pcmk__assert_alloc(len+2, sizeof(char *)); strv[0] = strdup("non-option ARGV-elements:\n"); for (gchar **s = options.remainder; *s; s++) { strv[i] = crm_strdup_printf("[%d of %d] %s\n", i, len, *s); i++; } strv[i] = NULL; exit_code = CRM_EX_USAGE; msg = g_strjoinv("", strv); g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "%s", msg); g_free(msg); /* Don't try to free the last element, which is just NULL. */ for(i = 0; i < len+1; i++) { free(strv[i]); } free(strv); goto done; } if (pcmk__str_eq(args->output_ty, "xml", pcmk__str_none)) { switch (options.rsc_cmd) { /* These are the only commands that have historically used the * elements in their XML schema. For all others, use the simple list * argument. */ case cmd_get_param: case cmd_list_instances: case cmd_list_standards: pcmk__output_enable_list_element(out); break; default: break; } } else if (pcmk__str_eq(args->output_ty, "text", pcmk__str_null_matches)) { switch (options.rsc_cmd) { case cmd_colocations: case cmd_list_resources: pcmk__output_text_set_fancy(out, true); break; default: break; } } if (args->version) { out->version(out, false); goto done; } if (options.cmdline_config) { /* A resource configuration was given on the command line. Sanity-check * the values and set error if they don't make sense. */ validate_cmdline_config(); if (error != NULL) { exit_code = CRM_EX_USAGE; goto done; } } else if (options.cmdline_params != NULL) { exit_code = CRM_EX_USAGE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, _("--option must be used with --validate and without -r")); g_hash_table_destroy(options.cmdline_params); options.cmdline_params = NULL; goto done; } if (is_resource_required() && (options.rsc_id == NULL)) { exit_code = CRM_EX_USAGE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, _("Must supply a resource id with -r")); goto done; } if (is_node_required() && (options.host_uname == NULL)) { exit_code = CRM_EX_USAGE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, _("Must supply a node name with -N")); goto done; } /* * Set up necessary connections */ // Establish a connection to the CIB if needed if (is_cib_required()) { cib_conn = cib_new(); if ((cib_conn == NULL) || (cib_conn->cmds == NULL)) { exit_code = CRM_EX_DISCONNECT; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, _("Could not create CIB connection")); goto done; } rc = cib__signon_attempts(cib_conn, cib_command, 5); rc = pcmk_legacy2rc(rc); if (rc != pcmk_rc_ok) { exit_code = pcmk_rc2exitc(rc); g_set_error(&error, PCMK__EXITC_ERROR, exit_code, _("Could not connect to the CIB: %s"), pcmk_rc_str(rc)); goto done; } } // Populate scheduler data from XML file if specified or CIB query otherwise if (is_scheduler_required()) { rc = initialize_scheduler_data(&cib_xml_orig); if (rc != pcmk_rc_ok) { exit_code = pcmk_rc2exitc(rc); goto done; } } find_flags = get_find_flags(); // If command requires that resource exist if specified, find it if ((find_flags != 0) && (options.rsc_id != NULL)) { rsc = pe_find_resource_with_flags(scheduler->priv->resources, options.rsc_id, find_flags); if (rsc == NULL) { exit_code = CRM_EX_NOSUCH; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, _("Resource '%s' not found"), options.rsc_id); goto done; } /* The --ban, --clear, --move, and --restart commands do not work with * instances of clone resourcs. */ if (pcmk__is_clone(rsc->priv->parent) && (strchr(options.rsc_id, ':') != NULL) && !accept_clone_instance()) { exit_code = CRM_EX_INVALID_PARAM; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, _("Cannot operate on clone resource instance '%s'"), options.rsc_id); goto done; } } // If user supplied a node name, check whether it exists if ((options.host_uname != NULL) && (scheduler != NULL)) { node = pcmk_find_node(scheduler, options.host_uname); if (node == NULL) { exit_code = CRM_EX_NOSUCH; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, _("Node '%s' not found"), options.host_uname); goto done; } } // Establish a connection to the controller if needed if (is_controller_required()) { rc = pcmk_new_ipc_api(&controld_api, pcmk_ipc_controld); if (rc != pcmk_rc_ok) { exit_code = pcmk_rc2exitc(rc); g_set_error(&error, PCMK__EXITC_ERROR, exit_code, _("Error connecting to the controller: %s"), pcmk_rc_str(rc)); goto done; } pcmk_register_ipc_callback(controld_api, controller_event_callback, NULL); rc = pcmk__connect_ipc(controld_api, pcmk_ipc_dispatch_main, 5); if (rc != pcmk_rc_ok) { exit_code = pcmk_rc2exitc(rc); g_set_error(&error, PCMK__EXITC_ERROR, exit_code, _("Error connecting to %s: %s"), pcmk_ipc_name(controld_api, true), pcmk_rc_str(rc)); goto done; } } /* * Handle requested command */ switch (options.rsc_cmd) { case cmd_list_resources: { GList *all = NULL; uint32_t show_opts = pcmk_show_inactive_rscs | pcmk_show_rsc_only | pcmk_show_pending; all = g_list_prepend(all, (gpointer) "*"); rc = out->message(out, "resource-list", scheduler, show_opts, true, all, all, false); g_list_free(all); if (rc == pcmk_rc_no_output) { rc = ENXIO; } break; } case cmd_list_instances: rc = out->message(out, "resource-names-list", scheduler->priv->resources); if (rc != pcmk_rc_ok) { rc = ENXIO; } break; case cmd_list_options: list_options(); break; case cmd_list_alternatives: rc = pcmk__list_alternatives(out, options.agent_spec); break; case cmd_list_agents: rc = pcmk__list_agents(out, options.agent_spec); break; case cmd_list_standards: rc = pcmk__list_standards(out); break; case cmd_list_providers: rc = pcmk__list_providers(out, options.agent_spec); break; case cmd_metadata: rc = show_metadata(out, options.agent_spec); break; case cmd_restart: /* We don't pass scheduler because rsc needs to stay valid for the * entire lifetime of cli_resource_restart(), but it will reset and * update the scheduler data multiple times, so it needs to use its * own copy. */ rc = cli_resource_restart(out, rsc, node, options.move_lifetime, options.timeout_ms, cib_conn, options.promoted_role_only, options.force); break; case cmd_wait: rc = wait_till_stable(out, options.timeout_ms, cib_conn); break; case cmd_execute_agent: if (options.cmdline_config) { exit_code = cli_resource_execute_from_params(out, NULL, options.v_class, options.v_provider, options.v_agent, options.operation, options.cmdline_params, options.override_params, options.timeout_ms, args->verbosity, options.force, options.check_level); } else { exit_code = cli_resource_execute(rsc, options.rsc_id, options.operation, options.override_params, options.timeout_ms, cib_conn, scheduler, args->verbosity, options.force, options.check_level); } goto done; case cmd_digests: node = pcmk_find_node(scheduler, options.host_uname); if (node == NULL) { rc = pcmk_rc_node_unknown; } else { rc = pcmk__resource_digests(out, rsc, node, options.override_params); } break; case cmd_colocations: rc = out->message(out, "locations-and-colocations", rsc, options.recursive, (bool) options.force); break; case cmd_cts: rc = pcmk_rc_ok; g_list_foreach(scheduler->priv->resources, (GFunc) cli_resource_print_cts, out); cli_resource_print_cts_constraints(scheduler); break; case cmd_fail: rc = cli_resource_fail(controld_api, options.host_uname, options.rsc_id, scheduler); if (rc == pcmk_rc_ok) { start_mainloop(controld_api); } break; case cmd_list_active_ops: rc = cli_resource_print_operations(options.rsc_id, options.host_uname, TRUE, scheduler); break; case cmd_list_all_ops: rc = cli_resource_print_operations(options.rsc_id, options.host_uname, FALSE, scheduler); break; case cmd_locate: { GList *nodes = cli_resource_search(rsc, options.rsc_id, scheduler); rc = out->message(out, "resource-search-list", nodes, options.rsc_id); g_list_free_full(nodes, free); break; } case cmd_query_xml: rc = cli_resource_print(rsc, scheduler, true); break; case cmd_query_xml_raw: rc = cli_resource_print(rsc, scheduler, false); break; case cmd_why: if ((options.host_uname != NULL) && (node == NULL)) { rc = pcmk_rc_node_unknown; } else { rc = out->message(out, "resource-reasons-list", scheduler->priv->resources, rsc, node); } break; case cmd_clear: rc = clear_constraints(out); break; case cmd_move: if (options.host_uname == NULL) { rc = ban_or_move(out, rsc, options.move_lifetime); } else { rc = cli_resource_move(rsc, options.rsc_id, options.host_uname, options.move_lifetime, cib_conn, scheduler, options.promoted_role_only, options.force); } if (rc == EINVAL) { exit_code = CRM_EX_USAGE; goto done; } break; case cmd_ban: if (options.host_uname == NULL) { rc = ban_or_move(out, rsc, options.move_lifetime); } else if (node == NULL) { rc = pcmk_rc_node_unknown; } else { rc = cli_resource_ban(out, options.rsc_id, node->priv->name, options.move_lifetime, cib_conn, options.promoted_role_only, PCMK_ROLE_PROMOTED); } if (rc == EINVAL) { exit_code = CRM_EX_USAGE; goto done; } break; case cmd_get_param: { unsigned int count = 0; GHashTable *params = NULL; pcmk_node_t *current = rsc->priv->fns->active_node(rsc, &count, NULL); bool free_params = true; const char* value = NULL; if (count > 1) { out->err(out, "%s is active on more than one node," " returning the default value for %s", rsc->id, pcmk__s(options.prop_name, "unspecified property")); current = NULL; } crm_debug("Looking up %s in %s", options.prop_name, rsc->id); if (pcmk__str_eq(options.attr_set_type, PCMK_XE_INSTANCE_ATTRIBUTES, pcmk__str_none)) { params = pe_rsc_params(rsc, current, scheduler); free_params = false; value = g_hash_table_lookup(params, options.prop_name); } else if (pcmk__str_eq(options.attr_set_type, PCMK_XE_META_ATTRIBUTES, pcmk__str_none)) { params = pcmk__strkey_table(free, free); get_meta_attributes(params, rsc, NULL, scheduler); value = g_hash_table_lookup(params, options.prop_name); } else if (pcmk__str_eq(options.attr_set_type, ATTR_SET_ELEMENT, pcmk__str_none)) { value = crm_element_value(rsc->priv->xml, options.prop_name); free_params = false; } else { const pcmk_rule_input_t rule_input = { .now = scheduler->priv->now, }; params = pcmk__strkey_table(free, free); pe__unpack_dataset_nvpairs(rsc->priv->xml, PCMK_XE_UTILIZATION, &rule_input, params, NULL, scheduler); value = g_hash_table_lookup(params, options.prop_name); } rc = out->message(out, "attribute-list", rsc, options.prop_name, value); if (free_params) { g_hash_table_destroy(params); } break; } case cmd_set_param: if (pcmk__str_empty(options.prop_value)) { exit_code = CRM_EX_USAGE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, _("You need to supply a value with the -v option")); goto done; } /* coverity[var_deref_model] False positive */ rc = cli_resource_update_attribute(rsc, options.rsc_id, options.prop_set, options.attr_set_type, options.prop_id, options.prop_name, options.prop_value, options.recursive, cib_conn, cib_xml_orig, options.force); break; case cmd_delete_param: /* coverity[var_deref_model] False positive */ rc = cli_resource_delete_attribute(rsc, options.rsc_id, options.prop_set, options.attr_set_type, options.prop_id, options.prop_name, cib_conn, cib_xml_orig, options.force); break; case cmd_cleanup: if (rsc == NULL) { rc = cli_cleanup_all(controld_api, options.host_uname, options.operation, options.interval_spec, scheduler); if (rc == pcmk_rc_ok) { start_mainloop(controld_api); } } else { cleanup(out, rsc, node); } break; case cmd_refresh: if (rsc == NULL) { rc = refresh(out); } else { refresh_resource(out, rsc, node); } break; case cmd_delete: /* rsc_id was already checked for NULL much earlier when validating * command line arguments. */ if (options.rsc_type == NULL) { exit_code = CRM_EX_USAGE; g_set_error(&error, PCMK__EXITC_ERROR, CRM_EX_USAGE, _("You need to specify a resource type with -t")); } else { rc = pcmk__resource_delete(cib_conn, cib_sync_call, options.rsc_id, options.rsc_type); if (rc != pcmk_rc_ok) { g_set_error(&error, PCMK__RC_ERROR, rc, _("Could not delete resource %s: %s"), options.rsc_id, pcmk_rc_str(rc)); } } break; default: exit_code = CRM_EX_USAGE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, _("Unimplemented command: %d"), (int) options.rsc_cmd); goto done; } /* Convert rc into an exit code. */ if (rc != pcmk_rc_ok && rc != pcmk_rc_no_output) { exit_code = pcmk_rc2exitc(rc); } /* * Clean up and exit */ done: /* When we get here, exit_code has been set one of two ways - either at one of * the spots where there's a "goto done" (which itself could have happened either * directly or by calling pcmk_rc2exitc), or just up above after any of the break * statements. * * Thus, we can use just exit_code here to decide what to do. */ if (exit_code != CRM_EX_OK && exit_code != CRM_EX_USAGE) { if (error != NULL) { char *msg = crm_strdup_printf("%s\nError performing operation: %s", error->message, crm_exit_str(exit_code)); g_clear_error(&error); g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "%s", msg); free(msg); } else { g_set_error(&error, PCMK__EXITC_ERROR, exit_code, _("Error performing operation: %s"), crm_exit_str(exit_code)); } } pcmk__xml_free(cib_xml_orig); g_free(options.host_uname); g_free(options.interval_spec); g_free(options.move_lifetime); g_free(options.operation); g_free(options.prop_id); free(options.prop_name); g_free(options.prop_set); g_free(options.prop_value); g_free(options.rsc_id); g_free(options.rsc_type); free(options.agent_spec); free(options.v_agent); free(options.v_class); free(options.v_provider); g_strfreev(options.remainder); if (options.override_params != NULL) { g_hash_table_destroy(options.override_params); } /* options.cmdline_params does not need to be destroyed here. See the * comments in cli_resource_execute_from_params. */ g_strfreev(processed_args); g_option_context_free(context); return bye(exit_code); } diff --git a/tools/crm_resource_runtime.c b/tools/crm_resource_runtime.c index 0c1384ddf4..ddd780f04c 100644 --- a/tools/crm_resource_runtime.c +++ b/tools/crm_resource_runtime.c @@ -1,2498 +1,2498 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include static GList * build_node_info_list(const pcmk_resource_t *rsc) { GList *retval = NULL; for (const GList *iter = rsc->priv->children; iter != NULL; iter = iter->next) { const pcmk_resource_t *child = (const pcmk_resource_t *) iter->data; for (const GList *iter2 = child->priv->active_nodes; iter2 != NULL; iter2 = iter2->next) { const pcmk_node_t *node = (const pcmk_node_t *) iter2->data; node_info_t *ni = pcmk__assert_alloc(1, sizeof(node_info_t)); ni->node_name = node->priv->name; if (pcmk_is_set(rsc->flags, pcmk__rsc_promotable) && (child->priv->fns->state(child, TRUE) == pcmk_role_promoted)) { ni->promoted = true; } retval = g_list_prepend(retval, ni); } } return retval; } GList * cli_resource_search(pcmk_resource_t *rsc, const char *requested_name, pcmk_scheduler_t *scheduler) { GList *retval = NULL; const pcmk_resource_t *parent = pe__const_top_resource(rsc, false); if (pcmk__is_clone(rsc)) { retval = build_node_info_list(rsc); /* The anonymous clone children's common ID is supplied */ } else if (pcmk__is_clone(parent) && !pcmk_is_set(rsc->flags, pcmk__rsc_unique) && (rsc->priv->history_id != NULL) && pcmk__str_eq(requested_name, rsc->priv->history_id, pcmk__str_none) && !pcmk__str_eq(requested_name, rsc->id, pcmk__str_none)) { retval = build_node_info_list(parent); } else { for (GList *iter = rsc->priv->active_nodes; iter != NULL; iter = iter->next) { pcmk_node_t *node = (pcmk_node_t *) iter->data; node_info_t *ni = pcmk__assert_alloc(1, sizeof(node_info_t)); ni->node_name = node->priv->name; if (rsc->priv->fns->state(rsc, TRUE) == pcmk_role_promoted) { ni->promoted = true; } retval = g_list_prepend(retval, ni); } } return retval; } // \return Standard Pacemaker return code static int find_resource_attr(pcmk__output_t *out, cib_t * the_cib, const char *attr, const char *rsc, const char *attr_set_type, const char *set_name, const char *attr_id, const char *attr_name, xmlNode **result) { xmlNode *xml_search; int rc = pcmk_rc_ok; GString *xpath = NULL; const char *xpath_base = NULL; if (result) { *result = NULL; } if(the_cib == NULL) { return ENOTCONN; } xpath_base = pcmk_cib_xpath_for(PCMK_XE_RESOURCES); if (xpath_base == NULL) { crm_err(PCMK_XE_RESOURCES " CIB element not known (bug?)"); return ENOMSG; } xpath = g_string_sized_new(1024); pcmk__g_strcat(xpath, xpath_base, "//*[@" PCMK_XA_ID "=\"", rsc, "\"]", NULL); if (attr_set_type != NULL) { pcmk__g_strcat(xpath, "/", attr_set_type, NULL); if (set_name != NULL) { pcmk__g_strcat(xpath, "[@" PCMK_XA_ID "=\"", set_name, "\"]", NULL); } } g_string_append(xpath, "//" PCMK_XE_NVPAIR); if (attr_id != NULL && attr_name!= NULL) { pcmk__g_strcat(xpath, "[@" PCMK_XA_ID "='", attr_id, "' " "and @" PCMK_XA_NAME "='", attr_name, "']", NULL); } else if (attr_id != NULL) { pcmk__g_strcat(xpath, "[@" PCMK_XA_ID "='", attr_id, "']", NULL); } else if (attr_name != NULL) { pcmk__g_strcat(xpath, "[@" PCMK_XA_NAME "='", attr_name, "']", NULL); } rc = the_cib->cmds->query(the_cib, xpath->str, &xml_search, cib_sync_call|cib_xpath); rc = pcmk_legacy2rc(rc); if (rc == pcmk_rc_ok) { crm_log_xml_debug(xml_search, "Match"); if (xml_search->children != NULL) { rc = ENOTUNIQ; pcmk__warn_multiple_name_matches(out, xml_search, attr_name); out->spacer(out); } } if (result) { *result = xml_search; } else { pcmk__xml_free(xml_search); } g_string_free(xpath, TRUE); return rc; } /* PRIVATE. Use the find_matching_attr_resources instead. */ static void find_matching_attr_resources_recursive(pcmk__output_t *out, GList /* */ **result, pcmk_resource_t *rsc, const char * attr_set, const char * attr_set_type, const char * attr_id, const char * attr_name, cib_t * cib, int depth) { int rc = pcmk_rc_ok; char *lookup_id = clone_strip(rsc->id); for (GList *gIter = rsc->priv->children; gIter != NULL; gIter = gIter->next) { find_matching_attr_resources_recursive(out, result, (pcmk_resource_t *) gIter->data, attr_set, attr_set_type, attr_id, attr_name, cib, depth+1); /* do it only once for clones */ if (pcmk__is_clone(rsc)) { break; } } rc = find_resource_attr(out, cib, PCMK_XA_ID, lookup_id, attr_set_type, attr_set, attr_id, attr_name, NULL); /* Post-order traversal. * The root is always on the list and it is the last item. */ if((0 == depth) || (pcmk_rc_ok == rc)) { /* push the head */ *result = g_list_append(*result, rsc); } free(lookup_id); } /* The result is a linearized pre-ordered tree of resources. */ static GList/**/ * find_matching_attr_resources(pcmk__output_t *out, pcmk_resource_t *rsc, const char * rsc_id, const char * attr_set, const char * attr_set_type, const char * attr_id, const char * attr_name, cib_t * cib, const char * cmd, gboolean force) { int rc = pcmk_rc_ok; char *lookup_id = NULL; GList * result = NULL; /* If --force is used, update only the requested resource (clone or primitive). * Otherwise, if the primitive has the attribute, use that. * Otherwise use the clone. */ if(force == TRUE) { return g_list_append(result, rsc); } if (pcmk__is_clone(rsc->priv->parent)) { int rc = find_resource_attr(out, cib, PCMK_XA_ID, rsc_id, attr_set_type, attr_set, attr_id, attr_name, NULL); if(rc != pcmk_rc_ok) { rsc = rsc->priv->parent; out->info(out, "Performing %s of '%s' on '%s', the parent of '%s'", cmd, attr_name, rsc->id, rsc_id); } return g_list_append(result, rsc); } else if ((rsc->priv->parent == NULL) && (rsc->priv->children != NULL) && pcmk__is_clone(rsc)) { pcmk_resource_t *child = rsc->priv->children->data; if (pcmk__is_primitive(child)) { lookup_id = clone_strip(child->id); /* Could be a cloned group! */ rc = find_resource_attr(out, cib, PCMK_XA_ID, lookup_id, attr_set_type, attr_set, attr_id, attr_name, NULL); if(rc == pcmk_rc_ok) { rsc = child; out->info(out, "A value for '%s' already exists in child '%s', performing %s on that instead of '%s'", attr_name, lookup_id, cmd, rsc_id); } free(lookup_id); } return g_list_append(result, rsc); } /* If the resource is a group ==> children inherit the attribute if defined. */ find_matching_attr_resources_recursive(out, &result, rsc, attr_set, attr_set_type, attr_id, attr_name, cib, 0); return result; } /*! * \internal * \brief Get a resource's XML by resource ID from a given CIB XML tree * * \param[in] cib_xml CIB XML to search * \param[in] rsc Resource whose XML to get * * \return Subtree of \p cib_xml belonging to \p rsc, or \c NULL if not found */ static xmlNode * get_cib_rsc(xmlNode *cib_xml, const pcmk_resource_t *rsc) { char *xpath = crm_strdup_printf("%s//*[@" PCMK_XA_ID "='%s']", pcmk_cib_xpath_for(PCMK_XE_RESOURCES), pcmk__xe_id(rsc->priv->xml)); xmlNode *rsc_xml = get_xpath_object(xpath, cib_xml, LOG_ERR); free(xpath); return rsc_xml; } static int update_element_attribute(pcmk__output_t *out, pcmk_resource_t *rsc, cib_t *cib, xmlNode *cib_xml_orig, const char *attr_name, const char *attr_value) { int rc = pcmk_rc_ok; xmlNode *rsc_xml = rsc->priv->xml; rsc_xml = get_cib_rsc(cib_xml_orig, rsc); if (rsc_xml == NULL) { return ENXIO; } crm_xml_add(rsc_xml, attr_name, attr_value); rc = cib->cmds->replace(cib, PCMK_XE_RESOURCES, rsc_xml, cib_sync_call); rc = pcmk_legacy2rc(rc); if (rc == pcmk_rc_ok) { out->info(out, "Set attribute: " PCMK_XA_NAME "=%s value=%s", attr_name, attr_value); } return rc; } static int resources_with_attr(pcmk__output_t *out, cib_t *cib, pcmk_resource_t *rsc, const char *requested_name, const char *attr_set, const char *attr_set_type, const char *attr_id, const char *attr_name, const char *top_id, gboolean force, GList **resources) { if (pcmk__str_eq(attr_set_type, PCMK_XE_INSTANCE_ATTRIBUTES, pcmk__str_casei)) { if (!force) { xmlNode *xml_search = NULL; int rc = pcmk_rc_ok; rc = find_resource_attr(out, cib, PCMK_XA_ID, top_id, PCMK_XE_META_ATTRIBUTES, attr_set, attr_id, attr_name, &xml_search); if (rc == pcmk_rc_ok || rc == ENOTUNIQ) { char *found_attr_id = NULL; found_attr_id = crm_element_value_copy(xml_search, PCMK_XA_ID); if (!out->is_quiet(out)) { out->err(out, "WARNING: There is already a meta attribute " "for '%s' called '%s' (id=%s)", top_id, attr_name, found_attr_id); out->err(out, " Delete '%s' first or use the force option " "to override", found_attr_id); } free(found_attr_id); pcmk__xml_free(xml_search); return ENOTUNIQ; } pcmk__xml_free(xml_search); } *resources = g_list_append(*resources, rsc); } else { *resources = find_matching_attr_resources(out, rsc, requested_name, attr_set, attr_set_type, attr_id, attr_name, cib, "update", force); } /* If the user specified attr_set or attr_id, the intent is to modify a * single resource, which will be the last item in the list. */ if ((attr_set != NULL) || (attr_id != NULL)) { GList *last = g_list_last(*resources); *resources = g_list_remove_link(*resources, last); g_list_free(*resources); *resources = last; } return pcmk_rc_ok; } static void free_attr_update_data(gpointer data) { attr_update_data_t *ud = data; if (ud == NULL) { return; } free(ud->attr_set_type); free(ud->attr_set_id); free(ud->attr_name); free(ud->attr_value); free(ud->given_rsc_id); free(ud->found_attr_id); free(ud); } static int update_attribute(pcmk_resource_t *rsc, const char *requested_name, const char *attr_set, const char *attr_set_type, const char *attr_id, const char *attr_name, const char *attr_value, gboolean recursive, cib_t *cib, xmlNode *cib_xml_orig, gboolean force, GList **results) { pcmk__output_t *out = rsc->priv->scheduler->priv->out; int rc = pcmk_rc_ok; GList/**/ *resources = NULL; const char *top_id = pe__const_top_resource(rsc, false)->id; if ((attr_id == NULL) && !force) { find_resource_attr(out, cib, PCMK_XA_ID, top_id, NULL, NULL, NULL, attr_name, NULL); } rc = resources_with_attr(out, cib, rsc, requested_name, attr_set, attr_set_type, attr_id, attr_name, top_id, force, &resources); if (rc != pcmk_rc_ok) { return rc; } for (GList *iter = resources; iter != NULL; iter = iter->next) { // @TODO Functionize loop body to simplify freeing allocated memory char *lookup_id = NULL; char *local_attr_set = NULL; char *found_attr_id = NULL; const char *rsc_attr_id = attr_id; const char *rsc_attr_set = attr_set; xmlNode *rsc_xml = rsc->priv->xml; xmlNode *xml_top = NULL; xmlNode *xml_obj = NULL; xmlNode *xml_search = NULL; rsc = (pcmk_resource_t *) iter->data; lookup_id = clone_strip(rsc->id); /* Could be a cloned group! */ rc = find_resource_attr(out, cib, PCMK_XA_ID, lookup_id, attr_set_type, attr_set, attr_id, attr_name, &xml_search); switch (rc) { case pcmk_rc_ok: found_attr_id = crm_element_value_copy(xml_search, PCMK_XA_ID); crm_debug("Found a match for " PCMK_XA_NAME "='%s': " PCMK_XA_ID "='%s'", attr_name, found_attr_id); rsc_attr_id = found_attr_id; break; case ENXIO: if (rsc_attr_set == NULL) { local_attr_set = crm_strdup_printf("%s-%s", lookup_id, attr_set_type); rsc_attr_set = local_attr_set; } if (rsc_attr_id == NULL) { found_attr_id = crm_strdup_printf("%s-%s", rsc_attr_set, attr_name); rsc_attr_id = found_attr_id; } rsc_xml = get_cib_rsc(cib_xml_orig, rsc); if (rsc_xml == NULL) { /* @TODO Warn and continue through the rest of the resources * and return the error at the end? This should never * happen, but if it does, then we could have a partial * update. */ free(lookup_id); free(found_attr_id); pcmk__xml_free(xml_search); g_list_free(resources); return ENXIO; } xml_top = pcmk__xe_create(NULL, (const char *) rsc_xml->name); crm_xml_add(xml_top, PCMK_XA_ID, lookup_id); xml_obj = pcmk__xe_create(xml_top, attr_set_type); crm_xml_add(xml_obj, PCMK_XA_ID, rsc_attr_set); break; default: free(lookup_id); free(found_attr_id); pcmk__xml_free(xml_search); g_list_free(resources); return rc; } xml_obj = crm_create_nvpair_xml(xml_obj, rsc_attr_id, attr_name, attr_value); if (xml_top == NULL) { xml_top = xml_obj; } crm_log_xml_debug(xml_top, "Update"); rc = cib->cmds->modify(cib, PCMK_XE_RESOURCES, xml_top, cib_sync_call); rc = pcmk_legacy2rc(rc); if (rc == pcmk_rc_ok) { attr_update_data_t *ud = pcmk__assert_alloc(1, sizeof(attr_update_data_t)); if (attr_set_type == NULL) { attr_set_type = (const char *) xml_search->parent->name; } if (rsc_attr_set == NULL) { rsc_attr_set = crm_element_value(xml_search->parent, PCMK_XA_ID); } ud->attr_set_type = pcmk__str_copy(attr_set_type); ud->attr_set_id = pcmk__str_copy(rsc_attr_set); ud->attr_name = pcmk__str_copy(attr_name); ud->attr_value = pcmk__str_copy(attr_value); ud->given_rsc_id = pcmk__str_copy(lookup_id); ud->found_attr_id = pcmk__str_copy(found_attr_id); ud->rsc = rsc; *results = g_list_append(*results, ud); } pcmk__xml_free(xml_top); pcmk__xml_free(xml_search); free(lookup_id); free(found_attr_id); free(local_attr_set); if (recursive && pcmk__str_eq(attr_set_type, PCMK_XE_META_ATTRIBUTES, pcmk__str_casei)) { /* We want to set the attribute only on resources explicitly * colocated with this one, so we use * rsc->priv->with_this_colocations directly rather than the * with_this_colocations() method. */ pcmk__set_rsc_flags(rsc, pcmk__rsc_detect_loop); for (GList *lpc = rsc->priv->with_this_colocations; lpc != NULL; lpc = lpc->next) { pcmk__colocation_t *cons = (pcmk__colocation_t *) lpc->data; crm_debug("Checking %s %d", cons->id, cons->score); if (pcmk_is_set(cons->dependent->flags, pcmk__rsc_detect_loop) || (cons->score <= 0)) { continue; } crm_debug("Setting %s=%s for dependent resource %s", attr_name, attr_value, cons->dependent->id); update_attribute(cons->dependent, cons->dependent->id, NULL, attr_set_type, NULL, attr_name, attr_value, recursive, cib, cib_xml_orig, force, results); } } } g_list_free(resources); return rc; } // \return Standard Pacemaker return code int cli_resource_update_attribute(pcmk_resource_t *rsc, const char *requested_name, const char *attr_set, const char *attr_set_type, const char *attr_id, const char *attr_name, const char *attr_value, gboolean recursive, cib_t *cib, xmlNode *cib_xml_orig, gboolean force) { static bool need_init = true; int rc = pcmk_rc_ok; GList *results = NULL; pcmk__output_t *out = rsc->priv->scheduler->priv->out; pcmk__assert(cib_xml_orig != NULL); /* If we were asked to update the attribute in a resource element (for * instance, ) there's really not much we need to do. */ if (pcmk__str_eq(attr_set_type, ATTR_SET_ELEMENT, pcmk__str_none)) { return update_element_attribute(out, rsc, cib, cib_xml_orig, attr_name, attr_value); } /* One time initialization - clear flags so we can detect loops */ if (need_init) { need_init = false; pcmk__unpack_constraints(rsc->priv->scheduler); pe__clear_resource_flags_on_all(rsc->priv->scheduler, pcmk__rsc_detect_loop); } rc = update_attribute(rsc, requested_name, attr_set, attr_set_type, attr_id, attr_name, attr_value, recursive, cib, cib_xml_orig, force, &results); if (rc == pcmk_rc_ok) { if (results == NULL) { return rc; } out->message(out, "attribute-changed-list", results); g_list_free_full(results, free_attr_update_data); } return rc; } // \return Standard Pacemaker return code int cli_resource_delete_attribute(pcmk_resource_t *rsc, const char *requested_name, const char *attr_set, const char *attr_set_type, const char *attr_id, const char *attr_name, cib_t *cib, xmlNode *cib_xml_orig, gboolean force) { pcmk__output_t *out = rsc->priv->scheduler->priv->out; int rc = pcmk_rc_ok; GList/**/ *resources = NULL; pcmk__assert((cib != NULL) && (cib_xml_orig != NULL)); if ((attr_id == NULL) && !force) { find_resource_attr(out, cib, PCMK_XA_ID, pe__const_top_resource(rsc, false)->id, NULL, NULL, NULL, attr_name, NULL); } if (pcmk__str_eq(attr_set_type, ATTR_SET_ELEMENT, pcmk__str_none)) { xmlNode *rsc_xml = rsc->priv->xml; rsc_xml = get_cib_rsc(cib_xml_orig, rsc); if (rsc_xml == NULL) { return ENXIO; } pcmk__xe_remove_attr(rsc_xml, attr_name); rc = cib->cmds->replace(cib, PCMK_XE_RESOURCES, rsc_xml, cib_sync_call); rc = pcmk_legacy2rc(rc); if (rc == pcmk_rc_ok) { out->info(out, "Deleted attribute: %s", attr_name); } return rc; } if (pcmk__str_eq(attr_set_type, PCMK_XE_META_ATTRIBUTES, pcmk__str_none)) { resources = find_matching_attr_resources(out, rsc, requested_name, attr_set, attr_set_type, attr_id, attr_name, cib, "delete", force); } else { resources = g_list_append(resources, rsc); } for (GList *iter = resources; iter != NULL; iter = iter->next) { char *lookup_id = NULL; xmlNode *xml_obj = NULL; xmlNode *xml_search = NULL; char *found_attr_id = NULL; const char *rsc_attr_id = attr_id; rsc = (pcmk_resource_t *) iter->data; /* @TODO Search the original CIB in find_resource_attr() for * future-proofing, to ensure that we're getting IDs of nvpairs that * exist in the CIB. */ lookup_id = clone_strip(rsc->id); rc = find_resource_attr(out, cib, PCMK_XA_ID, lookup_id, attr_set_type, attr_set, attr_id, attr_name, &xml_search); switch (rc) { case pcmk_rc_ok: found_attr_id = crm_element_value_copy(xml_search, PCMK_XA_ID); pcmk__xml_free(xml_search); break; case ENXIO: free(lookup_id); pcmk__xml_free(xml_search); continue; default: free(lookup_id); pcmk__xml_free(xml_search); g_list_free(resources); return rc; } if (rsc_attr_id == NULL) { rsc_attr_id = found_attr_id; } xml_obj = crm_create_nvpair_xml(NULL, rsc_attr_id, attr_name, NULL); crm_log_xml_debug(xml_obj, "Delete"); rc = cib->cmds->remove(cib, PCMK_XE_RESOURCES, xml_obj, cib_sync_call); rc = pcmk_legacy2rc(rc); if (rc == pcmk_rc_ok) { out->info(out, "Deleted '%s' option: " PCMK_XA_ID "=%s%s%s%s%s", lookup_id, found_attr_id, ((attr_set == NULL)? "" : " set="), pcmk__s(attr_set, ""), ((attr_name == NULL)? "" : " " PCMK_XA_NAME "="), pcmk__s(attr_name, "")); } free(lookup_id); pcmk__xml_free(xml_obj); free(found_attr_id); } g_list_free(resources); return rc; } // \return Standard Pacemaker return code static int send_lrm_rsc_op(pcmk_ipc_api_t *controld_api, bool do_fail_resource, const char *host_uname, const char *rsc_id, pcmk_scheduler_t *scheduler) { pcmk__output_t *out = scheduler->priv->out; const char *router_node = host_uname; const char *rsc_api_id = NULL; const char *rsc_long_id = NULL; const char *rsc_class = NULL; const char *rsc_provider = NULL; const char *rsc_type = NULL; bool cib_only = false; pcmk_resource_t *rsc = pe_find_resource(scheduler->priv->resources, rsc_id); if (rsc == NULL) { out->err(out, "Resource %s not found", rsc_id); return ENXIO; } else if (!pcmk__is_primitive(rsc)) { out->err(out, "We can only process primitive resources, not %s", rsc_id); return EINVAL; } rsc_class = crm_element_value(rsc->priv->xml, PCMK_XA_CLASS); rsc_provider = crm_element_value(rsc->priv->xml, PCMK_XA_PROVIDER); rsc_type = crm_element_value(rsc->priv->xml, PCMK_XA_TYPE); if ((rsc_class == NULL) || (rsc_type == NULL)) { out->err(out, "Resource %s does not have a class and type", rsc_id); return EINVAL; } { pcmk_node_t *node = pcmk_find_node(scheduler, host_uname); if (node == NULL) { out->err(out, "Node %s not found", host_uname); return pcmk_rc_node_unknown; } if (!(node->details->online)) { if (do_fail_resource) { out->err(out, "Node %s is not online", host_uname); return ENOTCONN; } else { cib_only = true; } } if (!cib_only && pcmk__is_pacemaker_remote_node(node)) { node = pcmk__current_node(node->priv->remote); if (node == NULL) { out->err(out, "No cluster connection to Pacemaker Remote node %s detected", host_uname); return ENOTCONN; } router_node = node->priv->name; } } if (rsc->priv->history_id != NULL) { rsc_api_id = rsc->priv->history_id; rsc_long_id = rsc->id; } else { rsc_api_id = rsc->id; } if (do_fail_resource) { return pcmk_controld_api_fail(controld_api, host_uname, router_node, rsc_api_id, rsc_long_id, rsc_class, rsc_provider, rsc_type); } else { return pcmk_controld_api_refresh(controld_api, host_uname, router_node, rsc_api_id, rsc_long_id, rsc_class, rsc_provider, rsc_type, cib_only); } } /*! * \internal * \brief Get resource name as used in failure-related node attributes * * \param[in] rsc Resource to check * * \return Newly allocated string containing resource's fail name * \note The caller is responsible for freeing the result. */ static inline char * rsc_fail_name(const pcmk_resource_t *rsc) { const char *name = pcmk__s(rsc->priv->history_id, rsc->id); if (pcmk_is_set(rsc->flags, pcmk__rsc_unique)) { return strdup(name); } return clone_strip(name); } // \return Standard Pacemaker return code static int clear_rsc_history(pcmk_ipc_api_t *controld_api, const char *host_uname, const char *rsc_id, pcmk_scheduler_t *scheduler) { int rc = pcmk_rc_ok; /* Erase the resource's entire LRM history in the CIB, even if we're only * clearing a single operation's fail count. If we erased only entries for a * single operation, we might wind up with a wrong idea of the current * resource state, and we might not re-probe the resource. */ rc = send_lrm_rsc_op(controld_api, false, host_uname, rsc_id, scheduler); if (rc != pcmk_rc_ok) { return rc; } crm_trace("Processing %d mainloop inputs", pcmk_controld_api_replies_expected(controld_api)); while (g_main_context_iteration(NULL, FALSE)) { crm_trace("Processed mainloop input, %d still remaining", pcmk_controld_api_replies_expected(controld_api)); } return rc; } // \return Standard Pacemaker return code static int clear_rsc_failures(pcmk__output_t *out, pcmk_ipc_api_t *controld_api, const char *node_name, const char *rsc_id, const char *operation, const char *interval_spec, pcmk_scheduler_t *scheduler) { int rc = pcmk_rc_ok; const char *failed_value = NULL; const char *failed_id = NULL; char *interval_ms_s = NULL; GHashTable *rscs = NULL; GHashTableIter iter; /* Create a hash table to use as a set of resources to clean. This lets us * clean each resource only once (per node) regardless of how many failed * operations it has. */ rscs = pcmk__strkey_table(NULL, NULL); // Normalize interval to milliseconds for comparison to history entry if (operation) { guint interval_ms = 0U; pcmk_parse_interval_spec(interval_spec, &interval_ms); interval_ms_s = crm_strdup_printf("%u", interval_ms); } for (xmlNode *xml_op = pcmk__xe_first_child(scheduler->priv->failed, NULL, NULL, NULL); xml_op != NULL; xml_op = pcmk__xe_next(xml_op, NULL)) { failed_id = crm_element_value(xml_op, PCMK__XA_RSC_ID); if (failed_id == NULL) { // Malformed history entry, should never happen continue; } // No resource specified means all resources match if (rsc_id) { pcmk_resource_t *fail_rsc = NULL; fail_rsc = pe_find_resource_with_flags(scheduler->priv->resources, failed_id, pcmk_rsc_match_history |pcmk_rsc_match_anon_basename); if ((fail_rsc == NULL) || !pcmk__str_eq(rsc_id, fail_rsc->id, pcmk__str_none)) { continue; } } // Host name should always have been provided by this point failed_value = crm_element_value(xml_op, PCMK_XA_UNAME); if (!pcmk__str_eq(node_name, failed_value, pcmk__str_casei)) { continue; } // No operation specified means all operations match if (operation) { failed_value = crm_element_value(xml_op, PCMK_XA_OPERATION); if (!pcmk__str_eq(operation, failed_value, pcmk__str_casei)) { continue; } // Interval (if operation was specified) defaults to 0 (not all) failed_value = crm_element_value(xml_op, PCMK_META_INTERVAL); if (!pcmk__str_eq(interval_ms_s, failed_value, pcmk__str_casei)) { continue; } } g_hash_table_add(rscs, (gpointer) failed_id); } free(interval_ms_s); g_hash_table_iter_init(&iter, rscs); while (g_hash_table_iter_next(&iter, (gpointer *) &failed_id, NULL)) { crm_debug("Erasing failures of %s on %s", failed_id, node_name); rc = clear_rsc_history(controld_api, node_name, failed_id, scheduler); if (rc != pcmk_rc_ok) { return rc; } } g_hash_table_destroy(rscs); return rc; } // \return Standard Pacemaker return code static int clear_rsc_fail_attrs(const pcmk_resource_t *rsc, const char *operation, const char *interval_spec, const pcmk_node_t *node) { int rc = pcmk_rc_ok; int attr_options = pcmk__node_attr_none; char *rsc_name = rsc_fail_name(rsc); if (pcmk__is_pacemaker_remote_node(node)) { attr_options |= pcmk__node_attr_remote; } rc = pcmk__attrd_api_clear_failures(NULL, node->priv->name, rsc_name, operation, interval_spec, NULL, attr_options); free(rsc_name); return rc; } // \return Standard Pacemaker return code int cli_resource_delete(pcmk_ipc_api_t *controld_api, const char *host_uname, const pcmk_resource_t *rsc, const char *operation, const char *interval_spec, bool just_failures, pcmk_scheduler_t *scheduler, gboolean force) { pcmk__output_t *out = scheduler->priv->out; int rc = pcmk_rc_ok; pcmk_node_t *node = NULL; if (rsc == NULL) { return ENXIO; } else if (rsc->priv->children != NULL) { for (const GList *lpc = rsc->priv->children; lpc != NULL; lpc = lpc->next) { const pcmk_resource_t *child = (const pcmk_resource_t *) lpc->data; rc = cli_resource_delete(controld_api, host_uname, child, operation, interval_spec, just_failures, scheduler, force); if (rc != pcmk_rc_ok) { return rc; } } return pcmk_rc_ok; } else if (host_uname == NULL) { GList *lpc = NULL; GList *nodes = g_hash_table_get_values(rsc->priv->probed_nodes); if(nodes == NULL && force) { nodes = pcmk__copy_node_list(scheduler->nodes, false); } else if ((nodes == NULL) && pcmk_is_set(rsc->flags, pcmk__rsc_exclusive_probes)) { GHashTableIter iter; pcmk_node_t *node = NULL; g_hash_table_iter_init(&iter, rsc->priv->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (void**)&node)) { if (node->assign->score >= 0) { nodes = g_list_prepend(nodes, node); } } } else if(nodes == NULL) { nodes = g_hash_table_get_values(rsc->priv->allowed_nodes); } for (lpc = nodes; lpc != NULL; lpc = lpc->next) { node = (pcmk_node_t *) lpc->data; if (node->details->online) { rc = cli_resource_delete(controld_api, node->priv->name, rsc, operation, interval_spec, just_failures, scheduler, force); } if (rc != pcmk_rc_ok) { g_list_free(nodes); return rc; } } g_list_free(nodes); return pcmk_rc_ok; } node = pcmk_find_node(scheduler, host_uname); if (node == NULL) { out->err(out, "Unable to clean up %s because node %s not found", rsc->id, host_uname); return ENODEV; } if (!pcmk_is_set(node->priv->flags, pcmk__node_probes_allowed)) { out->err(out, "Unable to clean up %s because resource discovery disabled on %s", rsc->id, host_uname); return EOPNOTSUPP; } if (controld_api == NULL) { out->err(out, "Dry run: skipping clean-up of %s on %s due to CIB_file", rsc->id, host_uname); return pcmk_rc_ok; } rc = clear_rsc_fail_attrs(rsc, operation, interval_spec, node); if (rc != pcmk_rc_ok) { out->err(out, "Unable to clean up %s failures on %s: %s", rsc->id, host_uname, pcmk_rc_str(rc)); return rc; } if (just_failures) { rc = clear_rsc_failures(out, controld_api, host_uname, rsc->id, operation, interval_spec, scheduler); } else { rc = clear_rsc_history(controld_api, host_uname, rsc->id, scheduler); } if (rc != pcmk_rc_ok) { out->err(out, "Cleaned %s failures on %s, but unable to clean history: %s", rsc->id, host_uname, pcmk_rc_str(rc)); } else { out->info(out, "Cleaned up %s on %s", rsc->id, host_uname); } return rc; } // \return Standard Pacemaker return code int cli_cleanup_all(pcmk_ipc_api_t *controld_api, const char *node_name, const char *operation, const char *interval_spec, pcmk_scheduler_t *scheduler) { pcmk__output_t *out = scheduler->priv->out; int rc = pcmk_rc_ok; int attr_options = pcmk__node_attr_none; const char *display_name = node_name? node_name : "all nodes"; if (controld_api == NULL) { out->info(out, "Dry run: skipping clean-up of %s due to CIB_file", display_name); return rc; } if (node_name) { pcmk_node_t *node = pcmk_find_node(scheduler, node_name); if (node == NULL) { out->err(out, "Unknown node: %s", node_name); return ENXIO; } if (pcmk__is_pacemaker_remote_node(node)) { attr_options |= pcmk__node_attr_remote; } } rc = pcmk__attrd_api_clear_failures(NULL, node_name, NULL, operation, interval_spec, NULL, attr_options); if (rc != pcmk_rc_ok) { out->err(out, "Unable to clean up all failures on %s: %s", display_name, pcmk_rc_str(rc)); return rc; } if (node_name) { rc = clear_rsc_failures(out, controld_api, node_name, NULL, operation, interval_spec, scheduler); if (rc != pcmk_rc_ok) { out->err(out, "Cleaned all resource failures on %s, but unable to clean history: %s", node_name, pcmk_rc_str(rc)); return rc; } } else { for (GList *iter = scheduler->nodes; iter; iter = iter->next) { pcmk_node_t *node = (pcmk_node_t *) iter->data; rc = clear_rsc_failures(out, controld_api, node->priv->name, NULL, operation, interval_spec, scheduler); if (rc != pcmk_rc_ok) { out->err(out, "Cleaned all resource failures on all nodes, but unable to clean history: %s", pcmk_rc_str(rc)); return rc; } } } out->info(out, "Cleaned up all resources on %s", display_name); return rc; } static void check_role(resource_checks_t *checks) { const char *role_s = g_hash_table_lookup(checks->rsc->priv->meta, PCMK_META_TARGET_ROLE); if (role_s == NULL) { return; } switch (pcmk_parse_role(role_s)) { case pcmk_role_stopped: checks->flags |= rsc_remain_stopped; break; case pcmk_role_unpromoted: if (pcmk_is_set(pe__const_top_resource(checks->rsc, false)->flags, pcmk__rsc_promotable)) { checks->flags |= rsc_unpromotable; } break; default: break; } } static void check_managed(resource_checks_t *checks) { const char *managed_s = g_hash_table_lookup(checks->rsc->priv->meta, PCMK_META_IS_MANAGED); if ((managed_s != NULL) && !crm_is_true(managed_s)) { checks->flags |= rsc_unmanaged; } } static void check_locked(resource_checks_t *checks) { const pcmk_node_t *lock_node = checks->rsc->priv->lock_node; if (lock_node != NULL) { checks->flags |= rsc_locked; checks->lock_node = lock_node->priv->name; } } static bool node_is_unhealthy(pcmk_node_t *node) { switch (pe__health_strategy(node->priv->scheduler)) { case pcmk__health_strategy_none: break; case pcmk__health_strategy_no_red: if (pe__node_health(node) < 0) { return true; } break; case pcmk__health_strategy_only_green: if (pe__node_health(node) <= 0) { return true; } break; case pcmk__health_strategy_progressive: case pcmk__health_strategy_custom: /* @TODO These are finite scores, possibly with rules, and possibly * combining with other scores, so attributing these as a cause is * nontrivial. */ break; } return false; } static void check_node_health(resource_checks_t *checks, pcmk_node_t *node) { if (node == NULL) { GHashTableIter iter; bool allowed = false; bool all_nodes_unhealthy = true; g_hash_table_iter_init(&iter, checks->rsc->priv->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (void **) &node)) { allowed = true; if (!node_is_unhealthy(node)) { all_nodes_unhealthy = false; break; } } if (allowed && all_nodes_unhealthy) { checks->flags |= rsc_node_health; } } else if (node_is_unhealthy(node)) { checks->flags |= rsc_node_health; } } int cli_resource_check(pcmk__output_t *out, pcmk_resource_t *rsc, pcmk_node_t *node) { resource_checks_t checks = { .rsc = rsc }; check_role(&checks); check_managed(&checks); check_locked(&checks); check_node_health(&checks, node); return out->message(out, "resource-check-list", &checks); } // \return Standard Pacemaker return code int cli_resource_fail(pcmk_ipc_api_t *controld_api, const char *host_uname, const char *rsc_id, pcmk_scheduler_t *scheduler) { crm_notice("Failing %s on %s", rsc_id, host_uname); return send_lrm_rsc_op(controld_api, true, host_uname, rsc_id, scheduler); } static GHashTable * generate_resource_params(pcmk_resource_t *rsc, pcmk_node_t *node, pcmk_scheduler_t *scheduler) { GHashTable *params = NULL; GHashTable *meta = NULL; GHashTable *combined = NULL; GHashTableIter iter; char *key = NULL; char *value = NULL; combined = pcmk__strkey_table(free, free); params = pe_rsc_params(rsc, node, scheduler); if (params != NULL) { g_hash_table_iter_init(&iter, params); while (g_hash_table_iter_next(&iter, (gpointer *) & key, (gpointer *) & value)) { pcmk__insert_dup(combined, key, value); } } meta = pcmk__strkey_table(free, free); get_meta_attributes(meta, rsc, NULL, scheduler); if (meta != NULL) { g_hash_table_iter_init(&iter, meta); while (g_hash_table_iter_next(&iter, (gpointer *) & key, (gpointer *) & value)) { char *crm_name = crm_meta_name(key); g_hash_table_insert(combined, crm_name, strdup(value)); } g_hash_table_destroy(meta); } return combined; } bool resource_is_running_on(pcmk_resource_t *rsc, const char *host) { bool found = true; GList *hIter = NULL; GList *hosts = NULL; if (rsc == NULL) { return false; } rsc->priv->fns->location(rsc, &hosts, pcmk__rsc_node_current); for (hIter = hosts; host != NULL && hIter != NULL; hIter = hIter->next) { pcmk_node_t *node = (pcmk_node_t *) hIter->data; if (pcmk__strcase_any_of(host, node->priv->name, node->priv->id, NULL)) { crm_trace("Resource %s is running on %s\n", rsc->id, host); goto done; } } if (host != NULL) { crm_trace("Resource %s is not running on: %s\n", rsc->id, host); found = false; } else if(host == NULL && hosts == NULL) { crm_trace("Resource %s is not running\n", rsc->id); found = false; } done: g_list_free(hosts); return found; } /*! * \internal * \brief Create a list of all resources active on host from a given list * * \param[in] host Name of host to check whether resources are active * \param[in] rsc_list List of resources to check * * \return New list of resources from list that are active on host */ static GList * get_active_resources(const char *host, GList *rsc_list) { GList *rIter = NULL; GList *active = NULL; for (rIter = rsc_list; rIter != NULL; rIter = rIter->next) { pcmk_resource_t *rsc = (pcmk_resource_t *) rIter->data; /* Expand groups to their members, because if we're restarting a member * other than the first, we can't otherwise tell which resources are * stopping and starting. */ if (pcmk__is_group(rsc)) { GList *member_active = NULL; member_active = get_active_resources(host, rsc->priv->children); active = g_list_concat(active, member_active); } else if (resource_is_running_on(rsc, host)) { active = g_list_append(active, strdup(rsc->id)); } } return active; } static void dump_list(GList *items, const char *tag) { int lpc = 0; GList *item = NULL; for (item = items; item != NULL; item = item->next) { crm_trace("%s[%d]: %s", tag, lpc, (char*)item->data); lpc++; } } static void display_list(pcmk__output_t *out, GList *items, const char *tag) { GList *item = NULL; for (item = items; item != NULL; item = item->next) { out->info(out, "%s%s", tag, (const char *)item->data); } } /*! * \internal * \brief Update scheduler XML input based on a CIB query and the current time * * The CIB XML is upgraded to the latest schema version. * * \param[in,out] out Output object * \param[in,out] scheduler Scheduler data to update * \param[in] cib Connection to the CIB manager * \param[out] cib_xml_orig Where to store CIB XML before any schema * upgrades (can be \c NULL) * * \return Standard Pacemaker return code */ int update_scheduler_input(pcmk__output_t *out, pcmk_scheduler_t *scheduler, cib_t *cib, xmlNode **cib_xml_orig) { xmlNode *queried_xml = NULL; xmlNode *updated_xml = NULL; int rc = pcmk_rc_ok; pcmk__assert((out != NULL) && (scheduler != NULL) && (scheduler->input == NULL) && (scheduler->priv->now == NULL) && (cib != NULL) && ((cib_xml_orig == NULL) || (*cib_xml_orig == NULL))); rc = cib->cmds->query(cib, NULL, &queried_xml, cib_sync_call); rc = pcmk_legacy2rc(rc); if (rc != pcmk_rc_ok) { out->err(out, "Could not obtain the current CIB: %s", pcmk_rc_str(rc)); goto done; } if (cib_xml_orig != NULL) { updated_xml = pcmk__xml_copy(NULL, queried_xml); } else { // No need to preserve the pre-upgrade CIB, so don't make a copy updated_xml = queried_xml; queried_xml = NULL; } rc = pcmk__update_configured_schema(&updated_xml, false); if (rc != pcmk_rc_ok) { out->err(out, "Could not upgrade the current CIB XML: %s", pcmk_rc_str(rc)); pcmk__xml_free(updated_xml); goto done; } scheduler->input = updated_xml; scheduler->priv->now = crm_time_new(NULL); done: if ((rc == pcmk_rc_ok) && (cib_xml_orig != NULL)) { *cib_xml_orig = queried_xml; } else { pcmk__xml_free(queried_xml); } return rc; } // \return Standard Pacemaker return code static int update_dataset(cib_t *cib, pcmk_scheduler_t *scheduler, xmlNode **cib_xml_orig, bool simulate) { char *pid = NULL; char *shadow_file = NULL; cib_t *shadow_cib = NULL; int rc = pcmk_rc_ok; pcmk__output_t *out = scheduler->priv->out; - pe_reset_working_set(scheduler); + pcmk_reset_scheduler(scheduler); pcmk__set_scheduler_flags(scheduler, pcmk__sched_no_counts); if(simulate) { bool prev_quiet = false; rc = update_scheduler_input(out, scheduler, cib, NULL); if (rc != pcmk_rc_ok) { goto done; } pid = pcmk__getpid_s(); shadow_cib = cib_shadow_new(pid); shadow_file = get_shadow_file(pid); if (shadow_cib == NULL) { out->err(out, "Could not create shadow cib: '%s'", pid); rc = ENXIO; goto done; } rc = pcmk__xml_write_file(scheduler->input, shadow_file, false); if (rc != pcmk_rc_ok) { out->err(out, "Could not populate shadow cib: %s", pcmk_rc_str(rc)); goto done; } rc = shadow_cib->cmds->signon(shadow_cib, crm_system_name, cib_command); rc = pcmk_legacy2rc(rc); if (rc != pcmk_rc_ok) { out->err(out, "Could not connect to shadow cib: %s", pcmk_rc_str(rc)); goto done; } pcmk__schedule_actions(scheduler->input, pcmk__sched_no_counts, scheduler); prev_quiet = out->is_quiet(out); out->quiet = true; pcmk__simulate_transition(scheduler, shadow_cib, NULL); out->quiet = prev_quiet; rc = update_dataset(shadow_cib, scheduler, cib_xml_orig, false); } else { xmlNode *xml = NULL; rc = update_scheduler_input(out, scheduler, cib, &xml); if (rc != pcmk_rc_ok) { goto done; } pcmk__xml_free(*cib_xml_orig); *cib_xml_orig = xml; cluster_status(scheduler); } done: // Do not free scheduler->input because rsc->priv->xml must remain valid cib_delete(shadow_cib); free(pid); if(shadow_file) { unlink(shadow_file); free(shadow_file); } return rc; } /*! * \internal * \brief Find the maximum stop timeout of a resource and its children (if any) * * \param[in,out] rsc Resource to get timeout for * * \return Maximum stop timeout for \p rsc (in milliseconds) */ static guint max_rsc_stop_timeout(pcmk_resource_t *rsc) { long long result_ll; guint max_delay = 0; xmlNode *config = NULL; GHashTable *meta = NULL; if (rsc == NULL) { return 0; } // If resource is collective, use maximum of its children's stop timeouts if (rsc->priv->children != NULL) { for (GList *iter = rsc->priv->children; iter != NULL; iter = iter->next) { pcmk_resource_t *child = iter->data; guint delay = max_rsc_stop_timeout(child); if (delay > max_delay) { pcmk__rsc_trace(rsc, "Maximum stop timeout for %s is now %s " "due to %s", rsc->id, pcmk__readable_interval(delay), child->id); max_delay = delay; } } return max_delay; } // Get resource's stop action configuration from CIB config = pcmk__find_action_config(rsc, PCMK_ACTION_STOP, 0, true); /* Get configured timeout for stop action (fully evaluated for rules, * defaults, etc.). * * @TODO This currently ignores node (which might matter for rules) */ meta = pcmk__unpack_action_meta(rsc, NULL, PCMK_ACTION_STOP, 0, config); if ((pcmk__scan_ll(g_hash_table_lookup(meta, PCMK_META_TIMEOUT), &result_ll, -1LL) == pcmk_rc_ok) && (result_ll >= 0)) { max_delay = (guint) QB_MIN(result_ll, UINT_MAX); } g_hash_table_destroy(meta); return max_delay; } /*! * \internal * \brief Find a reasonable waiting time for stopping any one resource in a list * * \param[in,out] scheduler Scheduler data * \param[in] resources List of names of resources that will be stopped * * \return Rough estimate of a reasonable time to wait (in seconds) to stop any * one resource in \p resources * \note This estimate is very rough, simply the maximum stop timeout of all * given resources and their children, plus a small fudge factor. It does * not account for children that must be stopped in sequence, action * throttling, or any demotions needed. It checks the stop timeout, even * if the resources in question are actually being started. */ static guint wait_time_estimate(pcmk_scheduler_t *scheduler, const GList *resources) { guint max_delay = 0U; // Find maximum stop timeout in milliseconds for (const GList *item = resources; item != NULL; item = item->next) { pcmk_resource_t *rsc = pe_find_resource(scheduler->priv->resources, (const char *) item->data); guint delay = max_rsc_stop_timeout(rsc); if (delay > max_delay) { pcmk__rsc_trace(rsc, "Wait time is now %s due to %s", pcmk__readable_interval(delay), rsc->id); max_delay = delay; } } return pcmk__timeout_ms2s(max_delay) + 5; } #define waiting_for_starts(d, r, h) ((d != NULL) || \ (!resource_is_running_on((r), (h)))) /*! * \internal * \brief Restart a resource (on a particular host if requested). * * \param[in,out] out Output object * \param[in,out] rsc The resource to restart * \param[in] node Node to restart resource on (NULL for all) * \param[in] move_lifetime If not NULL, how long constraint should * remain in effect (as ISO 8601 string) * \param[in] timeout_ms Consider failed if actions do not complete * in this time (specified in milliseconds, * but a two-second granularity is actually * used; if 0, it will be calculated based on * the resource timeout) * \param[in,out] cib Connection to the CIB manager * \param[in] promoted_role_only If true, limit to promoted instances * \param[in] force If true, apply only to requested instance * if part of a collective resource * * \return Standard Pacemaker return code (exits on certain failures) */ int cli_resource_restart(pcmk__output_t *out, pcmk_resource_t *rsc, const pcmk_node_t *node, const char *move_lifetime, guint timeout_ms, cib_t *cib, gboolean promoted_role_only, gboolean force) { int rc = pcmk_rc_ok; int lpc = 0; int before = 0; guint step_timeout_s = 0; guint sleep_interval = 2U; guint timeout = pcmk__timeout_ms2s(timeout_ms); bool stop_via_ban = false; char *rsc_id = NULL; char *lookup_id = NULL; char *orig_target_role = NULL; xmlNode *cib_xml_orig = NULL; GList *list_delta = NULL; GList *target_active = NULL; GList *current_active = NULL; GList *restart_target_active = NULL; pcmk_scheduler_t *scheduler = NULL; pcmk_resource_t *parent = uber_parent(rsc); bool running = false; const char *id = pcmk__s(rsc->priv->history_id, rsc->id); const char *host = node ? node->priv->name : NULL; /* If the implicit resource or primitive resource of a bundle is given, operate on the * bundle itself instead. */ if (pcmk__is_bundled(rsc)) { rsc = parent->priv->parent; } running = resource_is_running_on(rsc, host); if (pcmk__is_clone(parent) && !running) { if (pcmk__is_unique_clone(parent)) { lookup_id = strdup(rsc->id); } else { lookup_id = clone_strip(rsc->id); } rsc = parent->priv->fns->find_rsc(parent, lookup_id, node, pcmk_rsc_match_basename |pcmk_rsc_match_current_node); free(lookup_id); running = resource_is_running_on(rsc, host); } if (!running) { if (host) { out->err(out, "%s is not running on %s and so cannot be restarted", id, host); } else { out->err(out, "%s is not running anywhere and so cannot be restarted", id); } return ENXIO; } if (!pcmk_is_set(rsc->flags, pcmk__rsc_managed)) { out->err(out, "Unmanaged resources cannot be restarted."); return EAGAIN; } rsc_id = strdup(rsc->id); if (pcmk__is_unique_clone(parent)) { lookup_id = strdup(rsc->id); } else { lookup_id = clone_strip(rsc->id); } if (host) { if (pcmk__is_clone(rsc) || pe_bundle_replicas(rsc)) { stop_via_ban = true; } else if (pcmk__is_clone(parent)) { stop_via_ban = true; free(lookup_id); lookup_id = strdup(parent->id); } } /* grab full cib determine originally active resources disable or ban poll cib and watch for affected resources to get stopped without --timeout, calculate the stop timeout for each step and wait for that if we hit --timeout or the service timeout, re-enable or un-ban, report failure and indicate which resources we couldn't take down if everything stopped, re-enable or un-ban poll cib and watch for affected resources to get started without --timeout, calculate the start timeout for each step and wait for that if we hit --timeout or the service timeout, report (different) failure and indicate which resources we couldn't bring back up report success Optimizations: - use constraints to determine ordered list of affected resources - Allow a --no-deps option (aka. --force-restart) */ - scheduler = pe_new_working_set(); + scheduler = pcmk_new_scheduler(); if (scheduler == NULL) { rc = errno; out->err(out, "Could not allocate scheduler data: %s", pcmk_rc_str(rc)); goto done; } scheduler->priv->out = out; rc = update_dataset(cib, scheduler, &cib_xml_orig, false); if(rc != pcmk_rc_ok) { out->err(out, "Could not get new resource list: %s (%d)", pcmk_rc_str(rc), rc); goto done; } restart_target_active = get_active_resources(host, scheduler->priv->resources); current_active = get_active_resources(host, scheduler->priv->resources); dump_list(current_active, "Origin"); if (stop_via_ban) { /* Stop the clone or bundle instance by banning it from the host */ out->quiet = true; rc = cli_resource_ban(out, lookup_id, host, move_lifetime, cib, promoted_role_only, PCMK_ROLE_PROMOTED); } else { xmlNode *xml_search = NULL; /* Stop the resource by setting PCMK_META_TARGET_ROLE to Stopped. * Remember any existing PCMK_META_TARGET_ROLE so we can restore it * later (though it only makes any difference if it's Unpromoted). */ rc = find_resource_attr(out, cib, PCMK_XA_VALUE, lookup_id, NULL, NULL, NULL, PCMK_META_TARGET_ROLE, &xml_search); if (rc == pcmk_rc_ok) { orig_target_role = crm_element_value_copy(xml_search, PCMK_XA_VALUE); } pcmk__xml_free(xml_search); rc = cli_resource_update_attribute(rsc, rsc_id, NULL, PCMK_XE_META_ATTRIBUTES, NULL, PCMK_META_TARGET_ROLE, PCMK_ACTION_STOPPED, FALSE, cib, cib_xml_orig, force); } if(rc != pcmk_rc_ok) { out->err(out, "Could not set " PCMK_META_TARGET_ROLE " for %s: %s (%d)", rsc_id, pcmk_rc_str(rc), rc); if (current_active != NULL) { g_list_free_full(current_active, free); current_active = NULL; } if (restart_target_active != NULL) { g_list_free_full(restart_target_active, free); restart_target_active = NULL; } goto done; } rc = update_dataset(cib, scheduler, &cib_xml_orig, true); if(rc != pcmk_rc_ok) { out->err(out, "Could not determine which resources would be stopped"); goto failure; } target_active = get_active_resources(host, scheduler->priv->resources); dump_list(target_active, "Target"); list_delta = pcmk__subtract_lists(current_active, target_active, (GCompareFunc) strcmp); out->info(out, "Waiting for %d resources to stop:", g_list_length(list_delta)); display_list(out, list_delta, " * "); step_timeout_s = timeout / sleep_interval; while (list_delta != NULL) { before = g_list_length(list_delta); if(timeout_ms == 0) { step_timeout_s = wait_time_estimate(scheduler, list_delta) / sleep_interval; } /* We probably don't need the entire step timeout */ for(lpc = 0; (lpc < step_timeout_s) && (list_delta != NULL); lpc++) { sleep(sleep_interval); if(timeout) { timeout -= sleep_interval; crm_trace("%us remaining", timeout); } rc = update_dataset(cib, scheduler, &cib_xml_orig, false); if(rc != pcmk_rc_ok) { out->err(out, "Could not determine which resources were stopped"); goto failure; } if (current_active != NULL) { g_list_free_full(current_active, free); } current_active = get_active_resources(host, scheduler->priv->resources); g_list_free(list_delta); list_delta = pcmk__subtract_lists(current_active, target_active, (GCompareFunc) strcmp); dump_list(current_active, "Current"); dump_list(list_delta, "Delta"); } crm_trace("%d (was %d) resources remaining", g_list_length(list_delta), before); if(before == g_list_length(list_delta)) { /* aborted during stop phase, print the contents of list_delta */ out->err(out, "Could not complete shutdown of %s, %d resources remaining", rsc_id, g_list_length(list_delta)); display_list(out, list_delta, " * "); rc = ETIME; goto failure; } } if (stop_via_ban) { rc = cli_resource_clear(lookup_id, host, NULL, cib, true, force); } else if (orig_target_role) { rc = cli_resource_update_attribute(rsc, rsc_id, NULL, PCMK_XE_META_ATTRIBUTES, NULL, PCMK_META_TARGET_ROLE, orig_target_role, FALSE, cib, cib_xml_orig, force); free(orig_target_role); orig_target_role = NULL; } else { rc = cli_resource_delete_attribute(rsc, rsc_id, NULL, PCMK_XE_META_ATTRIBUTES, NULL, PCMK_META_TARGET_ROLE, cib, cib_xml_orig, force); } if(rc != pcmk_rc_ok) { out->err(out, "Could not unset " PCMK_META_TARGET_ROLE " for %s: %s (%d)", rsc_id, pcmk_rc_str(rc), rc); goto done; } if (target_active != NULL) { g_list_free_full(target_active, free); } target_active = restart_target_active; list_delta = pcmk__subtract_lists(target_active, current_active, (GCompareFunc) strcmp); out->info(out, "Waiting for %d resources to start again:", g_list_length(list_delta)); display_list(out, list_delta, " * "); step_timeout_s = timeout / sleep_interval; while (waiting_for_starts(list_delta, rsc, host)) { before = g_list_length(list_delta); if(timeout_ms == 0) { step_timeout_s = wait_time_estimate(scheduler, list_delta) / sleep_interval; } /* We probably don't need the entire step timeout */ for (lpc = 0; (lpc < step_timeout_s) && waiting_for_starts(list_delta, rsc, host); lpc++) { sleep(sleep_interval); if(timeout) { timeout -= sleep_interval; crm_trace("%ds remaining", timeout); } rc = update_dataset(cib, scheduler, &cib_xml_orig, false); if(rc != pcmk_rc_ok) { out->err(out, "Could not determine which resources were started"); goto failure; } /* It's OK if dependent resources moved to a different node, * so we check active resources on all nodes. */ if (current_active != NULL) { g_list_free_full(current_active, free); } current_active = get_active_resources(NULL, scheduler->priv->resources); g_list_free(list_delta); list_delta = pcmk__subtract_lists(target_active, current_active, (GCompareFunc) strcmp); dump_list(current_active, "Current"); dump_list(list_delta, "Delta"); } if(before == g_list_length(list_delta)) { /* aborted during start phase, print the contents of list_delta */ out->err(out, "Could not complete restart of %s, %d resources remaining", rsc_id, g_list_length(list_delta)); display_list(out, list_delta, " * "); rc = ETIME; goto failure; } } rc = pcmk_rc_ok; goto done; failure: if (stop_via_ban) { cli_resource_clear(lookup_id, host, NULL, cib, true, force); } else if (orig_target_role) { cli_resource_update_attribute(rsc, rsc_id, NULL, PCMK_XE_META_ATTRIBUTES, NULL, PCMK_META_TARGET_ROLE, orig_target_role, FALSE, cib, cib_xml_orig, force); free(orig_target_role); } else { cli_resource_delete_attribute(rsc, rsc_id, NULL, PCMK_XE_META_ATTRIBUTES, NULL, PCMK_META_TARGET_ROLE, cib, cib_xml_orig, force); } done: if (list_delta != NULL) { g_list_free(list_delta); } if (current_active != NULL) { g_list_free_full(current_active, free); } if (target_active != NULL && (target_active != restart_target_active)) { g_list_free_full(target_active, free); } if (restart_target_active != NULL) { g_list_free_full(restart_target_active, free); } free(rsc_id); free(lookup_id); - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); return rc; } static inline bool action_is_pending(const pcmk_action_t *action) { if (pcmk_any_flags_set(action->flags, pcmk__action_optional|pcmk__action_pseudo) || !pcmk_is_set(action->flags, pcmk__action_runnable) || pcmk__str_eq(PCMK_ACTION_NOTIFY, action->task, pcmk__str_casei)) { return false; } return true; } /*! * \internal * \brief Check whether any actions in a list are pending * * \param[in] actions List of actions to check * * \return true if any actions in the list are pending, otherwise false */ static bool actions_are_pending(const GList *actions) { for (const GList *action = actions; action != NULL; action = action->next) { const pcmk_action_t *a = (const pcmk_action_t *) action->data; if (action_is_pending(a)) { crm_notice("Waiting for %s (flags=%#.8x)", a->uuid, a->flags); return true; } } return false; } static void print_pending_actions(pcmk__output_t *out, GList *actions) { GList *action; out->info(out, "Pending actions:"); for (action = actions; action != NULL; action = action->next) { pcmk_action_t *a = (pcmk_action_t *) action->data; if (!action_is_pending(a)) { continue; } if (a->node) { out->info(out, "\tAction %d: %s\ton %s", a->id, a->uuid, pcmk__node_name(a->node)); } else { out->info(out, "\tAction %d: %s", a->id, a->uuid); } } } /* For --wait, timeout (in seconds) to use if caller doesn't specify one */ #define WAIT_DEFAULT_TIMEOUT_S (60 * 60) /* For --wait, how long to sleep between cluster state checks */ #define WAIT_SLEEP_S (2) /*! * \internal * \brief Wait until all pending cluster actions are complete * * This waits until either the CIB's transition graph is idle or a timeout is * reached. * * \param[in,out] out Output object * \param[in] timeout_ms Consider failed if actions do not complete in * this time (specified in milliseconds, but * one-second granularity is actually used; if 0, a * default will be used) * \param[in,out] cib Connection to the CIB manager * * \return Standard Pacemaker return code */ int wait_till_stable(pcmk__output_t *out, guint timeout_ms, cib_t * cib) { pcmk_scheduler_t *scheduler = NULL; xmlXPathObjectPtr search; int rc = pcmk_rc_ok; bool pending_unknown_state_resources; time_t expire_time = time(NULL); time_t time_diff; bool printed_version_warning = out->is_quiet(out); // i.e. don't print if quiet char *xpath = NULL; if (timeout_ms == 0) { expire_time += WAIT_DEFAULT_TIMEOUT_S; } else { expire_time += pcmk__timeout_ms2s(timeout_ms + 999); } - scheduler = pe_new_working_set(); + scheduler = pcmk_new_scheduler(); if (scheduler == NULL) { return ENOMEM; } xpath = crm_strdup_printf("/" PCMK_XE_CIB "/" PCMK_XE_STATUS "/" PCMK__XE_NODE_STATE "/" PCMK__XE_LRM "/" PCMK__XE_LRM_RESOURCES "/" PCMK__XE_LRM_RESOURCE "/" PCMK__XE_LRM_RSC_OP "[@" PCMK__XA_RC_CODE "='%d']", PCMK_OCF_UNKNOWN); do { /* Abort if timeout is reached */ time_diff = expire_time - time(NULL); if (time_diff <= 0) { print_pending_actions(out, scheduler->priv->actions); rc = ETIME; break; } crm_info("Waiting up to %lld seconds for cluster actions to complete", (long long) time_diff); if (rc == pcmk_rc_ok) { /* this avoids sleep on first loop iteration */ sleep(WAIT_SLEEP_S); } /* Get latest transition graph */ - pe_reset_working_set(scheduler); + pcmk_reset_scheduler(scheduler); rc = update_scheduler_input(out, scheduler, cib, NULL); if (rc != pcmk_rc_ok) { break; } pcmk__schedule_actions(scheduler->input, pcmk__sched_no_counts, scheduler); if (!printed_version_warning) { /* If the DC has a different version than the local node, the two * could come to different conclusions about what actions need to be * done. Warn the user in this case. * * @TODO A possible long-term solution would be to reimplement the * wait as a new controller operation that would be forwarded to the * DC. However, that would have potential problems of its own. */ const char *dc_version = NULL; dc_version = g_hash_table_lookup(scheduler->priv->options, PCMK_OPT_DC_VERSION); if (!pcmk__str_eq(dc_version, PACEMAKER_VERSION "-" BUILD_VERSION, pcmk__str_casei)) { out->info(out, "warning: wait option may not work properly in " "mixed-version cluster"); printed_version_warning = true; } } search = xpath_search(scheduler->input, xpath); pending_unknown_state_resources = (numXpathResults(search) > 0); freeXpathObject(search); } while (actions_are_pending(scheduler->priv->actions) || pending_unknown_state_resources); - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); free(xpath); return rc; } static const char * get_action(const char *rsc_action) { const char *action = NULL; if (pcmk__str_eq(rsc_action, "validate", pcmk__str_casei)) { action = PCMK_ACTION_VALIDATE_ALL; } else if (pcmk__str_eq(rsc_action, "force-check", pcmk__str_casei)) { action = PCMK_ACTION_MONITOR; } else if (pcmk__strcase_any_of(rsc_action, "force-start", "force-stop", "force-demote", "force-promote", NULL)) { action = rsc_action+6; } else { action = rsc_action; } return action; } /*! * \brief Set up environment variables as expected by resource agents * * When the cluster executes resource agents, it adds certain environment * variables (directly or via resource meta-attributes) expected by some * resource agents. Add the essential ones that many resource agents expect, so * the behavior is the same for command-line execution. * * \param[in,out] params Resource parameters that will be passed to agent * \param[in] timeout_ms Action timeout (in milliseconds) * \param[in] check_level OCF check level * \param[in] verbosity Verbosity level */ static void set_agent_environment(GHashTable *params, guint timeout_ms, int check_level, int verbosity) { g_hash_table_insert(params, crm_meta_name(PCMK_META_TIMEOUT), crm_strdup_printf("%u", timeout_ms)); pcmk__insert_dup(params, PCMK_XA_CRM_FEATURE_SET, CRM_FEATURE_SET); if (check_level >= 0) { char *level = crm_strdup_printf("%d", check_level); setenv("OCF_CHECK_LEVEL", level, 1); free(level); } pcmk__set_env_option(PCMK__ENV_DEBUG, ((verbosity > 0)? "1" : "0"), true); if (verbosity > 1) { setenv("OCF_TRACE_RA", "1", 1); } /* A resource agent using the standard ocf-shellfuncs library will not print * messages to stderr if it doesn't have a controlling terminal (e.g. if * crm_resource is called via script or ssh). This forces it to do so. */ setenv("OCF_TRACE_FILE", "/dev/stderr", 0); } /*! * \internal * \brief Apply command-line overrides to resource parameters * * \param[in,out] params Parameters to be passed to agent * \param[in] overrides Parameters to override (or NULL if none) */ static void apply_overrides(GHashTable *params, GHashTable *overrides) { if (overrides != NULL) { GHashTableIter iter; char *name = NULL; char *value = NULL; g_hash_table_iter_init(&iter, overrides); while (g_hash_table_iter_next(&iter, (gpointer *) &name, (gpointer *) &value)) { pcmk__insert_dup(params, name, value); } } } crm_exit_t cli_resource_execute_from_params(pcmk__output_t *out, const char *rsc_name, const char *rsc_class, const char *rsc_prov, const char *rsc_type, const char *rsc_action, GHashTable *params, GHashTable *override_hash, guint timeout_ms, int resource_verbose, gboolean force, int check_level) { const char *class = rsc_class; const char *action = get_action(rsc_action); crm_exit_t exit_code = CRM_EX_OK; svc_action_t *op = NULL; // If no timeout was provided, use the same default as the cluster if (timeout_ms == 0U) { timeout_ms = PCMK_DEFAULT_ACTION_TIMEOUT_MS; } set_agent_environment(params, timeout_ms, check_level, resource_verbose); apply_overrides(params, override_hash); op = services__create_resource_action(rsc_name? rsc_name : "test", rsc_class, rsc_prov, rsc_type, action, 0, QB_MIN(timeout_ms, INT_MAX), params, 0); if (op == NULL) { out->err(out, "Could not execute %s using %s%s%s:%s: %s", action, rsc_class, (rsc_prov? ":" : ""), (rsc_prov? rsc_prov : ""), rsc_type, strerror(ENOMEM)); g_hash_table_destroy(params); return CRM_EX_OSERR; } #if PCMK__ENABLE_SERVICE if (pcmk__str_eq(rsc_class, PCMK_RESOURCE_CLASS_SERVICE, pcmk__str_casei)) { class = resources_find_service_class(rsc_type); } #endif if (!pcmk_is_set(pcmk_get_ra_caps(class), pcmk_ra_cap_cli_exec)) { services__format_result(op, CRM_EX_UNIMPLEMENT_FEATURE, PCMK_EXEC_ERROR, "Manual execution of the %s standard is " "unsupported", pcmk__s(class, "unspecified")); } if (op->rc != PCMK_OCF_UNKNOWN) { exit_code = op->rc; goto done; } services_action_sync(op); // Map results to OCF codes for consistent reporting to user { enum ocf_exitcode ocf_code = services_result2ocf(class, action, op->rc); // Cast variable instead of function return to keep compilers happy exit_code = (crm_exit_t) ocf_code; } done: out->message(out, "resource-agent-action", resource_verbose, rsc_class, rsc_prov, rsc_type, rsc_name, rsc_action, override_hash, exit_code, op->status, services__exit_reason(op), op->stdout_data, op->stderr_data); services_action_free(op); return exit_code; } /*! * \internal * \brief Get the timeout the cluster would use for an action * * \param[in] rsc Resource that action is for * \param[in] action Name of action */ static guint get_action_timeout(pcmk_resource_t *rsc, const char *action) { long long timeout_ms = -1LL; xmlNode *op = pcmk__find_action_config(rsc, action, 0, true); GHashTable *meta = pcmk__unpack_action_meta(rsc, NULL, action, 0, op); if ((pcmk__scan_ll(g_hash_table_lookup(meta, PCMK_META_TIMEOUT), &timeout_ms, -1LL) != pcmk_rc_ok) || (timeout_ms <= 0LL)) { timeout_ms = PCMK_DEFAULT_ACTION_TIMEOUT_MS; } g_hash_table_destroy(meta); return (guint) QB_MIN(timeout_ms, UINT_MAX); } crm_exit_t cli_resource_execute(pcmk_resource_t *rsc, const char *requested_name, const char *rsc_action, GHashTable *override_hash, guint timeout_ms, cib_t *cib, pcmk_scheduler_t *scheduler, int resource_verbose, gboolean force, int check_level) { pcmk__output_t *out = scheduler->priv->out; crm_exit_t exit_code = CRM_EX_OK; const char *rid = requested_name; const char *rtype = NULL; const char *rprov = NULL; const char *rclass = NULL; GHashTable *params = NULL; if (pcmk__strcase_any_of(rsc_action, "force-start", "force-demote", "force-promote", NULL)) { if (pcmk__is_clone(rsc)) { GList *nodes = cli_resource_search(rsc, requested_name, scheduler); if(nodes != NULL && force == FALSE) { out->err(out, "It is not safe to %s %s here: the cluster claims it is already active", rsc_action, rsc->id); out->err(out, "Try setting " PCMK_META_TARGET_ROLE "=" PCMK_ROLE_STOPPED " first or specifying the force option"); return CRM_EX_UNSAFE; } g_list_free_full(nodes, free); } } if (pcmk__is_clone(rsc)) { /* Grab the first child resource in the hope it's not a group */ rsc = rsc->priv->children->data; } if (pcmk__is_group(rsc)) { out->err(out, "Sorry, the %s option doesn't support group resources", rsc_action); return CRM_EX_UNIMPLEMENT_FEATURE; } else if (pcmk__is_bundled(rsc)) { out->err(out, "Sorry, the %s option doesn't support bundled resources", rsc_action); return CRM_EX_UNIMPLEMENT_FEATURE; } rclass = crm_element_value(rsc->priv->xml, PCMK_XA_CLASS); rprov = crm_element_value(rsc->priv->xml, PCMK_XA_PROVIDER); rtype = crm_element_value(rsc->priv->xml, PCMK_XA_TYPE); params = generate_resource_params(rsc, NULL /* @TODO use local node */, scheduler); if (timeout_ms == 0U) { timeout_ms = get_action_timeout(rsc, get_action(rsc_action)); } if (!pcmk__is_anonymous_clone(rsc->priv->parent)) { rid = rsc->id; } exit_code = cli_resource_execute_from_params(out, rid, rclass, rprov, rtype, rsc_action, params, override_hash, timeout_ms, resource_verbose, force, check_level); return exit_code; } // \return Standard Pacemaker return code int cli_resource_move(const pcmk_resource_t *rsc, const char *rsc_id, const char *host_name, const char *move_lifetime, cib_t *cib, pcmk_scheduler_t *scheduler, gboolean promoted_role_only, gboolean force) { pcmk__output_t *out = scheduler->priv->out; int rc = pcmk_rc_ok; unsigned int count = 0; pcmk_node_t *current = NULL; pcmk_node_t *dest = pcmk_find_node(scheduler, host_name); bool cur_is_dest = false; if (dest == NULL) { return pcmk_rc_node_unknown; } if (promoted_role_only && !pcmk_is_set(rsc->flags, pcmk__rsc_promotable)) { const pcmk_resource_t *p = pe__const_top_resource(rsc, false); if (pcmk_is_set(p->flags, pcmk__rsc_promotable)) { out->info(out, "Using parent '%s' for move instead of '%s'.", rsc->id, rsc_id); rsc_id = p->id; rsc = p; } else { out->info(out, "Ignoring --promoted option: %s is not promotable", rsc_id); promoted_role_only = FALSE; } } current = pe__find_active_requires(rsc, &count); if (pcmk_is_set(rsc->flags, pcmk__rsc_promotable)) { unsigned int promoted_count = 0; pcmk_node_t *promoted_node = NULL; for (const GList *iter = rsc->priv->children; iter != NULL; iter = iter->next) { const pcmk_resource_t *child = (const pcmk_resource_t *) iter->data; enum rsc_role_e child_role = child->priv->fns->state(child, TRUE); if (child_role == pcmk_role_promoted) { rsc = child; promoted_node = pcmk__current_node(child); promoted_count++; } } if (promoted_role_only || (promoted_count != 0)) { count = promoted_count; current = promoted_node; } } if (count > 1) { if (pcmk__is_clone(rsc)) { current = NULL; } else { return pcmk_rc_multiple; } } if (pcmk__same_node(current, dest)) { cur_is_dest = true; if (force) { crm_info("%s is already %s on %s, reinforcing placement with location constraint.", rsc_id, promoted_role_only?"promoted":"active", pcmk__node_name(dest)); } else { return pcmk_rc_already; } } /* Clear any previous prefer constraints across all nodes. */ cli_resource_clear(rsc_id, NULL, scheduler->nodes, cib, false, force); /* Clear any previous ban constraints on 'dest'. */ cli_resource_clear(rsc_id, dest->priv->name, scheduler->nodes, cib, true, force); /* Record an explicit preference for 'dest' */ rc = cli_resource_prefer(out, rsc_id, dest->priv->name, move_lifetime, cib, promoted_role_only, PCMK_ROLE_PROMOTED); crm_trace("%s%s now prefers %s%s", rsc->id, (promoted_role_only? " (promoted)" : ""), pcmk__node_name(dest), force?"(forced)":""); /* only ban the previous location if current location != destination location. * it is possible to use -M to enforce a location without regard of where the * resource is currently located */ if (force && !cur_is_dest) { /* Ban the original location if possible */ if(current) { (void)cli_resource_ban(out, rsc_id, current->priv->name, move_lifetime, cib, promoted_role_only, PCMK_ROLE_PROMOTED); } else if(count > 1) { out->info(out, "Resource '%s' is currently %s in %d locations. " "One may now move to %s", rsc_id, (promoted_role_only? "promoted" : "active"), count, pcmk__node_name(dest)); out->info(out, "To prevent '%s' from being %s at a specific location, " "specify a node.", rsc_id, (promoted_role_only? "promoted" : "active")); } else { crm_trace("Not banning %s from its current location: not active", rsc_id); } } return rc; } diff --git a/tools/crm_simulate.c b/tools/crm_simulate.c index df5359cb4f..8dc8b0b670 100644 --- a/tools/crm_simulate.c +++ b/tools/crm_simulate.c @@ -1,585 +1,585 @@ /* * Copyright 2009-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #define SUMMARY "crm_simulate - simulate a Pacemaker cluster's response to events" struct { char *dot_file; char *graph_file; gchar *input_file; pcmk_injections_t *injections; unsigned int flags; gchar *output_file; long long repeat; gboolean store; gchar *test_dir; char *use_date; char *xml_file; } options = { .flags = pcmk_sim_show_pending | pcmk_sim_sanitized, .repeat = 1 }; uint32_t section_opts = 0; char *temp_shadow = NULL; crm_exit_t exit_code = CRM_EX_OK; #define INDENT " " static pcmk__supported_format_t formats[] = { PCMK__SUPPORTED_FORMAT_NONE, PCMK__SUPPORTED_FORMAT_TEXT, PCMK__SUPPORTED_FORMAT_XML, { NULL, NULL, NULL } }; static gboolean all_actions_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { options.flags |= pcmk_sim_all_actions; return TRUE; } static gboolean attrs_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { section_opts |= pcmk_section_attributes; return TRUE; } static gboolean failcounts_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { section_opts |= pcmk_section_failcounts | pcmk_section_failures; return TRUE; } static gboolean in_place_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { options.store = TRUE; options.flags |= pcmk_sim_process | pcmk_sim_simulate; return TRUE; } static gboolean live_check_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { if (options.xml_file) { free(options.xml_file); } options.xml_file = NULL; options.flags &= ~pcmk_sim_sanitized; return TRUE; } static gboolean node_down_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { options.injections->node_down = g_list_append(options.injections->node_down, g_strdup(optarg)); return TRUE; } static gboolean node_fail_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { options.injections->node_fail = g_list_append(options.injections->node_fail, g_strdup(optarg)); return TRUE; } static gboolean node_up_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { pcmk__simulate_node_config = true; options.injections->node_up = g_list_append(options.injections->node_up, g_strdup(optarg)); return TRUE; } static gboolean op_fail_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { options.flags |= pcmk_sim_process | pcmk_sim_simulate; options.injections->op_fail = g_list_append(options.injections->op_fail, g_strdup(optarg)); return TRUE; } static gboolean op_inject_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { options.injections->op_inject = g_list_append(options.injections->op_inject, g_strdup(optarg)); return TRUE; } static gboolean pending_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { options.flags |= pcmk_sim_show_pending; return TRUE; } static gboolean process_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { options.flags |= pcmk_sim_process; return TRUE; } static gboolean quorum_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { pcmk__str_update(&options.injections->quorum, optarg); return TRUE; } static gboolean save_dotfile_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { options.flags |= pcmk_sim_process; pcmk__str_update(&options.dot_file, optarg); return TRUE; } static gboolean save_graph_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { options.flags |= pcmk_sim_process; pcmk__str_update(&options.graph_file, optarg); return TRUE; } static gboolean show_scores_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { options.flags |= pcmk_sim_process | pcmk_sim_show_scores; return TRUE; } static gboolean simulate_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { options.flags |= pcmk_sim_process | pcmk_sim_simulate; return TRUE; } static gboolean ticket_activate_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { options.injections->ticket_activate = g_list_append(options.injections->ticket_activate, g_strdup(optarg)); return TRUE; } static gboolean ticket_grant_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { options.injections->ticket_grant = g_list_append(options.injections->ticket_grant, g_strdup(optarg)); return TRUE; } static gboolean ticket_revoke_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { options.injections->ticket_revoke = g_list_append(options.injections->ticket_revoke, g_strdup(optarg)); return TRUE; } static gboolean ticket_standby_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { options.injections->ticket_standby = g_list_append(options.injections->ticket_standby, g_strdup(optarg)); return TRUE; } static gboolean utilization_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { options.flags |= pcmk_sim_process | pcmk_sim_show_utilization; return TRUE; } static gboolean watchdog_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { pcmk__str_update(&options.injections->watchdog, optarg); return TRUE; } static gboolean xml_file_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { pcmk__str_update(&options.xml_file, optarg); options.flags |= pcmk_sim_sanitized; return TRUE; } static gboolean xml_pipe_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { pcmk__str_update(&options.xml_file, "-"); options.flags |= pcmk_sim_sanitized; return TRUE; } static GOptionEntry operation_entries[] = { { "run", 'R', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, process_cb, "Process the supplied input and show what actions the cluster will take in response", NULL }, { "simulate", 'S', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, simulate_cb, "Like --run, but also simulate taking those actions and show the resulting new status", NULL }, { "in-place", 'X', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, in_place_cb, "Like --simulate, but also store the results back to the input file", NULL }, { "show-attrs", 'A', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, attrs_cb, "Show node attributes", NULL }, { "show-failcounts", 'c', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, failcounts_cb, "Show resource fail counts", NULL }, { "show-scores", 's', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, show_scores_cb, "Show allocation scores", NULL }, { "show-utilization", 'U', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, utilization_cb, "Show utilization information", NULL }, { "profile", 'P', 0, G_OPTION_ARG_FILENAME, &options.test_dir, "Process all the XML files in the named directory to create profiling data", "DIR" }, { "repeat", 'N', 0, G_OPTION_ARG_INT, &options.repeat, "With --profile, repeat each test N times and print timings", "N" }, /* Deprecated */ { "pending", 'j', G_OPTION_FLAG_NO_ARG|G_OPTION_FLAG_HIDDEN, G_OPTION_ARG_CALLBACK, pending_cb, "Display pending state if '" PCMK_META_RECORD_PENDING "' is enabled", NULL }, { NULL } }; static GOptionEntry synthetic_entries[] = { { "node-up", 'u', 0, G_OPTION_ARG_CALLBACK, node_up_cb, "Simulate bringing a node online", "NODE" }, { "node-down", 'd', 0, G_OPTION_ARG_CALLBACK, node_down_cb, "Simulate taking a node offline", "NODE" }, { "node-fail", 'f', 0, G_OPTION_ARG_CALLBACK, node_fail_cb, "Simulate a node failing", "NODE" }, { "op-inject", 'i', 0, G_OPTION_ARG_CALLBACK, op_inject_cb, "Generate a failure for the cluster to react to in the simulation.\n" INDENT "See `Operation Specification` help for more information.", "OPSPEC" }, { "op-fail", 'F', 0, G_OPTION_ARG_CALLBACK, op_fail_cb, "If the specified task occurs during the simulation, have it fail with return code ${rc}.\n" INDENT "The transition will normally stop at the failed action.\n" INDENT "Save the result with --save-output and re-run with --xml-file.\n" INDENT "See `Operation Specification` help for more information.", "OPSPEC" }, { "set-datetime", 't', 0, G_OPTION_ARG_STRING, &options.use_date, "Set date/time (ISO 8601 format, see https://en.wikipedia.org/wiki/ISO_8601)", "DATETIME" }, { "quorum", 'q', 0, G_OPTION_ARG_CALLBACK, quorum_cb, "Set to '1' (or 'true') to indicate cluster has quorum", "QUORUM" }, { "watchdog", 'w', 0, G_OPTION_ARG_CALLBACK, watchdog_cb, "Set to '1' (or 'true') to indicate cluster has an active watchdog device", "DEVICE" }, { "ticket-grant", 'g', 0, G_OPTION_ARG_CALLBACK, ticket_grant_cb, "Simulate granting a ticket", "TICKET" }, { "ticket-revoke", 'r', 0, G_OPTION_ARG_CALLBACK, ticket_revoke_cb, "Simulate revoking a ticket", "TICKET" }, { "ticket-standby", 'b', 0, G_OPTION_ARG_CALLBACK, ticket_standby_cb, "Simulate making a ticket standby", "TICKET" }, { "ticket-activate", 'e', 0, G_OPTION_ARG_CALLBACK, ticket_activate_cb, "Simulate activating a ticket", "TICKET" }, { NULL } }; static GOptionEntry artifact_entries[] = { { "save-input", 'I', 0, G_OPTION_ARG_FILENAME, &options.input_file, "Save the input configuration to the named file", "FILE" }, { "save-output", 'O', 0, G_OPTION_ARG_FILENAME, &options.output_file, "Save the output configuration to the named file", "FILE" }, { "save-graph", 'G', 0, G_OPTION_ARG_CALLBACK, save_graph_cb, "Save the transition graph (XML format) to the named file", "FILE" }, { "save-dotfile", 'D', 0, G_OPTION_ARG_CALLBACK, save_dotfile_cb, "Save the transition graph (DOT format) to the named file", "FILE" }, { "all-actions", 'a', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, all_actions_cb, "Display all possible actions in DOT graph (even if not part of transition)", NULL }, { NULL } }; static GOptionEntry source_entries[] = { { "live-check", 'L', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, live_check_cb, "Connect to CIB manager and use the current CIB contents as input", NULL }, { "xml-file", 'x', 0, G_OPTION_ARG_CALLBACK, xml_file_cb, "Retrieve XML from the named file", "FILE" }, { "xml-pipe", 'p', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, xml_pipe_cb, "Retrieve XML from stdin", NULL }, { NULL } }; static int setup_input(pcmk__output_t *out, const char *input, const char *output, GError **error) { int rc = pcmk_rc_ok; xmlNode *cib_object = NULL; char *local_output = NULL; if (input == NULL) { /* Use live CIB */ rc = cib__signon_query(out, NULL, &cib_object); if (rc != pcmk_rc_ok) { // cib__signon_query() outputs any relevant error return rc; } } else if (pcmk__str_eq(input, "-", pcmk__str_casei)) { cib_object = pcmk__xml_read(NULL); } else { cib_object = pcmk__xml_read(input); } if (cib_object == NULL) { rc = pcmk_rc_bad_input; g_set_error(error, PCMK__EXITC_ERROR, pcmk_rc2exitc(rc), "Could not read input XML: %s", pcmk_rc_str(rc)); return rc; } if (pcmk_find_cib_element(cib_object, PCMK_XE_STATUS) == NULL) { pcmk__xe_create(cib_object, PCMK_XE_STATUS); } rc = pcmk__update_configured_schema(&cib_object, false); if (rc != pcmk_rc_ok) { pcmk__xml_free(cib_object); return rc; } if (!pcmk__validate_xml(cib_object, NULL, NULL, NULL)) { pcmk__xml_free(cib_object); return pcmk_rc_schema_validation; } if (output == NULL) { char *pid = pcmk__getpid_s(); local_output = get_shadow_file(pid); temp_shadow = strdup(local_output); output = local_output; free(pid); } rc = pcmk__xml_write_file(cib_object, output, false); if (rc != pcmk_rc_ok) { g_set_error(error, PCMK__EXITC_ERROR, CRM_EX_CANTCREAT, "Could not create '%s': %s", output, pcmk_rc_str(rc)); } else { setenv("CIB_file", output, 1); } pcmk__xml_free(cib_object); free(local_output); return rc; } static GOptionContext * build_arg_context(pcmk__common_args_t *args, GOptionGroup **group) { GOptionContext *context = NULL; GOptionEntry extra_prog_entries[] = { { "quiet", 'Q', 0, G_OPTION_ARG_NONE, &(args->quiet), "Display only essential output", NULL }, { NULL } }; const char *description = "Operation Specification:\n\n" "The OPSPEC in any command line option is of the form\n" "${resource}_${task}_${interval_in_ms}@${node}=${rc}\n" "(memcached_monitor_20000@bart.example.com=7, for example).\n" "${rc} is an OCF return code. For more information on these\n" "return codes, refer to https://clusterlabs.org/pacemaker/doc/2.1/Pacemaker_Administration/html/agents.html#ocf-return-codes\n\n" "Examples:\n\n" "Pretend a recurring monitor action found memcached stopped on node\n" "fred.example.com and, during recovery, that the memcached stop\n" "action failed:\n\n" "\tcrm_simulate -LS --op-inject memcached:0_monitor_20000@bart.example.com=7 " "--op-fail memcached:0_stop_0@fred.example.com=1 --save-output /tmp/memcached-test.xml\n\n" "Now see what the reaction to the stop failed would be:\n\n" "\tcrm_simulate -S --xml-file /tmp/memcached-test.xml\n\n"; context = pcmk__build_arg_context(args, "text (default), xml", group, NULL); pcmk__add_main_args(context, extra_prog_entries); g_option_context_set_description(context, description); pcmk__add_arg_group(context, "operations", "Operations:", "Show operations options", operation_entries); pcmk__add_arg_group(context, "synthetic", "Synthetic Cluster Events:", "Show synthetic cluster event options", synthetic_entries); pcmk__add_arg_group(context, "artifact", "Artifact Options:", "Show artifact options", artifact_entries); pcmk__add_arg_group(context, "source", "Data Source:", "Show data source options", source_entries); return context; } int main(int argc, char **argv) { int rc = pcmk_rc_ok; pcmk_scheduler_t *scheduler = NULL; pcmk__output_t *out = NULL; GError *error = NULL; GOptionGroup *output_group = NULL; pcmk__common_args_t *args = pcmk__new_common_args(SUMMARY); gchar **processed_args = pcmk__cmdline_preproc(argv, "bdefgiqrtuwxDFGINOP"); GOptionContext *context = build_arg_context(args, &output_group); options.injections = calloc(1, sizeof(pcmk_injections_t)); if (options.injections == NULL) { rc = ENOMEM; goto done; } /* This must come before g_option_context_parse_strv. */ options.xml_file = strdup("-"); pcmk__register_formats(output_group, formats); if (!g_option_context_parse_strv(context, &processed_args, &error)) { exit_code = CRM_EX_USAGE; goto done; } pcmk__cli_init_logging("crm_simulate", args->verbosity); rc = pcmk__output_new(&out, args->output_ty, args->output_dest, argv); if (rc != pcmk_rc_ok) { fprintf(stderr, "Error creating output format %s: %s\n", args->output_ty, pcmk_rc_str(rc)); exit_code = CRM_EX_ERROR; goto done; } if (pcmk__str_eq(args->output_ty, "text", pcmk__str_null_matches) && !(pcmk_is_set(options.flags, pcmk_sim_show_scores) && args->quiet)) { pcmk__output_text_set_fancy(out, true); } pe__register_messages(out); pcmk__register_lib_messages(out); out->quiet = args->quiet; if (args->version) { out->version(out, false); goto done; } if (args->verbosity > 0) { options.flags |= pcmk_sim_verbose; } - scheduler = pe_new_working_set(); + scheduler = pcmk_new_scheduler(); if (scheduler == NULL) { rc = ENOMEM; g_set_error(&error, PCMK__RC_ERROR, rc, "Could not allocate scheduler data"); goto done; } if (pcmk_is_set(options.flags, pcmk_sim_show_scores)) { pcmk__set_scheduler_flags(scheduler, pcmk__sched_output_scores); } if (pcmk_is_set(options.flags, pcmk_sim_show_utilization)) { pcmk__set_scheduler_flags(scheduler, pcmk__sched_show_utilization); } if (options.test_dir != NULL) { scheduler->priv->out = out; pcmk__profile_dir(options.test_dir, options.repeat, scheduler, options.use_date); rc = pcmk_rc_ok; goto done; } rc = setup_input(out, options.xml_file, options.store? options.xml_file : options.output_file, &error); if (rc != pcmk_rc_ok) { goto done; } rc = pcmk__simulate(scheduler, out, options.injections, options.flags, section_opts, options.use_date, options.input_file, options.graph_file, options.dot_file); done: pcmk__output_and_clear_error(&error, NULL); /* There sure is a lot to free in options. */ free(options.dot_file); free(options.graph_file); g_free(options.input_file); g_free(options.output_file); g_free(options.test_dir); free(options.use_date); free(options.xml_file); pcmk_free_injections(options.injections); pcmk__free_arg_context(context); g_strfreev(processed_args); if (scheduler != NULL) { - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); } fflush(stderr); if (temp_shadow) { unlink(temp_shadow); free(temp_shadow); } if (rc != pcmk_rc_ok) { exit_code = pcmk_rc2exitc(rc); } if (out != NULL) { out->finish(out, exit_code, true, NULL); pcmk__output_free(out); } pcmk__unregister_formats(); crm_exit(exit_code); } diff --git a/tools/crm_ticket.c b/tools/crm_ticket.c index 4a2e904344..251ed8449f 100644 --- a/tools/crm_ticket.c +++ b/tools/crm_ticket.c @@ -1,666 +1,666 @@ /* * Copyright 2012-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include GError *error = NULL; #define SUMMARY "Perform tasks related to cluster tickets\n\n" \ "Allows ticket attributes to be queried, modified and deleted." struct { gchar *attr_default; gchar *attr_id; char *attr_name; char *attr_value; gboolean force; char *get_attr_name; gboolean quiet; gchar *set_name; char ticket_cmd; gchar *ticket_id; gchar *xml_file; } options = { .ticket_cmd = 'S' }; GList *attr_delete; GHashTable *attr_set; bool modified = false; int cib_options = cib_sync_call; static pcmk__output_t *out = NULL; #define INDENT " " static pcmk__supported_format_t formats[] = { PCMK__SUPPORTED_FORMAT_NONE, PCMK__SUPPORTED_FORMAT_TEXT, PCMK__SUPPORTED_FORMAT_XML, { NULL, NULL, NULL } }; static gboolean attr_value_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { pcmk__str_update(&options.attr_value, optarg); if (!options.attr_name || !options.attr_value) { return TRUE; } pcmk__insert_dup(attr_set, options.attr_name, options.attr_value); pcmk__str_update(&options.attr_name, NULL); pcmk__str_update(&options.attr_value, NULL); modified = true; return TRUE; } static gboolean command_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { if (pcmk__str_any_of(option_name, "--info", "-l", NULL)) { options.ticket_cmd = 'l'; } else if (pcmk__str_any_of(option_name, "--details", "-L", NULL)) { options.ticket_cmd = 'L'; } else if (pcmk__str_any_of(option_name, "--raw", "-w", NULL)) { options.ticket_cmd = 'w'; } else if (pcmk__str_any_of(option_name, "--query-xml", "-q", NULL)) { options.ticket_cmd = 'q'; } else if (pcmk__str_any_of(option_name, "--constraints", "-c", NULL)) { options.ticket_cmd = 'c'; } else if (pcmk__str_any_of(option_name, "--cleanup", "-C", NULL)) { options.ticket_cmd = 'C'; } return TRUE; } static gboolean delete_attr_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { attr_delete = g_list_append(attr_delete, strdup(optarg)); modified = true; return TRUE; } static gboolean get_attr_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { pcmk__str_update(&options.get_attr_name, optarg); options.ticket_cmd = 'G'; return TRUE; } static gboolean grant_standby_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { if (pcmk__str_any_of(option_name, "--grant", "-g", NULL)) { pcmk__insert_dup(attr_set, PCMK__XA_GRANTED, PCMK_VALUE_TRUE); modified = true; } else if (pcmk__str_any_of(option_name, "--revoke", "-r", NULL)) { pcmk__insert_dup(attr_set, PCMK__XA_GRANTED, PCMK_VALUE_FALSE); modified = true; } else if (pcmk__str_any_of(option_name, "--standby", "-s", NULL)) { pcmk__insert_dup(attr_set, PCMK_XA_STANDBY, PCMK_VALUE_TRUE); modified = true; } else if (pcmk__str_any_of(option_name, "--activate", "-a", NULL)) { pcmk__insert_dup(attr_set, PCMK_XA_STANDBY, PCMK_VALUE_FALSE); modified = true; } return TRUE; } static gboolean set_attr_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **err) { pcmk__str_update(&options.attr_name, optarg); if (!options.attr_name || !options.attr_value) { return TRUE; } pcmk__insert_dup(attr_set, options.attr_name, options.attr_value); pcmk__str_update(&options.attr_name, NULL); pcmk__str_update(&options.attr_value, NULL); modified = true; return TRUE; } static GOptionEntry query_entries[] = { { "info", 'l', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "Display the information of ticket(s)", NULL }, { "details", 'L', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "Display the details of ticket(s)", NULL }, { "raw", 'w', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "Display the IDs of ticket(s)", NULL }, { "query-xml", 'q', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "Query the XML of ticket(s)", NULL }, { "constraints", 'c', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "Display the " PCMK_XE_RSC_TICKET " constraints that apply to ticket(s)", NULL }, { NULL } }; static GOptionEntry command_entries[] = { { "grant", 'g', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, grant_standby_cb, "Grant a ticket to this cluster site", NULL }, { "revoke", 'r', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, grant_standby_cb, "Revoke a ticket from this cluster site", NULL }, { "standby", 's', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, grant_standby_cb, "Tell this cluster site this ticket is standby", NULL }, { "activate", 'a', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, grant_standby_cb, "Tell this cluster site this ticket is active", NULL }, { NULL } }; static GOptionEntry advanced_entries[] = { { "get-attr", 'G', 0, G_OPTION_ARG_CALLBACK, get_attr_cb, "Display the named attribute for a ticket", "ATTRIBUTE" }, { "set-attr", 'S', 0, G_OPTION_ARG_CALLBACK, set_attr_cb, "Set the named attribute for a ticket", "ATTRIBUTE" }, { "delete-attr", 'D', 0, G_OPTION_ARG_CALLBACK, delete_attr_cb, "Delete the named attribute for a ticket", "ATTRIBUTE" }, { "cleanup", 'C', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, command_cb, "Delete all state of a ticket at this cluster site", NULL }, { NULL} }; static GOptionEntry addl_entries[] = { { "attr-value", 'v', 0, G_OPTION_ARG_CALLBACK, attr_value_cb, "Attribute value to use with -S", "VALUE" }, { "default", 'd', 0, G_OPTION_ARG_STRING, &options.attr_default, "(Advanced) Default attribute value to display if none is found\n" INDENT "(for use with -G)", "VALUE" }, { "force", 'f', 0, G_OPTION_ARG_NONE, &options.force, "(Advanced) Force the action to be performed", NULL }, { "ticket", 't', 0, G_OPTION_ARG_STRING, &options.ticket_id, "Ticket ID", "ID" }, { "xml-file", 'x', G_OPTION_FLAG_HIDDEN, G_OPTION_ARG_STRING, &options.xml_file, NULL, NULL }, { NULL } }; static GOptionEntry deprecated_entries[] = { { "set-name", 'n', 0, G_OPTION_ARG_STRING, &options.set_name, "(Advanced) ID of the " PCMK_XE_INSTANCE_ATTRIBUTES " object to change", "ID" }, { "nvpair", 'i', 0, G_OPTION_ARG_STRING, &options.attr_id, "(Advanced) ID of the nvpair object to change/delete", "ID" }, { "quiet", 'Q', 0, G_OPTION_ARG_NONE, &options.quiet, "Print only the value on stdout", NULL }, { NULL } }; static void ticket_grant_warning(gchar *ticket_id) { out->err(out, "This command cannot help you verify whether '%s' has " "been already granted elsewhere.\n" "If you really want to grant '%s' to this site now, and " "you know what you are doing,\n" "please specify --force.", ticket_id, ticket_id); } static void ticket_revoke_warning(gchar *ticket_id) { out->err(out, "Revoking '%s' can trigger the specified '" PCMK_XA_LOSS_POLICY "'(s) relating to '%s'.\n\n" "You can check that with:\n" "crm_ticket --ticket %s --constraints\n\n" "Otherwise before revoking '%s', you may want to make '%s'" "standby with:\n" "crm_ticket --ticket %s --standby\n\n" "If you really want to revoke '%s' from this site now, and " "you know what you are doing,\n" "please specify --force.", ticket_id, ticket_id, ticket_id, ticket_id, ticket_id, ticket_id, ticket_id); } static GOptionContext * build_arg_context(pcmk__common_args_t *args, GOptionGroup **group) { GOptionContext *context = NULL; const char *description = "Examples:\n\n" "Display the info of tickets:\n\n" "\tcrm_ticket --info\n\n" "Display the detailed info of tickets:\n\n" "\tcrm_ticket --details\n\n" "Display the XML of 'ticketA':\n\n" "\tcrm_ticket --ticket ticketA --query-xml\n\n" "Display the " PCMK_XE_RSC_TICKET " constraints that apply to 'ticketA':\n\n" "\tcrm_ticket --ticket ticketA --constraints\n\n" "Grant 'ticketA' to this cluster site:\n\n" "\tcrm_ticket --ticket ticketA --grant\n\n" "Revoke 'ticketA' from this cluster site:\n\n" "\tcrm_ticket --ticket ticketA --revoke\n\n" "Make 'ticketA' standby (the cluster site will treat a granted\n" "'ticketA' as 'standby', and the dependent resources will be\n" "stopped or demoted gracefully without triggering loss-policies):\n\n" "\tcrm_ticket --ticket ticketA --standby\n\n" "Activate 'ticketA' from being standby:\n\n" "\tcrm_ticket --ticket ticketA --activate\n\n" "Get the value of the 'granted' attribute for 'ticketA':\n\n" "\tcrm_ticket --ticket ticketA --get-attr granted\n\n" "Set the value of the 'standby' attribute for 'ticketA':\n\n" "\tcrm_ticket --ticket ticketA --set-attr standby --attr-value true\n\n" "Delete the 'granted' attribute for 'ticketA':\n\n" "\tcrm_ticket --ticket ticketA --delete-attr granted\n\n" "Erase the operation history of 'ticketA' at this cluster site,\n" "causing the cluster site to 'forget' the existing ticket state:\n\n" "\tcrm_ticket --ticket ticketA --cleanup\n\n"; context = pcmk__build_arg_context(args, "text (default), xml", group, NULL); g_option_context_set_description(context, description); pcmk__add_arg_group(context, "queries", "Queries:", "Show queries", query_entries); pcmk__add_arg_group(context, "commands", "Commands:", "Show command options", command_entries); pcmk__add_arg_group(context, "advanced", "Advanced Options:", "Show advanced options", advanced_entries); pcmk__add_arg_group(context, "additional", "Additional Options:", "Show additional options", addl_entries); pcmk__add_arg_group(context, "deprecated", "Deprecated Options:", "Show deprecated options", deprecated_entries); return context; } int main(int argc, char **argv) { pcmk_scheduler_t *scheduler = NULL; xmlNode *cib_xml_copy = NULL; cib_t *cib_conn = NULL; crm_exit_t exit_code = CRM_EX_OK; int rc = pcmk_rc_ok; GOptionGroup *output_group = NULL; pcmk__common_args_t *args = NULL; GOptionContext *context = NULL; gchar **processed_args = NULL; attr_set = pcmk__strkey_table(free, free); attr_delete = NULL; args = pcmk__new_common_args(SUMMARY); context = build_arg_context(args, &output_group); processed_args = pcmk__cmdline_preproc(argv, "dintvxCDGS"); pcmk__register_formats(output_group, formats); if (!g_option_context_parse_strv(context, &processed_args, &error)) { exit_code = CRM_EX_USAGE; goto done; } pcmk__cli_init_logging("crm_ticket", args->verbosity); rc = pcmk__output_new(&out, args->output_ty, args->output_dest, argv); if (rc != pcmk_rc_ok) { exit_code = pcmk_rc2exitc(rc); g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Error creating output format %s: %s", args->output_ty, pcmk_rc_str(rc)); goto done; } pe__register_messages(out); pcmk__register_lib_messages(out); if (args->version) { out->version(out, false); goto done; } - scheduler = pe_new_working_set(); + scheduler = pcmk_new_scheduler(); if (scheduler == NULL) { rc = errno; exit_code = pcmk_rc2exitc(rc); g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Could not allocate scheduler data: %s", pcmk_rc_str(rc)); goto done; } pcmk__set_scheduler_flags(scheduler, pcmk__sched_no_counts); cib_conn = cib_new(); if (cib_conn == NULL) { exit_code = CRM_EX_DISCONNECT; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Could not connect to the CIB manager"); goto done; } rc = cib__signon_attempts(cib_conn, cib_command, 5); rc = pcmk_legacy2rc(rc); if (rc != pcmk_rc_ok) { exit_code = pcmk_rc2exitc(rc); g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Could not connect to the CIB: %s", pcmk_rc_str(rc)); goto done; } if (options.xml_file != NULL) { cib_xml_copy = pcmk__xml_read(options.xml_file); } else { rc = cib_conn->cmds->query(cib_conn, NULL, &cib_xml_copy, cib_sync_call); rc = pcmk_legacy2rc(rc); if (rc != pcmk_rc_ok) { exit_code = pcmk_rc2exitc(rc); g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Could not get local CIB: %s", pcmk_rc_str(rc)); goto done; } } rc = pcmk__update_configured_schema(&cib_xml_copy, false); if (rc != pcmk_rc_ok) { exit_code = pcmk_rc2exitc(rc); g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Could not update local CIB to latest schema version"); goto done; } scheduler->input = cib_xml_copy; scheduler->priv->now = crm_time_new(NULL); cluster_status(scheduler); /* For recording the tickets that are referenced in PCMK_XE_RSC_TICKET * constraints but have never been granted yet. */ pcmk__unpack_constraints(scheduler); if (options.ticket_cmd == 'l' || options.ticket_cmd == 'L' || options.ticket_cmd == 'w') { bool raw = false; bool details = false; if (options.ticket_cmd == 'L') { details = true; } else if (options.ticket_cmd == 'w') { raw = true; } rc = pcmk__ticket_info(out, scheduler, options.ticket_id, details, raw); exit_code = pcmk_rc2exitc(rc); if (rc == ENXIO) { g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "No such ticket '%s'", options.ticket_id); } else if (rc != pcmk_rc_ok) { g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Could not get ticket info: %s", pcmk_rc_str(rc)); } } else if (options.ticket_cmd == 'q') { rc = pcmk__ticket_state(out, cib_conn, options.ticket_id); if (rc != pcmk_rc_ok && rc != pcmk_rc_duplicate_id) { exit_code = pcmk_rc2exitc(rc); g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Could not query ticket XML: %s", pcmk_rc_str(rc)); } else { exit_code = CRM_EX_OK; } } else if (options.ticket_cmd == 'c') { rc = pcmk__ticket_constraints(out, cib_conn, options.ticket_id); exit_code = pcmk_rc2exitc(rc); if (rc != pcmk_rc_ok) { g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Could not show ticket constraints: %s", pcmk_rc_str(rc)); } } else if (options.ticket_cmd == 'G') { if (options.ticket_id == NULL) { exit_code = CRM_EX_NOSUCH; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Must supply ticket ID with -t"); goto done; } rc = pcmk__ticket_get_attr(out, scheduler, options.ticket_id, options.get_attr_name, options.attr_default); exit_code = pcmk_rc2exitc(rc); } else if (options.ticket_cmd == 'C') { if (options.ticket_id == NULL) { exit_code = CRM_EX_USAGE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Must supply ticket ID with -t"); goto done; } rc = pcmk__ticket_delete(out, cib_conn, scheduler, options.ticket_id, options.force); exit_code = pcmk_rc2exitc(rc); switch (rc) { case ENXIO: g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "No such ticket '%s'", options.ticket_id); break; case EACCES: ticket_revoke_warning(options.ticket_id); break; case pcmk_rc_ok: case pcmk_rc_duplicate_id: break; default: g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Could not clean up ticket: %s", pcmk_rc_str(rc)); break; } } else if (modified) { if (options.ticket_id == NULL) { exit_code = CRM_EX_USAGE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Must supply ticket ID with -t"); goto done; } if (options.attr_value && (pcmk__str_empty(options.attr_name))) { exit_code = CRM_EX_USAGE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Must supply attribute name with -S for -v %s", options.attr_value); goto done; } if (options.attr_name && (pcmk__str_empty(options.attr_value))) { exit_code = CRM_EX_USAGE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Must supply attribute value with -v for -S %s", options.attr_value); goto done; } if (attr_delete != NULL) { rc = pcmk__ticket_remove_attr(out, cib_conn, scheduler, options.ticket_id, attr_delete, options.force); if (rc == EACCES) { ticket_revoke_warning(options.ticket_id); exit_code = CRM_EX_UNSAFE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Ticket modification not allowed without --force"); goto done; } } else { rc = pcmk__ticket_set_attr(out, cib_conn, scheduler, options.ticket_id, attr_set, options.force); if (rc == EACCES) { const char *value = NULL; value = g_hash_table_lookup(attr_set, PCMK__XA_GRANTED); if (crm_is_true(value)) { ticket_grant_warning(options.ticket_id); } else { ticket_revoke_warning(options.ticket_id); } exit_code = CRM_EX_UNSAFE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Ticket modification not allowed without --force"); goto done; } } exit_code = pcmk_rc2exitc(rc); if (rc != pcmk_rc_ok && error == NULL) { g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Could not modify ticket: %s", pcmk_rc_str(rc)); } } else if (options.ticket_cmd == 'S') { /* Correct usage was handled in the "if (modified)" block above, so * this is just for reporting usage errors */ if (pcmk__str_empty(options.attr_name)) { // We only get here if ticket_cmd was left as default exit_code = CRM_EX_USAGE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Must supply a command"); goto done; } if (options.ticket_id == NULL) { exit_code = CRM_EX_USAGE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Must supply ticket ID with -t"); goto done; } if (pcmk__str_empty(options.attr_value)) { exit_code = CRM_EX_USAGE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Must supply value with -v for -S %s", options.attr_name); goto done; } } else { exit_code = CRM_EX_USAGE; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Unknown command: %c", options.ticket_cmd); } done: if (attr_set) { g_hash_table_destroy(attr_set); } attr_set = NULL; if (attr_delete) { g_list_free_full(attr_delete, free); } attr_delete = NULL; - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); scheduler = NULL; cib__clean_up_connection(&cib_conn); g_strfreev(processed_args); pcmk__free_arg_context(context); g_free(options.attr_default); g_free(options.attr_id); free(options.attr_name); free(options.attr_value); free(options.get_attr_name); g_free(options.set_name); g_free(options.ticket_id); g_free(options.xml_file); pcmk__output_and_clear_error(&error, out); if (out != NULL) { out->finish(out, exit_code, true, NULL); pcmk__output_free(out); } pcmk__unregister_formats(); crm_exit(exit_code); } diff --git a/tools/crm_verify.c b/tools/crm_verify.c index c491bde365..3dd9763c0a 100644 --- a/tools/crm_verify.c +++ b/tools/crm_verify.c @@ -1,300 +1,299 @@ /* * Copyright 2004-2024 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include const char *SUMMARY = "Check a Pacemaker configuration for errors\n\n" "Check the well-formedness of a complete Pacemaker XML configuration,\n" "its conformance to the configured schema, and the presence of common\n" "misconfigurations. Problems reported as errors must be fixed before the\n" "cluster will work properly. It is left to the administrator to decide\n" "whether to fix problems reported as warnings."; struct { char *cib_save; gboolean use_live_cib; char *xml_file; gboolean xml_stdin; char *xml_string; unsigned int verbosity; } options; static GOptionEntry data_entries[] = { { "live-check", 'L', 0, G_OPTION_ARG_NONE, &options.use_live_cib, "Check the configuration used by the running cluster", NULL }, { "xml-file", 'x', 0, G_OPTION_ARG_FILENAME, &options.xml_file, "Check the configuration in the named file", "FILE" }, { "xml-pipe", 'p', 0, G_OPTION_ARG_NONE, &options.xml_stdin, "Check the configuration piped in via stdin", NULL }, { "xml-text", 'X', 0, G_OPTION_ARG_STRING, &options.xml_string, "Check the configuration in the supplied string", "XML" }, { NULL } }; static GOptionEntry addl_entries[] = { { "save-xml", 'S', G_OPTION_FLAG_NONE, G_OPTION_ARG_FILENAME, &options.cib_save, "Save verified XML to named file (most useful with -L)", "FILE" }, { NULL } }; static pcmk__supported_format_t formats[] = { PCMK__SUPPORTED_FORMAT_NONE, PCMK__SUPPORTED_FORMAT_TEXT, PCMK__SUPPORTED_FORMAT_XML, { NULL, NULL, NULL } }; static GOptionContext * build_arg_context(pcmk__common_args_t *args, GOptionGroup **group) { GOptionContext *context = NULL; const char *description = "Examples:\n\n" "Check the consistency of the configuration in the running cluster:\n\n" "\tcrm_verify --live-check\n\n" "Check the consistency of the configuration in a given file and " "produce quiet output:\n\n" "\tcrm_verify --xml-file file.xml --quiet\n\n" "Check the consistency of the configuration in a given file and " "produce verbose output:\n\n" "\tcrm_verify --xml-file file.xml --verbose\n\n"; GOptionEntry extra_prog_entries[] = { { "quiet", 'q', 0, G_OPTION_ARG_NONE, &(args->quiet), "Don't print verify information", NULL }, { NULL } }; context = pcmk__build_arg_context(args, "text (default), xml", group, NULL); pcmk__add_main_args(context, extra_prog_entries); g_option_context_set_description(context, description); pcmk__add_arg_group(context, "data", "Data sources:", "Show data options", data_entries); pcmk__add_arg_group(context, "additional", "Additional options:", "Show additional options", addl_entries); return context; } /*! * \internal * \brief Output a configuration error * * \param[in] ctx Output object * \param[in] msg printf(3)-style format string * \param[in] ... Format string arguments */ G_GNUC_PRINTF(2, 3) static void output_config_error(void *ctx, const char *msg, ...) { va_list ap; char *buf = NULL; pcmk__output_t *out = ctx; va_start(ap, msg); pcmk__assert(vasprintf(&buf, msg, ap) > 0); if (options.verbosity > 0) { out->err(out, "error: %s", buf); } va_end(ap); } /*! * \internal * \brief Output a configuration warning * * \param[in] ctx Output object * \param[in] msg printf(3)-style format string * \param[in] ... Format string arguments */ G_GNUC_PRINTF(2, 3) static void output_config_warning(void *ctx, const char *msg, ...) { va_list ap; char *buf = NULL; pcmk__output_t *out = ctx; va_start(ap, msg); pcmk__assert(vasprintf(&buf, msg, ap) > 0); if (options.verbosity > 0) { out->err(out, "warning: %s", buf); } va_end(ap); } int main(int argc, char **argv) { pcmk_scheduler_t *scheduler = NULL; int rc = pcmk_rc_ok; crm_exit_t exit_code = CRM_EX_OK; GError *error = NULL; pcmk__output_t *out = NULL; const char *cib_source = NULL; xmlNode *cib_object = NULL; GOptionGroup *output_group = NULL; pcmk__common_args_t *args = pcmk__new_common_args(SUMMARY); gchar **processed_args = pcmk__cmdline_preproc(argv, "xSX"); GOptionContext *context = build_arg_context(args, &output_group); pcmk__register_formats(output_group, formats); if (!g_option_context_parse_strv(context, &processed_args, &error)) { exit_code = CRM_EX_USAGE; goto done; } if (args->verbosity > 0) { args->verbosity -= args->quiet; } pcmk__cli_init_logging("crm_verify", args->verbosity); rc = pcmk__output_new(&out, args->output_ty, args->output_dest, argv); if (rc != pcmk_rc_ok) { exit_code = CRM_EX_ERROR; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Error creating output format %s: %s", args->output_ty, pcmk_rc_str(rc)); goto done; } if (args->version) { out->version(out, false); goto done; } pcmk__register_lib_messages(out); pcmk__set_config_error_handler(output_config_error, out); pcmk__set_config_warning_handler(output_config_warning, out); if (pcmk__str_eq(args->output_ty, "xml", pcmk__str_none)) { args->verbosity = 1; } options.verbosity = args->verbosity; if (options.xml_file != NULL) { cib_source = options.xml_file; } else if (options.xml_string != NULL) { cib_source = options.xml_string; } else if (options.xml_stdin) { cib_source = "-"; } else if (options.use_live_cib) { cib_source = NULL; } else { rc = ENODATA; g_set_error(&error, PCMK__RC_ERROR, rc, "No input specified"); goto done; } rc = pcmk__parse_cib(out, cib_source, &cib_object); if (rc != pcmk_rc_ok) { g_set_error(&error, PCMK__RC_ERROR, rc, "Verification failed: %s", pcmk_rc_str(rc)); goto done; } if (options.cib_save != NULL) { pcmk__xml_write_file(cib_object, options.cib_save, false); } - scheduler = pe_new_working_set(); - + scheduler = pcmk_new_scheduler(); if (scheduler == NULL) { rc = errno; g_set_error(&error, PCMK__RC_ERROR, rc, "Could not allocate scheduler data: %s", pcmk_rc_str(rc)); goto done; } scheduler->priv->out = out; rc = pcmk__verify(scheduler, out, &cib_object); if ((rc == pcmk_rc_schema_validation) && !args->quiet) { const char *failure_type = ""; const char *verbose_hint = ""; if (pcmk__config_has_error) { failure_type = " (with errors)"; } else if (pcmk__config_has_warning) { failure_type = " (with warnings)"; } if (options.verbosity == 0) { verbose_hint = " (-V may provide more detail)"; } out->err(out, "Configuration invalid%s%s", failure_type, verbose_hint); } - pe_free_working_set(scheduler); + pcmk_free_scheduler(scheduler); done: g_strfreev(processed_args); pcmk__free_arg_context(context); free(options.cib_save); free(options.xml_file); free(options.xml_string); if (cib_object != NULL) { pcmk__xml_free(cib_object); } if (exit_code == CRM_EX_OK) { exit_code = pcmk_rc2exitc(rc); } pcmk__output_and_clear_error(&error, out); if (out != NULL) { out->finish(out, exit_code, true, NULL); pcmk__output_free(out); } pcmk__unregister_formats(); crm_exit(exit_code); }