diff --git a/daemons/fenced/pacemaker-fenced.c b/daemons/fenced/pacemaker-fenced.c index d6ff3c32ee..bab27ef154 100644 --- a/daemons/fenced/pacemaker-fenced.c +++ b/daemons/fenced/pacemaker-fenced.c @@ -1,979 +1,981 @@ /* * 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 // PRIu32, PRIx32 #include #include #include #include #include #include #include #include #include #include #include #include #define SUMMARY "daemon for executing fencing devices in a Pacemaker cluster" char *stonith_our_uname = NULL; long long stonith_watchdog_timeout_ms = 0; GList *stonith_watchdog_targets = NULL; static GMainLoop *mainloop = NULL; gboolean stand_alone = FALSE; gboolean stonith_shutdown_flag = FALSE; static qb_ipcs_service_t *ipcs = NULL; static pcmk__output_t *out = NULL; pcmk__supported_format_t formats[] = { PCMK__SUPPORTED_FORMAT_NONE, PCMK__SUPPORTED_FORMAT_TEXT, PCMK__SUPPORTED_FORMAT_XML, { NULL, NULL, NULL } }; static struct { bool no_cib_connect; gchar **log_files; } options; crm_exit_t exit_code = CRM_EX_OK; static void stonith_cleanup(void); static int32_t st_ipc_accept(qb_ipcs_connection_t * c, uid_t uid, gid_t gid) { if (stonith_shutdown_flag) { crm_info("Ignoring new client [%d] during shutdown", pcmk__client_pid(c)); return -ECONNREFUSED; } if (pcmk__new_client(c, uid, gid) == NULL) { return -ENOMEM; } return 0; } /* Exit code means? */ static int32_t st_ipc_dispatch(qb_ipcs_connection_t * qbc, void *data, size_t size) { uint32_t id = 0; uint32_t flags = 0; int call_options = 0; xmlNode *request = NULL; pcmk__client_t *c = pcmk__find_client(qbc); const char *op = NULL; if (c == NULL) { crm_info("Invalid client: %p", qbc); return 0; } request = pcmk__client_data2xml(c, data, &id, &flags); if (request == NULL) { pcmk__ipc_send_ack(c, id, flags, PCMK__XE_NACK, NULL, CRM_EX_PROTOCOL); return 0; } op = crm_element_value(request, PCMK__XA_CRM_TASK); if(pcmk__str_eq(op, CRM_OP_RM_NODE_CACHE, pcmk__str_casei)) { crm_xml_add(request, PCMK__XA_T, PCMK__VALUE_STONITH_NG); crm_xml_add(request, PCMK__XA_ST_OP, op); crm_xml_add(request, PCMK__XA_ST_CLIENTID, c->id); crm_xml_add(request, PCMK__XA_ST_CLIENTNAME, pcmk__client_name(c)); crm_xml_add(request, PCMK__XA_ST_CLIENTNODE, stonith_our_uname); send_cluster_message(NULL, crm_msg_stonith_ng, request, FALSE); free_xml(request); return 0; } if (c->name == NULL) { const char *value = crm_element_value(request, PCMK__XA_ST_CLIENTNAME); c->name = crm_strdup_printf("%s.%u", pcmk__s(value, "unknown"), c->pid); } crm_element_value_int(request, PCMK__XA_ST_CALLOPT, &call_options); crm_trace("Flags %#08" PRIx32 "/%#08x for command %" PRIu32 " from client %s", flags, call_options, id, pcmk__client_name(c)); if (pcmk_is_set(call_options, st_opt_sync_call)) { CRM_ASSERT(flags & crm_ipc_client_response); CRM_LOG_ASSERT(c->request_id == 0); /* This means the client has two synchronous events in-flight */ c->request_id = id; /* Reply only to the last one */ } crm_xml_add(request, PCMK__XA_ST_CLIENTID, c->id); crm_xml_add(request, PCMK__XA_ST_CLIENTNAME, pcmk__client_name(c)); crm_xml_add(request, PCMK__XA_ST_CLIENTNODE, stonith_our_uname); crm_log_xml_trace(request, "ipc-received"); stonith_command(c, id, flags, request, NULL); free_xml(request); return 0; } /* Error code means? */ static int32_t st_ipc_closed(qb_ipcs_connection_t * c) { pcmk__client_t *client = pcmk__find_client(c); if (client == NULL) { return 0; } crm_trace("Connection %p closed", c); pcmk__free_client(client); /* 0 means: yes, go ahead and destroy the connection */ return 0; } static void st_ipc_destroy(qb_ipcs_connection_t * c) { crm_trace("Connection %p destroyed", c); st_ipc_closed(c); } static void stonith_peer_callback(xmlNode * msg, void *private_data) { const char *remote_peer = crm_element_value(msg, PCMK__XA_SRC); const char *op = crm_element_value(msg, PCMK__XA_ST_OP); if (pcmk__str_eq(op, STONITH_OP_POKE, pcmk__str_none)) { return; } crm_log_xml_trace(msg, "Peer[inbound]"); stonith_command(NULL, 0, 0, msg, remote_peer); } #if SUPPORT_COROSYNC static void stonith_peer_ais_callback(cpg_handle_t handle, const struct cpg_name *groupName, uint32_t nodeid, uint32_t pid, void *msg, size_t msg_len) { uint32_t kind = 0; xmlNode *xml = NULL; const char *from = NULL; char *data = pcmk_message_common_cs(handle, nodeid, pid, msg, &kind, &from); if(data == NULL) { return; } if (kind == crm_class_cluster) { xml = pcmk__xml_parse(data); if (xml == NULL) { crm_err("Invalid XML: '%.120s'", data); free(data); return; } crm_xml_add(xml, PCMK__XA_SRC, from); stonith_peer_callback(xml, NULL); } free_xml(xml); free(data); return; } static void stonith_peer_cs_destroy(gpointer user_data) { crm_crit("Lost connection to cluster layer, shutting down"); stonith_shutdown(0); } #endif void do_local_reply(const xmlNode *notify_src, pcmk__client_t *client, int call_options) { /* send callback to originating child */ int local_rc = pcmk_rc_ok; int rid = 0; uint32_t ipc_flags = crm_ipc_server_event; if (pcmk_is_set(call_options, st_opt_sync_call)) { CRM_LOG_ASSERT(client->request_id); rid = client->request_id; client->request_id = 0; ipc_flags = crm_ipc_flags_none; } local_rc = pcmk__ipc_send_xml(client, rid, notify_src, ipc_flags); if (local_rc == pcmk_rc_ok) { crm_trace("Sent response %d to client %s", rid, pcmk__client_name(client)); } else { crm_warn("%synchronous reply to client %s failed: %s", (pcmk_is_set(call_options, st_opt_sync_call)? "S" : "As"), pcmk__client_name(client), pcmk_rc_str(local_rc)); } } uint64_t get_stonith_flag(const char *name) { if (pcmk__str_eq(name, PCMK__VALUE_ST_NOTIFY_FENCE, pcmk__str_none)) { return st_callback_notify_fence; } else if (pcmk__str_eq(name, STONITH_OP_DEVICE_ADD, pcmk__str_casei)) { return st_callback_device_add; } else if (pcmk__str_eq(name, STONITH_OP_DEVICE_DEL, pcmk__str_casei)) { return st_callback_device_del; } else if (pcmk__str_eq(name, PCMK__VALUE_ST_NOTIFY_HISTORY, pcmk__str_none)) { return st_callback_notify_history; } else if (pcmk__str_eq(name, PCMK__VALUE_ST_NOTIFY_HISTORY_SYNCED, pcmk__str_none)) { return st_callback_notify_history_synced; } return st_callback_unknown; } static void stonith_notify_client(gpointer key, gpointer value, gpointer user_data) { const xmlNode *update_msg = user_data; pcmk__client_t *client = value; const char *type = NULL; CRM_CHECK(client != NULL, return); CRM_CHECK(update_msg != NULL, return); type = crm_element_value(update_msg, PCMK__XA_SUBT); CRM_CHECK(type != NULL, crm_log_xml_err(update_msg, "notify"); return); if (client->ipcs == NULL) { crm_trace("Skipping client with NULL channel"); return; } if (pcmk_is_set(client->flags, get_stonith_flag(type))) { int rc = pcmk__ipc_send_xml(client, 0, update_msg, crm_ipc_server_event); if (rc != pcmk_rc_ok) { crm_warn("%s notification of client %s failed: %s " CRM_XS " id=%.8s rc=%d", type, pcmk__client_name(client), pcmk_rc_str(rc), client->id, rc); } else { crm_trace("Sent %s notification to client %s", type, pcmk__client_name(client)); } } } void do_stonith_async_timeout_update(const char *client_id, const char *call_id, int timeout) { pcmk__client_t *client = NULL; xmlNode *notify_data = NULL; if (!timeout || !call_id || !client_id) { return; } client = pcmk__find_client_by_id(client_id); if (!client) { return; } notify_data = create_xml_node(NULL, PCMK__XE_ST_ASYNC_TIMEOUT_VALUE); crm_xml_add(notify_data, PCMK__XA_T, PCMK__VALUE_ST_ASYNC_TIMEOUT_VALUE); crm_xml_add(notify_data, PCMK__XA_ST_CALLID, call_id); crm_xml_add_int(notify_data, PCMK__XA_ST_TIMEOUT, timeout); crm_trace("timeout update is %d for client %s and call id %s", timeout, client_id, call_id); if (client) { pcmk__ipc_send_xml(client, 0, notify_data, crm_ipc_server_event); } free_xml(notify_data); } /*! * \internal * \brief Notify relevant IPC clients of a fencing operation result * * \param[in] type Notification type * \param[in] result Result of fencing operation (assume success if NULL) * \param[in] data If not NULL, add to notification as call data */ void fenced_send_notification(const char *type, const pcmk__action_result_t *result, xmlNode *data) { /* TODO: Standardize the contents of data */ xmlNode *update_msg = create_xml_node(NULL, PCMK__XE_NOTIFY); CRM_LOG_ASSERT(type != NULL); crm_xml_add(update_msg, PCMK__XA_T, PCMK__VALUE_ST_NOTIFY); crm_xml_add(update_msg, PCMK__XA_SUBT, type); crm_xml_add(update_msg, PCMK__XA_ST_OP, type); stonith__xe_set_result(update_msg, result); if (data != NULL) { add_message_xml(update_msg, PCMK__XA_ST_CALLDATA, data); } crm_trace("Notifying clients"); pcmk__foreach_ipc_client(stonith_notify_client, update_msg); free_xml(update_msg); crm_trace("Notify complete"); } /*! * \internal * \brief Send notifications for a configuration change to subscribed clients * * \param[in] op Notification type (\c STONITH_OP_DEVICE_ADD, * \c STONITH_OP_DEVICE_DEL, \c STONITH_OP_LEVEL_ADD, or * \c STONITH_OP_LEVEL_DEL) * \param[in] result Operation result * \param[in] desc Description of what changed (either device ID or string * representation of level * ([])) */ void fenced_send_config_notification(const char *op, const pcmk__action_result_t *result, const char *desc) { xmlNode *notify_data = create_xml_node(NULL, op); CRM_CHECK(notify_data != NULL, return); crm_xml_add(notify_data, PCMK__XA_ST_DEVICE_ID, desc); fenced_send_notification(op, result, notify_data); free_xml(notify_data); } /*! * \internal * \brief Check whether a node does watchdog-fencing * * \param[in] node Name of node to check * * \return TRUE if node found in stonith_watchdog_targets * or stonith_watchdog_targets is empty indicating * all nodes are doing watchdog-fencing */ gboolean node_does_watchdog_fencing(const char *node) { return ((stonith_watchdog_targets == NULL) || pcmk__str_in_list(node, stonith_watchdog_targets, pcmk__str_casei)); } void stonith_shutdown(int nsig) { crm_info("Terminating with %d clients", pcmk__ipc_client_count()); stonith_shutdown_flag = TRUE; if (mainloop != NULL && g_main_loop_is_running(mainloop)) { g_main_loop_quit(mainloop); } } static void stonith_cleanup(void) { fenced_cib_cleanup(); if (ipcs) { qb_ipcs_destroy(ipcs); } crm_peer_destroy(); pcmk__client_cleanup(); free_stonith_remote_op_list(); free_topology_list(); free_device_list(); free_metadata_cache(); fenced_unregister_handlers(); free(stonith_our_uname); stonith_our_uname = NULL; } static gboolean stand_alone_cpg_cb(const gchar *option_name, const gchar *optarg, gpointer data, GError **error) { stand_alone = FALSE; options.no_cib_connect = true; return TRUE; } struct qb_ipcs_service_handlers ipc_callbacks = { .connection_accept = st_ipc_accept, .connection_created = NULL, .msg_process = st_ipc_dispatch, .connection_closed = st_ipc_closed, .connection_destroyed = st_ipc_destroy }; /*! * \internal * \brief Callback for peer status changes * * \param[in] type What changed * \param[in] node What peer had the change * \param[in] data Previous value of what changed */ static void st_peer_update_callback(enum crm_status_type type, crm_node_t * node, const void *data) { if ((type != crm_status_processes) && !pcmk_is_set(node->flags, crm_remote_node)) { /* * This is a hack until we can send to a nodeid and/or we fix node name lookups * These messages are ignored in stonith_peer_callback() */ xmlNode *query = create_xml_node(NULL, PCMK__XE_STONITH_COMMAND); crm_xml_add(query, PCMK__XA_T, PCMK__VALUE_STONITH_NG); crm_xml_add(query, PCMK__XA_ST_OP, STONITH_OP_POKE); crm_debug("Broadcasting our uname because of node %u", node->id); send_cluster_message(NULL, crm_msg_stonith_ng, query, FALSE); free_xml(query); } } static pcmk__cluster_option_t fencer_options[] = { /* name, old name, type, allowed values, * default value, validator, * flags, * short description, * long description */ { PCMK_STONITH_HOST_ARGUMENT, NULL, "string", NULL, "port", NULL, pcmk__opt_advanced, N_("An alternate parameter to supply instead of 'port'"), N_("Some devices do not support the standard 'port' parameter or may " "provide additional ones. Use this to specify an alternate, device-" "specific, parameter that should indicate the machine to be " "fenced. A value of \"none\" can be used to tell the cluster not " "to supply any additional parameters."), }, { PCMK_STONITH_HOST_MAP, NULL, "string", NULL, NULL, NULL, pcmk__opt_none, N_("A mapping of node names to port numbers for devices that do not " "support node names."), N_("For example, \"node1:1;node2:2,3\" would tell the cluster to use " "port 1 for node1 and ports 2 and 3 for node2."), }, { PCMK_STONITH_HOST_LIST, NULL, "string", NULL, NULL, NULL, pcmk__opt_none, N_("A list of nodes that can be targeted by this device (optional " "unless pcmk_host_list=\"static-list\")"), N_("For example, \"node1,node2,node3\"."), }, { PCMK_STONITH_HOST_CHECK, NULL, "select", "dynamic-list, static-list, status, none", NULL, NULL, pcmk__opt_none, N_("How to determine which nodes can be targeted by the device"), N_("Use \"dynamic-list\" to query the device via the 'list' command; " "\"static-list\" to check the pcmk_host_list attribute; " "\"status\" to query the device via the 'status' command; or " "\"none\" to assume every device can fence every node. " "The default value is \"static-list\" if pcmk_host_map or " "pcmk_host_list is set; otherwise \"dynamic-list\" if the device " "supports the list operation; otherwise \"status\" if the device " "supports the status operation; otherwise \"none\""), }, { PCMK_STONITH_DELAY_MAX, NULL, "time", NULL, "0s", NULL, pcmk__opt_none, N_("Enable a delay of no more than the time specified before executing " "fencing actions."), N_("Enable a delay of no more than the time specified before executing " "fencing actions. Pacemaker derives the overall delay by taking " "the value of pcmk_delay_base and adding a random delay value such " "that the sum is kept below this maximum."), }, { PCMK_STONITH_DELAY_BASE, NULL, "string", NULL, "0s", NULL, pcmk__opt_none, N_("Enable a base delay for fencing actions and specify base delay " "value."), N_("This enables a static delay for fencing actions, which can help " "avoid \"death matches\" where two nodes try to fence each other " "at the same time. If pcmk_delay_max is also used, a random delay " "will be added such that the total delay is kept below that value. " "This can be set to a single time value to apply to any node " "targeted by this device (useful if a separate device is " "configured for each target), or to a node map (for example, " "\"node1:1s;node2:5\") to set a different value for each target."), }, { PCMK_STONITH_ACTION_LIMIT, NULL, "integer", NULL, "1", NULL, pcmk__opt_none, N_("The maximum number of actions can be performed in parallel on this " "device"), N_("Cluster property concurrent-fencing=\"true\" needs to be " "configured first. Then use this to specify the maximum number of " "actions can be performed in parallel on this device. A value of " "-1 means an unlimited number of actions can be performed in " "parallel."), }, { "pcmk_reboot_action", NULL, "string", NULL, PCMK_ACTION_REBOOT, NULL, pcmk__opt_advanced, N_("An alternate command to run instead of 'reboot'"), N_("Some devices do not support the standard commands or may provide " "additional ones. Use this to specify an alternate, device-" "specific, command that implements the 'reboot' action."), }, { "pcmk_reboot_timeout", NULL, "time", NULL, "60s", NULL, pcmk__opt_advanced, N_("Specify an alternate timeout to use for 'reboot' actions instead " "of stonith-timeout"), N_("Some devices need much more/less time to complete than normal. " "Use this to specify an alternate, device-specific, timeout for " "'reboot' actions."), }, { "pcmk_reboot_retries", NULL, "integer", NULL, "2", NULL, pcmk__opt_advanced, N_("The maximum number of times to try the 'reboot' command within the " "timeout period"), N_("Some devices do not support multiple connections. Operations may " "\"fail\" if the device is busy with another task. In that case, " "Pacemaker will automatically retry the operation if there is time " "remaining. Use this option to alter the number of times Pacemaker " "tries a 'reboot' action before giving up."), }, { "pcmk_off_action", NULL, "string", NULL, PCMK_ACTION_OFF, NULL, pcmk__opt_advanced, N_("An alternate command to run instead of 'off'"), N_("Some devices do not support the standard commands or may provide " "additional ones. Use this to specify an alternate, device-" "specific, command that implements the 'off' action."), }, { "pcmk_off_timeout", NULL, "time", NULL, "60s", NULL, pcmk__opt_advanced, N_("Specify an alternate timeout to use for 'off' actions instead of " "stonith-timeout"), N_("Some devices need much more/less time to complete than normal. " "Use this to specify an alternate, device-specific, timeout for " "'off' actions."), }, { "pcmk_off_retries", NULL, "integer", NULL, "2", NULL, pcmk__opt_advanced, N_("The maximum number of times to try the 'off' command within the " "timeout period"), N_("Some devices do not support multiple connections. Operations may " "\"fail\" if the device is busy with another task. In that case, " "Pacemaker will automatically retry the operation if there is time " "remaining. Use this option to alter the number of times Pacemaker " "tries a 'off' action before giving up."), }, { "pcmk_on_action", NULL, "string", NULL, PCMK_ACTION_ON, NULL, pcmk__opt_advanced, N_("An alternate command to run instead of 'on'"), N_("Some devices do not support the standard commands or may provide " "additional ones. Use this to specify an alternate, device-" "specific, command that implements the 'on' action."), }, { "pcmk_on_timeout", NULL, "time", NULL, "60s", NULL, pcmk__opt_advanced, N_("Specify an alternate timeout to use for 'on' actions instead of " "stonith-timeout"), N_("Some devices need much more/less time to complete than normal. " "Use this to specify an alternate, device-specific, timeout for " "'on' actions."), }, { "pcmk_on_retries", NULL, "integer", NULL, "2", NULL, pcmk__opt_advanced, N_("The maximum number of times to try the 'on' command within the " "timeout period"), N_("Some devices do not support multiple connections. Operations may " "\"fail\" if the device is busy with another task. In that case, " "Pacemaker will automatically retry the operation if there is time " "remaining. Use this option to alter the number of times Pacemaker " "tries a 'on' action before giving up."), }, { "pcmk_list_action", NULL, "string", NULL, PCMK_ACTION_LIST, NULL, pcmk__opt_advanced, N_("An alternate command to run instead of 'list'"), N_("Some devices do not support the standard commands or may provide " "additional ones. Use this to specify an alternate, device-" "specific, command that implements the 'list' action."), }, { "pcmk_list_timeout", NULL, "time", NULL, "60s", NULL, pcmk__opt_advanced, N_("Specify an alternate timeout to use for 'list' actions instead of " "stonith-timeout"), N_("Some devices need much more/less time to complete than normal. " "Use this to specify an alternate, device-specific, timeout for " "'list' actions."), }, { "pcmk_list_retries", NULL, "integer", NULL, "2", NULL, pcmk__opt_advanced, N_("The maximum number of times to try the 'list' command within the " "timeout period"), N_("Some devices do not support multiple connections. Operations may " "\"fail\" if the device is busy with another task. In that case, " "Pacemaker will automatically retry the operation if there is time " "remaining. Use this option to alter the number of times Pacemaker " "tries a 'list' action before giving up."), }, { "pcmk_monitor_action", NULL, "string", NULL, PCMK_ACTION_MONITOR, NULL, pcmk__opt_advanced, N_("An alternate command to run instead of 'monitor'"), N_("Some devices do not support the standard commands or may provide " "additional ones. Use this to specify an alternate, device-" "specific, command that implements the 'monitor' action."), }, { "pcmk_monitor_timeout", NULL, "time", NULL, "60s", NULL, pcmk__opt_advanced, N_("Specify an alternate timeout to use for 'monitor' actions instead " "of stonith-timeout"), N_("Some devices need much more/less time to complete than normal. " "Use this to specify an alternate, device-specific, timeout for " "'monitor' actions."), }, { "pcmk_monitor_retries", NULL, "integer", NULL, "2", NULL, pcmk__opt_advanced, N_("The maximum number of times to try the 'monitor' command within " "the timeout period"), N_("Some devices do not support multiple connections. Operations may " "\"fail\" if the device is busy with another task. In that case, " "Pacemaker will automatically retry the operation if there is time " "remaining. Use this option to alter the number of times Pacemaker " "tries a 'monitor' action before giving up."), }, { "pcmk_status_action", NULL, "string", NULL, PCMK_ACTION_STATUS, NULL, pcmk__opt_advanced, N_("An alternate command to run instead of 'status'"), N_("Some devices do not support the standard commands or may provide " "additional ones. Use this to specify an alternate, device-" "specific, command that implements the 'status' action."), }, { "pcmk_status_timeout", NULL, "time", NULL, "60s", NULL, pcmk__opt_advanced, N_("Specify an alternate timeout to use for 'status' actions instead " "of stonith-timeout"), N_("Some devices need much more/less time to complete than normal. " "Use this to specify an alternate, device-specific, timeout for " "'status' actions."), }, { "pcmk_status_retries", NULL, "integer", NULL, "2", NULL, pcmk__opt_advanced, N_("The maximum number of times to try the 'status' command within " "the timeout period"), N_("Some devices do not support multiple connections. Operations may " "\"fail\" if the device is busy with another task. In that case, " "Pacemaker will automatically retry the operation if there is time " "remaining. Use this option to alter the number of times Pacemaker " "tries a 'status' action before giving up."), }, { NULL, }, }; static int fencer_metadata(void) { // @TODO Use pcmk__daemon_metadata when fencer_options moves to options.c const char *name = "pacemaker-fenced"; const char *desc_short = N_("Instance attributes available for all " "\"stonith\"-class resources"); const char *desc_long = N_("Instance attributes available for all " "\"stonith\"-class resources and used by " "Pacemaker's fence daemon, formerly known as " "stonithd"); pcmk__output_t *tmp_out = NULL; xmlNode *top = NULL; const xmlNode *metadata = NULL; - gchar *metadata_s = NULL; + GString *metadata_s = NULL; int rc = pcmk__output_new(&tmp_out, "xml", "/dev/null", NULL); if (rc != pcmk_rc_ok) { return rc; } out->message(tmp_out, "option-list", name, desc_short, desc_long, pcmk__opt_none, fencer_options, true); tmp_out->finish(tmp_out, CRM_EX_OK, false, (void **) &top); metadata = first_named_child(top, PCMK_XE_RESOURCE_AGENT); - metadata_s = pcmk__xml_dump(metadata, - pcmk__xml_fmt_pretty|pcmk__xml_fmt_text); - out->output_xml(out, PCMK_XE_METADATA, metadata_s); + metadata_s = g_string_sized_new(16384); + pcmk__xml_string(metadata, pcmk__xml_fmt_pretty|pcmk__xml_fmt_text, + metadata_s, 0); + + out->output_xml(out, PCMK_XE_METADATA, metadata_s->str); pcmk__output_free(tmp_out); free_xml(top); - g_free(metadata_s); + g_string_free(metadata_s, TRUE); return pcmk_rc_ok; } static GOptionEntry entries[] = { { "stand-alone", 's', G_OPTION_FLAG_NONE, G_OPTION_ARG_NONE, &stand_alone, N_("Deprecated (will be removed in a future release)"), NULL }, { "stand-alone-w-cpg", 'c', G_OPTION_FLAG_NO_ARG, G_OPTION_ARG_CALLBACK, stand_alone_cpg_cb, N_("Intended for use in regression testing only"), NULL }, { "logfile", 'l', G_OPTION_FLAG_NONE, G_OPTION_ARG_FILENAME_ARRAY, &options.log_files, N_("Send logs to the additional named logfile"), NULL }, { NULL } }; static GOptionContext * build_arg_context(pcmk__common_args_t *args, GOptionGroup **group) { GOptionContext *context = NULL; context = pcmk__build_arg_context(args, "text (default), xml", group, "[metadata]"); pcmk__add_main_args(context, entries); return context; } int main(int argc, char **argv) { int rc = pcmk_rc_ok; crm_cluster_t *cluster = NULL; crm_ipc_t *old_instance = 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, "l"); GOptionContext *context = build_arg_context(args, &output_group); crm_log_preinit(NULL, argc, argv); pcmk__register_formats(output_group, formats); if (!g_option_context_parse_strv(context, &processed_args, &error)) { exit_code = CRM_EX_USAGE; goto done; } 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; } if ((g_strv_length(processed_args) >= 2) && pcmk__str_eq(processed_args[1], "metadata", pcmk__str_none)) { rc = fencer_metadata(); if (rc != pcmk_rc_ok) { exit_code = CRM_EX_FATAL; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Unable to display metadata: %s", pcmk_rc_str(rc)); } goto done; } // Open additional log files pcmk__add_logfiles(options.log_files, out); crm_log_init(NULL, LOG_INFO + args->verbosity, TRUE, (args->verbosity > 0), argc, argv, FALSE); crm_notice("Starting Pacemaker fencer"); old_instance = crm_ipc_new("stonith-ng", 0); if (old_instance == NULL) { /* crm_ipc_new() will have already logged an error message with * crm_err() */ exit_code = CRM_EX_FATAL; goto done; } if (pcmk__connect_generic_ipc(old_instance) == pcmk_rc_ok) { // IPC endpoint already up crm_ipc_close(old_instance); crm_ipc_destroy(old_instance); crm_err("pacemaker-fenced is already active, aborting startup"); goto done; } else { // Not up or not authentic, we'll proceed either way crm_ipc_destroy(old_instance); old_instance = NULL; } mainloop_add_signal(SIGTERM, stonith_shutdown); crm_peer_init(); rc = fenced_scheduler_init(); if (rc != pcmk_rc_ok) { exit_code = CRM_EX_FATAL; g_set_error(&error, PCMK__EXITC_ERROR, exit_code, "Error initializing scheduler data: %s", pcmk_rc_str(rc)); goto done; } cluster = pcmk_cluster_new(); if (!stand_alone) { #if SUPPORT_COROSYNC if (is_corosync_cluster()) { cluster->destroy = stonith_peer_cs_destroy; cluster->cpg.cpg_deliver_fn = stonith_peer_ais_callback; cluster->cpg.cpg_confchg_fn = pcmk_cpg_membership; } #endif // SUPPORT_COROSYNC crm_set_status_callback(&st_peer_update_callback); if (crm_cluster_connect(cluster) == FALSE) { exit_code = CRM_EX_FATAL; crm_crit("Cannot sign in to the cluster... terminating"); goto done; } pcmk__str_update(&stonith_our_uname, cluster->uname); if (!options.no_cib_connect) { setup_cib(); } } else { pcmk__str_update(&stonith_our_uname, "localhost"); crm_warn("Stand-alone mode is deprecated and will be removed " "in a future release"); } init_device_list(); init_topology_list(); pcmk__serve_fenced_ipc(&ipcs, &ipc_callbacks); // Create the mainloop and run it... mainloop = g_main_loop_new(NULL, FALSE); crm_notice("Pacemaker fencer successfully started and accepting connections"); g_main_loop_run(mainloop); done: g_strfreev(processed_args); pcmk__free_arg_context(context); g_strfreev(options.log_files); stonith_cleanup(); pcmk_cluster_free(cluster); fenced_scheduler_cleanup(); 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/include/crm/common/xml_io_internal.h b/include/crm/common/xml_io_internal.h index 18a84840c2..6c2b625504 100644 --- a/include/crm/common/xml_io_internal.h +++ b/include/crm/common/xml_io_internal.h @@ -1,35 +1,34 @@ /* * Copyright 2017-2024 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__XML_IO_INTERNAL__H #define PCMK__XML_IO_INTERNAL__H /* * Internal-only wrappers for and extensions to libxml2 I/O */ #include // bool +#include // GString #include // xmlNode xmlNode *pcmk__xml_read(const char *filename); xmlNode *pcmk__xml_parse(const char *input); void pcmk__xml_string(const xmlNode *data, uint32_t options, GString *buffer, int depth); int pcmk__xml2fd(int fd, xmlNode *cur); int pcmk__xml_write_fd(const xmlNode *xml, const char *filename, int fd, bool compress, unsigned int *nbytes); int pcmk__xml_write_file(const xmlNode *xml, const char *filename, bool compress, unsigned int *nbytes); -gchar *pcmk__xml_dump(const xmlNode *xml, uint32_t flags); - #endif // PCMK__XML_IO_INTERNAL__H diff --git a/lib/cluster/cpg.c b/lib/cluster/cpg.c index 082cdc1f9e..d592561685 100644 --- a/lib/cluster/cpg.c +++ b/lib/cluster/cpg.c @@ -1,1094 +1,1096 @@ /* * 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 #include #include #include #include #include #include #include /* PCMK__SPECIAL_PID* */ #include "crmcluster_private.h" /* @TODO Once we can update the public API to require crm_cluster_t* in more * functions, we can ditch this in favor of cluster->cpg_handle. */ static cpg_handle_t pcmk_cpg_handle = 0; // @TODO These could be moved to crm_cluster_t* at that time as well static bool cpg_evicted = false; static GList *cs_message_queue = NULL; static int cs_message_timer = 0; struct pcmk__cpg_host_s { uint32_t id; uint32_t pid; gboolean local; enum crm_ais_msg_types type; uint32_t size; char uname[MAX_NAME]; } __attribute__ ((packed)); typedef struct pcmk__cpg_host_s pcmk__cpg_host_t; struct pcmk__cpg_msg_s { struct qb_ipc_response_header header __attribute__ ((aligned(8))); uint32_t id; gboolean is_compressed; pcmk__cpg_host_t host; pcmk__cpg_host_t sender; uint32_t size; uint32_t compressed_size; /* 584 bytes */ char data[0]; } __attribute__ ((packed)); typedef struct pcmk__cpg_msg_s pcmk__cpg_msg_t; static void crm_cs_flush(gpointer data); #define msg_data_len(msg) (msg->is_compressed?msg->compressed_size:msg->size) #define cs_repeat(rc, counter, max, code) do { \ rc = code; \ if ((rc == CS_ERR_TRY_AGAIN) || (rc == CS_ERR_QUEUE_FULL)) { \ counter++; \ crm_debug("Retrying operation after %ds", counter); \ sleep(counter); \ } else { \ break; \ } \ } while (counter < max) /*! * \brief Disconnect from Corosync CPG * * \param[in,out] cluster Cluster to disconnect */ void cluster_disconnect_cpg(crm_cluster_t *cluster) { pcmk_cpg_handle = 0; if (cluster->cpg_handle) { crm_trace("Disconnecting CPG"); cpg_leave(cluster->cpg_handle, &cluster->group); cpg_finalize(cluster->cpg_handle); cluster->cpg_handle = 0; } else { crm_info("No CPG connection"); } } /*! * \brief Get the local Corosync node ID (via CPG) * * \param[in] handle CPG connection to use (or 0 to use new connection) * * \return Corosync ID of local node (or 0 if not known) */ uint32_t get_local_nodeid(cpg_handle_t handle) { cs_error_t rc = CS_OK; int retries = 0; static uint32_t local_nodeid = 0; cpg_handle_t local_handle = handle; cpg_model_v1_data_t cpg_model_info = {CPG_MODEL_V1, NULL, NULL, NULL, 0}; int fd = -1; uid_t found_uid = 0; gid_t found_gid = 0; pid_t found_pid = 0; int rv; if(local_nodeid != 0) { return local_nodeid; } if(handle == 0) { crm_trace("Creating connection"); cs_repeat(rc, retries, 5, cpg_model_initialize(&local_handle, CPG_MODEL_V1, (cpg_model_data_t *)&cpg_model_info, NULL)); if (rc != CS_OK) { crm_err("Could not connect to the CPG API: %s (%d)", cs_strerror(rc), rc); return 0; } rc = cpg_fd_get(local_handle, &fd); if (rc != CS_OK) { crm_err("Could not obtain the CPG API connection: %s (%d)", cs_strerror(rc), rc); goto bail; } /* CPG provider run as root (in given user namespace, anyway)? */ if (!(rv = crm_ipc_is_authentic_process(fd, (uid_t) 0,(gid_t) 0, &found_pid, &found_uid, &found_gid))) { crm_err("CPG provider is not authentic:" " process %lld (uid: %lld, gid: %lld)", (long long) PCMK__SPECIAL_PID_AS_0(found_pid), (long long) found_uid, (long long) found_gid); goto bail; } else if (rv < 0) { crm_err("Could not verify authenticity of CPG provider: %s (%d)", strerror(-rv), -rv); goto bail; } } if (rc == CS_OK) { retries = 0; crm_trace("Performing lookup"); cs_repeat(rc, retries, 5, cpg_local_get(local_handle, &local_nodeid)); } if (rc != CS_OK) { crm_err("Could not get local node id from the CPG API: %s (%d)", pcmk__cs_err_str(rc), rc); } bail: if(handle == 0) { crm_trace("Closing connection"); cpg_finalize(local_handle); } crm_debug("Local nodeid is %u", local_nodeid); return local_nodeid; } /*! * \internal * \brief Callback function for Corosync message queue timer * * \param[in] data CPG handle * * \return FALSE (to indicate to glib that timer should not be removed) */ static gboolean crm_cs_flush_cb(gpointer data) { cs_message_timer = 0; crm_cs_flush(data); return FALSE; } // Send no more than this many CPG messages in one flush #define CS_SEND_MAX 200 /*! * \internal * \brief Send messages in Corosync CPG message queue * * \param[in] data CPG handle */ static void crm_cs_flush(gpointer data) { unsigned int sent = 0; guint queue_len = 0; cs_error_t rc = 0; cpg_handle_t *handle = (cpg_handle_t *) data; if (*handle == 0) { crm_trace("Connection is dead"); return; } queue_len = g_list_length(cs_message_queue); if (((queue_len % 1000) == 0) && (queue_len > 1)) { crm_err("CPG queue has grown to %d", queue_len); } else if (queue_len == CS_SEND_MAX) { crm_warn("CPG queue has grown to %d", queue_len); } if (cs_message_timer != 0) { /* There is already a timer, wait until it goes off */ crm_trace("Timer active %d", cs_message_timer); return; } while ((cs_message_queue != NULL) && (sent < CS_SEND_MAX)) { struct iovec *iov = cs_message_queue->data; rc = cpg_mcast_joined(*handle, CPG_TYPE_AGREED, iov, 1); if (rc != CS_OK) { break; } sent++; crm_trace("CPG message sent, size=%llu", (unsigned long long) iov->iov_len); cs_message_queue = g_list_remove(cs_message_queue, iov); free(iov->iov_base); free(iov); } queue_len -= sent; do_crm_log((queue_len > 5)? LOG_INFO : LOG_TRACE, "Sent %u CPG message%s (%d still queued): %s (rc=%d)", sent, pcmk__plural_s(sent), queue_len, pcmk__cs_err_str(rc), (int) rc); if (cs_message_queue) { uint32_t delay_ms = 100; if (rc != CS_OK) { /* Proportionally more if sending failed but cap at 1s */ delay_ms = QB_MIN(1000, CS_SEND_MAX + (10 * queue_len)); } cs_message_timer = g_timeout_add(delay_ms, crm_cs_flush_cb, data); } } /*! * \internal * \brief Dispatch function for CPG handle * * \param[in,out] user_data Cluster object * * \return 0 on success, -1 on error (per mainloop_io_t interface) */ static int pcmk_cpg_dispatch(gpointer user_data) { cs_error_t rc = CS_OK; crm_cluster_t *cluster = (crm_cluster_t *) user_data; rc = cpg_dispatch(cluster->cpg_handle, CS_DISPATCH_ONE); if (rc != CS_OK) { crm_err("Connection to the CPG API failed: %s (%d)", pcmk__cs_err_str(rc), rc); cpg_finalize(cluster->cpg_handle); cluster->cpg_handle = 0; return -1; } else if (cpg_evicted) { crm_err("Evicted from CPG membership"); return -1; } return 0; } static inline const char * ais_dest(const pcmk__cpg_host_t *host) { if (host->local) { return "local"; } else if (host->size > 0) { return host->uname; } else { return ""; } } static inline const char * msg_type2text(enum crm_ais_msg_types type) { const char *text = "unknown"; switch (type) { case crm_msg_none: text = "unknown"; break; case crm_msg_ais: text = "ais"; break; case crm_msg_cib: text = "cib"; break; case crm_msg_crmd: text = "crmd"; break; case crm_msg_pe: text = "pengine"; break; case crm_msg_te: text = "tengine"; break; case crm_msg_lrmd: text = "lrmd"; break; case crm_msg_attrd: text = "attrd"; break; case crm_msg_stonithd: text = "stonithd"; break; case crm_msg_stonith_ng: text = "stonith-ng"; break; } return text; } /*! * \internal * \brief Check whether a Corosync CPG message is valid * * \param[in] msg Corosync CPG message to check * * \return true if \p msg is valid, otherwise false */ static bool check_message_sanity(const pcmk__cpg_msg_t *msg) { int32_t payload_size = msg->header.size - sizeof(pcmk__cpg_msg_t); if (payload_size < 1) { crm_err("%sCPG message %d from %s invalid: " "Claimed size of %d bytes is too small " CRM_XS " from %s[%u] to %s@%s", (msg->is_compressed? "Compressed " : ""), msg->id, ais_dest(&(msg->sender)), (int) msg->header.size, msg_type2text(msg->sender.type), msg->sender.pid, msg_type2text(msg->host.type), ais_dest(&(msg->host))); return false; } if (msg->header.error != CS_OK) { crm_err("%sCPG message %d from %s invalid: " "Sender indicated error %d " CRM_XS " from %s[%u] to %s@%s", (msg->is_compressed? "Compressed " : ""), msg->id, ais_dest(&(msg->sender)), msg->header.error, msg_type2text(msg->sender.type), msg->sender.pid, msg_type2text(msg->host.type), ais_dest(&(msg->host))); return false; } if (msg_data_len(msg) != payload_size) { crm_err("%sCPG message %d from %s invalid: " "Total size %d inconsistent with payload size %d " CRM_XS " from %s[%u] to %s@%s", (msg->is_compressed? "Compressed " : ""), msg->id, ais_dest(&(msg->sender)), (int) msg->header.size, (int) msg_data_len(msg), msg_type2text(msg->sender.type), msg->sender.pid, msg_type2text(msg->host.type), ais_dest(&(msg->host))); return false; } if (!msg->is_compressed && /* msg->size != (strlen(msg->data) + 1) would be a stronger check, * but checking the last byte or two should be quick */ (((msg->size > 1) && (msg->data[msg->size - 2] == '\0')) || (msg->data[msg->size - 1] != '\0'))) { crm_err("CPG message %d from %s invalid: " "Payload does not end at byte %llu " CRM_XS " from %s[%u] to %s@%s", msg->id, ais_dest(&(msg->sender)), (unsigned long long) msg->size, msg_type2text(msg->sender.type), msg->sender.pid, msg_type2text(msg->host.type), ais_dest(&(msg->host))); return false; } crm_trace("Verified %d-byte %sCPG message %d from %s[%u]@%s to %s@%s", (int) msg->header.size, (msg->is_compressed? "compressed " : ""), msg->id, msg_type2text(msg->sender.type), msg->sender.pid, ais_dest(&(msg->sender)), msg_type2text(msg->host.type), ais_dest(&(msg->host))); return true; } /*! * \brief Extract text data from a Corosync CPG message * * \param[in] handle CPG connection (to get local node ID if not known) * \param[in] nodeid Corosync ID of node that sent message * \param[in] pid Process ID of message sender (for logging only) * \param[in,out] content CPG message * \param[out] kind If not NULL, will be set to CPG header ID * (which should be an enum crm_ais_msg_class value, * currently always crm_class_cluster) * \param[out] from If not NULL, will be set to sender uname * (valid for the lifetime of \p content) * * \return Newly allocated string with message data * \note It is the caller's responsibility to free the return value with free(). */ char * pcmk_message_common_cs(cpg_handle_t handle, uint32_t nodeid, uint32_t pid, void *content, uint32_t *kind, const char **from) { char *data = NULL; pcmk__cpg_msg_t *msg = (pcmk__cpg_msg_t *) content; if(handle) { // Do filtering and field massaging uint32_t local_nodeid = get_local_nodeid(handle); const char *local_name = get_local_node_name(); if (msg->sender.id > 0 && msg->sender.id != nodeid) { crm_err("Nodeid mismatch from %d.%d: claimed nodeid=%u", nodeid, pid, msg->sender.id); return NULL; } else if (msg->host.id != 0 && (local_nodeid != msg->host.id)) { /* Not for us */ crm_trace("Not for us: %u != %u", msg->host.id, local_nodeid); return NULL; } else if (msg->host.size != 0 && !pcmk__str_eq(msg->host.uname, local_name, pcmk__str_casei)) { /* Not for us */ crm_trace("Not for us: %s != %s", msg->host.uname, local_name); return NULL; } msg->sender.id = nodeid; if (msg->sender.size == 0) { crm_node_t *peer = pcmk__get_node(nodeid, NULL, NULL, pcmk__node_search_cluster); if (peer == NULL) { crm_err("Peer with nodeid=%u is unknown", nodeid); } else if (peer->uname == NULL) { crm_err("No uname for peer with nodeid=%u", nodeid); } else { crm_notice("Fixing uname for peer with nodeid=%u", nodeid); msg->sender.size = strlen(peer->uname); memset(msg->sender.uname, 0, MAX_NAME); memcpy(msg->sender.uname, peer->uname, msg->sender.size); } } } crm_trace("Got new%s message (size=%d, %d, %d)", msg->is_compressed ? " compressed" : "", msg_data_len(msg), msg->size, msg->compressed_size); if (kind != NULL) { *kind = msg->header.id; } if (from != NULL) { *from = msg->sender.uname; } if (msg->is_compressed && msg->size > 0) { int rc = BZ_OK; char *uncompressed = NULL; unsigned int new_size = msg->size + 1; if (!check_message_sanity(msg)) { goto badmsg; } crm_trace("Decompressing message data"); uncompressed = calloc(1, new_size); rc = BZ2_bzBuffToBuffDecompress(uncompressed, &new_size, msg->data, msg->compressed_size, 1, 0); rc = pcmk__bzlib2rc(rc); if (rc != pcmk_rc_ok) { crm_err("Decompression failed: %s " CRM_XS " rc=%d", pcmk_rc_str(rc), rc); free(uncompressed); goto badmsg; } CRM_ASSERT(new_size == msg->size); data = uncompressed; } else if (!check_message_sanity(msg)) { goto badmsg; } else { data = strdup(msg->data); } // Is this necessary? pcmk__get_node(msg->sender.id, msg->sender.uname, NULL, pcmk__node_search_cluster); crm_trace("Payload: %.200s", data); return data; badmsg: crm_err("Invalid message (id=%d, dest=%s:%s, from=%s:%s.%d):" " min=%d, total=%d, size=%d, bz2_size=%d", msg->id, ais_dest(&(msg->host)), msg_type2text(msg->host.type), ais_dest(&(msg->sender)), msg_type2text(msg->sender.type), msg->sender.pid, (int)sizeof(pcmk__cpg_msg_t), msg->header.size, msg->size, msg->compressed_size); free(data); return NULL; } /*! * \internal * \brief Compare cpg_address objects by node ID * * \param[in] first First cpg_address structure to compare * \param[in] second Second cpg_address structure to compare * * \return Negative number if first's node ID is lower, * positive number if first's node ID is greater, * or 0 if both node IDs are equal */ static int cmp_member_list_nodeid(const void *first, const void *second) { const struct cpg_address *const a = *((const struct cpg_address **) first), *const b = *((const struct cpg_address **) second); if (a->nodeid < b->nodeid) { return -1; } else if (a->nodeid > b->nodeid) { return 1; } /* don't bother with "reason" nor "pid" */ return 0; } /*! * \internal * \brief Get a readable string equivalent of a cpg_reason_t value * * \param[in] reason CPG reason value * * \return Readable string suitable for logging */ static const char * cpgreason2str(cpg_reason_t reason) { switch (reason) { case CPG_REASON_JOIN: return " via cpg_join"; case CPG_REASON_LEAVE: return " via cpg_leave"; case CPG_REASON_NODEDOWN: return " via cluster exit"; case CPG_REASON_NODEUP: return " via cluster join"; case CPG_REASON_PROCDOWN: return " for unknown reason"; default: break; } return ""; } /*! * \internal * \brief Get a log-friendly node name * * \param[in] peer Node to check * * \return Node's uname, or readable string if not known */ static inline const char * peer_name(const crm_node_t *peer) { if (peer == NULL) { return "unknown node"; } else if (peer->uname == NULL) { return "peer node"; } else { return peer->uname; } } /*! * \internal * \brief Process a CPG peer's leaving the cluster * * \param[in] cpg_group_name CPG group name (for logging) * \param[in] event_counter Event number (for logging) * \param[in] local_nodeid Node ID of local node * \param[in] cpg_peer CPG peer that left * \param[in] sorted_member_list List of remaining members, qsort()-ed by ID * \param[in] member_list_entries Number of entries in \p sorted_member_list */ static void node_left(const char *cpg_group_name, int event_counter, uint32_t local_nodeid, const struct cpg_address *cpg_peer, const struct cpg_address **sorted_member_list, size_t member_list_entries) { crm_node_t *peer = pcmk__search_node_caches(cpg_peer->nodeid, NULL, pcmk__node_search_cluster); const struct cpg_address **rival = NULL; /* Most CPG-related Pacemaker code assumes that only one process on a node * can be in the process group, but Corosync does not impose this * limitation, and more than one can be a member in practice due to a * daemon attempting to start while another instance is already running. * * Check for any such duplicate instances, because we don't want to process * their leaving as if our actual peer left. If the peer that left still has * an entry in sorted_member_list (with a different PID), we will ignore the * leaving. * * @TODO Track CPG members' PIDs so we can tell exactly who left. */ if (peer != NULL) { rival = bsearch(&cpg_peer, sorted_member_list, member_list_entries, sizeof(const struct cpg_address *), cmp_member_list_nodeid); } if (rival == NULL) { crm_info("Group %s event %d: %s (node %u pid %u) left%s", cpg_group_name, event_counter, peer_name(peer), cpg_peer->nodeid, cpg_peer->pid, cpgreason2str(cpg_peer->reason)); if (peer != NULL) { crm_update_peer_proc(__func__, peer, crm_proc_cpg, PCMK_VALUE_OFFLINE); } } else if (cpg_peer->nodeid == local_nodeid) { crm_warn("Group %s event %d: duplicate local pid %u left%s", cpg_group_name, event_counter, cpg_peer->pid, cpgreason2str(cpg_peer->reason)); } else { crm_warn("Group %s event %d: " "%s (node %u) duplicate pid %u left%s (%u remains)", cpg_group_name, event_counter, peer_name(peer), cpg_peer->nodeid, cpg_peer->pid, cpgreason2str(cpg_peer->reason), (*rival)->pid); } } /*! * \brief Handle a CPG configuration change event * * \param[in] handle CPG connection * \param[in] cpg_name CPG group name * \param[in] member_list List of current CPG members * \param[in] member_list_entries Number of entries in \p member_list * \param[in] left_list List of CPG members that left * \param[in] left_list_entries Number of entries in \p left_list * \param[in] joined_list List of CPG members that joined * \param[in] joined_list_entries Number of entries in \p joined_list */ void pcmk_cpg_membership(cpg_handle_t handle, const struct cpg_name *groupName, const struct cpg_address *member_list, size_t member_list_entries, const struct cpg_address *left_list, size_t left_list_entries, const struct cpg_address *joined_list, size_t joined_list_entries) { int i; gboolean found = FALSE; static int counter = 0; uint32_t local_nodeid = get_local_nodeid(handle); const struct cpg_address **sorted; sorted = malloc(member_list_entries * sizeof(const struct cpg_address *)); CRM_ASSERT(sorted != NULL); for (size_t iter = 0; iter < member_list_entries; iter++) { sorted[iter] = member_list + iter; } /* so that the cross-matching multiply-subscribed nodes is then cheap */ qsort(sorted, member_list_entries, sizeof(const struct cpg_address *), cmp_member_list_nodeid); for (i = 0; i < left_list_entries; i++) { node_left(groupName->value, counter, local_nodeid, &left_list[i], sorted, member_list_entries); } free(sorted); sorted = NULL; for (i = 0; i < joined_list_entries; i++) { crm_info("Group %s event %d: node %u pid %u joined%s", groupName->value, counter, joined_list[i].nodeid, joined_list[i].pid, cpgreason2str(joined_list[i].reason)); } for (i = 0; i < member_list_entries; i++) { crm_node_t *peer = pcmk__get_node(member_list[i].nodeid, NULL, NULL, pcmk__node_search_cluster); if (member_list[i].nodeid == local_nodeid && member_list[i].pid != getpid()) { // See the note in node_left() crm_warn("Group %s event %d: detected duplicate local pid %u", groupName->value, counter, member_list[i].pid); continue; } crm_info("Group %s event %d: %s (node %u pid %u) is member", groupName->value, counter, peer_name(peer), member_list[i].nodeid, member_list[i].pid); /* If the caller left auto-reaping enabled, this will also update the * state to member. */ peer = crm_update_peer_proc(__func__, peer, crm_proc_cpg, PCMK_VALUE_ONLINE); if (peer && peer->state && strcmp(peer->state, CRM_NODE_MEMBER)) { /* The node is a CPG member, but we currently think it's not a * cluster member. This is possible only if auto-reaping was * disabled. The node may be joining, and we happened to get the CPG * notification before the quorum notification; or the node may have * just died, and we are processing its final messages; or a bug * has affected the peer cache. */ time_t now = time(NULL); if (peer->when_lost == 0) { // Track when we first got into this contradictory state peer->when_lost = now; } else if (now > (peer->when_lost + 60)) { // If it persists for more than a minute, update the state crm_warn("Node %u is member of group %s but was believed offline", member_list[i].nodeid, groupName->value); pcmk__update_peer_state(__func__, peer, CRM_NODE_MEMBER, 0); } } if (local_nodeid == member_list[i].nodeid) { found = TRUE; } } if (!found) { crm_err("Local node was evicted from group %s", groupName->value); cpg_evicted = true; } counter++; } /*! * \brief Connect to Corosync CPG * * \param[in,out] cluster Cluster object * * \return TRUE on success, otherwise FALSE */ gboolean cluster_connect_cpg(crm_cluster_t *cluster) { cs_error_t rc; int fd = -1; int retries = 0; uint32_t id = 0; crm_node_t *peer = NULL; cpg_handle_t handle = 0; const char *message_name = pcmk__message_name(crm_system_name); uid_t found_uid = 0; gid_t found_gid = 0; pid_t found_pid = 0; int rv; struct mainloop_fd_callbacks cpg_fd_callbacks = { .dispatch = pcmk_cpg_dispatch, .destroy = cluster->destroy, }; cpg_model_v1_data_t cpg_model_info = { .model = CPG_MODEL_V1, .cpg_deliver_fn = cluster->cpg.cpg_deliver_fn, .cpg_confchg_fn = cluster->cpg.cpg_confchg_fn, .cpg_totem_confchg_fn = NULL, .flags = 0, }; cpg_evicted = false; cluster->group.length = 0; cluster->group.value[0] = 0; /* group.value is char[128] */ strncpy(cluster->group.value, message_name, 127); cluster->group.value[127] = 0; cluster->group.length = 1 + QB_MIN(127, strlen(cluster->group.value)); cs_repeat(rc, retries, 30, cpg_model_initialize(&handle, CPG_MODEL_V1, (cpg_model_data_t *)&cpg_model_info, NULL)); if (rc != CS_OK) { crm_err("Could not connect to the CPG API: %s (%d)", cs_strerror(rc), rc); goto bail; } rc = cpg_fd_get(handle, &fd); if (rc != CS_OK) { crm_err("Could not obtain the CPG API connection: %s (%d)", cs_strerror(rc), rc); goto bail; } /* CPG provider run as root (in given user namespace, anyway)? */ if (!(rv = crm_ipc_is_authentic_process(fd, (uid_t) 0,(gid_t) 0, &found_pid, &found_uid, &found_gid))) { crm_err("CPG provider is not authentic:" " process %lld (uid: %lld, gid: %lld)", (long long) PCMK__SPECIAL_PID_AS_0(found_pid), (long long) found_uid, (long long) found_gid); rc = CS_ERR_ACCESS; goto bail; } else if (rv < 0) { crm_err("Could not verify authenticity of CPG provider: %s (%d)", strerror(-rv), -rv); rc = CS_ERR_ACCESS; goto bail; } id = get_local_nodeid(handle); if (id == 0) { crm_err("Could not get local node id from the CPG API"); goto bail; } cluster->nodeid = id; retries = 0; cs_repeat(rc, retries, 30, cpg_join(handle, &cluster->group)); if (rc != CS_OK) { crm_err("Could not join the CPG group '%s': %d", message_name, rc); goto bail; } pcmk_cpg_handle = handle; cluster->cpg_handle = handle; mainloop_add_fd("corosync-cpg", G_PRIORITY_MEDIUM, fd, cluster, &cpg_fd_callbacks); bail: if (rc != CS_OK) { cpg_finalize(handle); return FALSE; } peer = pcmk__get_node(id, NULL, NULL, pcmk__node_search_cluster); crm_update_peer_proc(__func__, peer, crm_proc_cpg, PCMK_VALUE_ONLINE); return TRUE; } /*! * \internal * \brief Send an XML message via Corosync CPG * * \param[in] msg XML message to send * \param[in] node Cluster node to send message to * \param[in] dest Type of message to send * * \return TRUE on success, otherwise FALSE */ bool pcmk__cpg_send_xml(const xmlNode *msg, const crm_node_t *node, enum crm_ais_msg_types dest) { bool rc = true; - gchar *data = pcmk__xml_dump(msg, 0); + GString *data = g_string_sized_new(1024); - rc = send_cluster_text(crm_class_cluster, data, FALSE, node, dest); - g_free(data); + pcmk__xml_string(msg, 0, data, 0); + + rc = send_cluster_text(crm_class_cluster, data->str, FALSE, node, dest); + g_string_free(data, TRUE); return rc; } /*! * \internal * \brief Send string data via Corosync CPG * * \param[in] msg_class Message class (to set as CPG header ID) * \param[in] data Data to send * \param[in] local What to set as host "local" value (which is never used) * \param[in] node Cluster node to send message to * \param[in] dest Type of message to send * * \return TRUE on success, otherwise FALSE */ gboolean send_cluster_text(enum crm_ais_msg_class msg_class, const char *data, gboolean local, const crm_node_t *node, enum crm_ais_msg_types dest) { static int msg_id = 0; static int local_pid = 0; static int local_name_len = 0; static const char *local_name = NULL; char *target = NULL; struct iovec *iov; pcmk__cpg_msg_t *msg = NULL; enum crm_ais_msg_types sender = text2msg_type(crm_system_name); switch (msg_class) { case crm_class_cluster: break; default: crm_err("Invalid message class: %d", msg_class); return FALSE; } CRM_CHECK(dest != crm_msg_ais, return FALSE); if (local_name == NULL) { local_name = get_local_node_name(); } if ((local_name_len == 0) && (local_name != NULL)) { local_name_len = strlen(local_name); } if (data == NULL) { data = ""; } if (local_pid == 0) { local_pid = getpid(); } if (sender == crm_msg_none) { sender = local_pid; } msg = calloc(1, sizeof(pcmk__cpg_msg_t)); msg_id++; msg->id = msg_id; msg->header.id = msg_class; msg->header.error = CS_OK; msg->host.type = dest; msg->host.local = local; if (node) { if (node->uname) { target = strdup(node->uname); msg->host.size = strlen(node->uname); memset(msg->host.uname, 0, MAX_NAME); memcpy(msg->host.uname, node->uname, msg->host.size); } else { target = crm_strdup_printf("%u", node->id); } msg->host.id = node->id; } else { target = strdup("all"); } msg->sender.id = 0; msg->sender.type = sender; msg->sender.pid = local_pid; msg->sender.size = local_name_len; memset(msg->sender.uname, 0, MAX_NAME); if ((local_name != NULL) && (msg->sender.size != 0)) { memcpy(msg->sender.uname, local_name, msg->sender.size); } msg->size = 1 + strlen(data); msg->header.size = sizeof(pcmk__cpg_msg_t) + msg->size; if (msg->size < CRM_BZ2_THRESHOLD) { msg = pcmk__realloc(msg, msg->header.size); memcpy(msg->data, data, msg->size); } else { char *compressed = NULL; unsigned int new_size = 0; char *uncompressed = strdup(data); if (pcmk__compress(uncompressed, (unsigned int) msg->size, 0, &compressed, &new_size) == pcmk_rc_ok) { msg->header.size = sizeof(pcmk__cpg_msg_t) + new_size; msg = pcmk__realloc(msg, msg->header.size); memcpy(msg->data, compressed, new_size); msg->is_compressed = TRUE; msg->compressed_size = new_size; } else { // cppcheck seems not to understand the abort logic in pcmk__realloc // cppcheck-suppress memleak msg = pcmk__realloc(msg, msg->header.size); memcpy(msg->data, data, msg->size); } free(uncompressed); free(compressed); } iov = calloc(1, sizeof(struct iovec)); iov->iov_base = msg; iov->iov_len = msg->header.size; if (msg->compressed_size) { crm_trace("Queueing CPG message %u to %s (%llu bytes, %d bytes compressed payload): %.200s", msg->id, target, (unsigned long long) iov->iov_len, msg->compressed_size, data); } else { crm_trace("Queueing CPG message %u to %s (%llu bytes, %d bytes payload): %.200s", msg->id, target, (unsigned long long) iov->iov_len, msg->size, data); } free(target); cs_message_queue = g_list_append(cs_message_queue, iov); crm_cs_flush(&pcmk_cpg_handle); return TRUE; } /*! * \brief Get the message type equivalent of a string * * \param[in] text String of message type * * \return Message type equivalent of \p text */ enum crm_ais_msg_types text2msg_type(const char *text) { int type = crm_msg_none; CRM_CHECK(text != NULL, return type); text = pcmk__message_name(text); if (pcmk__str_eq(text, "ais", pcmk__str_casei)) { type = crm_msg_ais; } else if (pcmk__str_eq(text, CRM_SYSTEM_CIB, pcmk__str_casei)) { type = crm_msg_cib; } else if (pcmk__strcase_any_of(text, CRM_SYSTEM_CRMD, CRM_SYSTEM_DC, NULL)) { type = crm_msg_crmd; } else if (pcmk__str_eq(text, CRM_SYSTEM_TENGINE, pcmk__str_casei)) { type = crm_msg_te; } else if (pcmk__str_eq(text, CRM_SYSTEM_PENGINE, pcmk__str_casei)) { type = crm_msg_pe; } else if (pcmk__str_eq(text, CRM_SYSTEM_LRMD, pcmk__str_casei)) { type = crm_msg_lrmd; } else if (pcmk__str_eq(text, CRM_SYSTEM_STONITHD, pcmk__str_casei)) { type = crm_msg_stonithd; } else if (pcmk__str_eq(text, "stonith-ng", pcmk__str_casei)) { type = crm_msg_stonith_ng; } else if (pcmk__str_eq(text, "attrd", pcmk__str_casei)) { type = crm_msg_attrd; } else { /* This will normally be a transient client rather than * a cluster daemon. Set the type to the pid of the client */ int scan_rc = sscanf(text, "%d", &type); if (scan_rc != 1 || type <= crm_msg_stonith_ng) { /* Ensure it's sane */ type = crm_msg_none; } } return type; } diff --git a/lib/common/digest.c b/lib/common/digest.c index 933769fde3..9a06ff44a7 100644 --- a/lib/common/digest.c +++ b/lib/common/digest.c @@ -1,336 +1,336 @@ /* * Copyright 2015-2024 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 "crmcommon_private.h" #define BEST_EFFORT_STATUS 0 /*! * \internal * \brief Dump XML in a format used with v1 digests * * \param[in] xml Root of XML to dump * * \return Newly allocated buffer containing dumped XML */ static GString * dump_xml_for_digest(xmlNodePtr xml) { GString *buffer = g_string_sized_new(1024); /* for compatibility with the old result which is used for v1 digests */ g_string_append_c(buffer, ' '); pcmk__xml_string(xml, 0, buffer, 0); g_string_append_c(buffer, '\n'); return buffer; } /*! * \brief Calculate and return v1 digest of XML tree * * \param[in] input Root of XML to digest * \param[in] sort Whether to sort the XML before calculating digest * \param[in] ignored Not used * * \return Newly allocated string containing digest * \note Example return value: "c048eae664dba840e1d2060f00299e9d" */ static char * calculate_xml_digest_v1(xmlNode *input, gboolean sort, gboolean ignored) { char *digest = NULL; GString *buffer = NULL; xmlNode *copy = NULL; if (sort) { crm_trace("Sorting xml..."); copy = sorted_xml(input, NULL, TRUE); crm_trace("Done"); input = copy; } buffer = dump_xml_for_digest(input); CRM_CHECK(buffer->len > 0, free_xml(copy); g_string_free(buffer, TRUE); return NULL); digest = crm_md5sum((const char *) buffer->str); crm_log_xml_trace(input, "digest:source"); g_string_free(buffer, TRUE); free_xml(copy); return digest; } /*! * \brief Calculate and return v2 digest of XML tree * * \param[in] source Root of XML to digest * \param[in] do_filter Whether to filter certain XML attributes * * \return Newly allocated string containing digest */ static char * calculate_xml_digest_v2(const xmlNode *source, gboolean do_filter) { char *digest = NULL; - gchar *buf = NULL; + GString *buf = g_string_sized_new(1024); crm_trace("Begin digest %s", do_filter?"filtered":""); - buf = pcmk__xml_dump(source, (do_filter? pcmk__xml_fmt_filtered : 0)); - digest = crm_md5sum(buf); + pcmk__xml_string(source, (do_filter? pcmk__xml_fmt_filtered : 0), buf, 0); + digest = crm_md5sum(buf->str); pcmk__if_tracing( { char *trace_file = crm_strdup_printf("%s/digest-%s", pcmk__get_tmpdir(), digest); crm_trace("Saving %s.%s.%s to %s", crm_element_value(source, PCMK_XA_ADMIN_EPOCH), crm_element_value(source, PCMK_XA_EPOCH), crm_element_value(source, PCMK_XA_NUM_UPDATES), trace_file); save_xml_to_file(source, "digest input", trace_file); free(trace_file); }, {} ); crm_trace("End digest"); - g_free(buf); + g_string_free(buf, TRUE); return digest; } /*! * \brief Calculate and return digest of XML tree, suitable for storing on disk * * \param[in] input Root of XML to digest * * \return Newly allocated string containing digest */ char * calculate_on_disk_digest(xmlNode *input) { /* Always use the v1 format for on-disk digests * a) it's a compatibility nightmare * b) we only use this once at startup, all other * invocations are in a separate child process */ return calculate_xml_digest_v1(input, FALSE, FALSE); } /*! * \brief Calculate and return digest of XML operation * * \param[in] input Root of XML to digest * \param[in] version Unused * * \return Newly allocated string containing digest */ char * calculate_operation_digest(xmlNode *input, const char *version) { /* We still need the sorting for operation digests */ return calculate_xml_digest_v1(input, TRUE, FALSE); } /*! * \brief Calculate and return digest of XML tree * * \param[in] input Root of XML to digest * \param[in] sort Whether to sort XML before calculating digest * \param[in] do_filter Whether to filter certain XML attributes * \param[in] version CRM feature set version (used to select v1/v2 digest) * * \return Newly allocated string containing digest */ char * calculate_xml_versioned_digest(xmlNode *input, gboolean sort, gboolean do_filter, const char *version) { /* * @COMPAT digests (on-disk or in diffs/patchsets) created <1.1.4; * removing this affects even full-restart upgrades from old versions * * The sorting associated with v1 digest creation accounted for 23% of * the CIB manager's CPU usage on the server. v2 drops this. * * The filtering accounts for an additional 2.5% and we may want to * remove it in future. * * v2 also uses the xmlBuffer contents directly to avoid additional copying */ if (version == NULL || compare_version("3.0.5", version) > 0) { crm_trace("Using v1 digest algorithm for %s", pcmk__s(version, "unknown feature set")); return calculate_xml_digest_v1(input, sort, do_filter); } crm_trace("Using v2 digest algorithm for %s", pcmk__s(version, "unknown feature set")); return calculate_xml_digest_v2(input, do_filter); } /*! * \internal * \brief Check whether calculated digest of given XML matches expected digest * * \param[in] input Root of XML tree to digest * \param[in] expected Expected digest in on-disk format * * \return true if digests match, false on mismatch or error */ bool pcmk__verify_digest(xmlNode *input, const char *expected) { char *calculated = NULL; bool passed; if (input != NULL) { calculated = calculate_on_disk_digest(input); if (calculated == NULL) { crm_perror(LOG_ERR, "Could not calculate digest for comparison"); return false; } } passed = pcmk__str_eq(expected, calculated, pcmk__str_casei); if (passed) { crm_trace("Digest comparison passed: %s", calculated); } else { crm_err("Digest comparison failed: expected %s, calculated %s", expected, calculated); } free(calculated); return passed; } /*! * \internal * \brief Check whether an XML attribute should be excluded from CIB digests * * \param[in] name XML attribute name * * \return true if XML attribute should be excluded from CIB digest calculation */ bool pcmk__xa_filterable(const char *name) { static const char *filter[] = { PCMK_XA_CRM_DEBUG_ORIGIN, PCMK_XA_CIB_LAST_WRITTEN, PCMK_XA_UPDATE_ORIGIN, PCMK_XA_UPDATE_CLIENT, PCMK_XA_UPDATE_USER, }; for (int i = 0; i < PCMK__NELEM(filter); i++) { if (strcmp(name, filter[i]) == 0) { return true; } } return false; } char * crm_md5sum(const char *buffer) { int lpc = 0, len = 0; char *digest = NULL; unsigned char raw_digest[MD5_DIGEST_SIZE]; if (buffer == NULL) { buffer = ""; } len = strlen(buffer); crm_trace("Beginning digest of %d bytes", len); digest = malloc(2 * MD5_DIGEST_SIZE + 1); if (digest) { md5_buffer(buffer, len, raw_digest); for (lpc = 0; lpc < MD5_DIGEST_SIZE; lpc++) { sprintf(digest + (2 * lpc), "%02x", raw_digest[lpc]); } digest[(2 * MD5_DIGEST_SIZE)] = 0; crm_trace("Digest %s.", digest); } else { crm_err("Could not create digest"); } return digest; } // Return true if a is an attribute that should be filtered static bool should_filter_for_digest(xmlAttrPtr a, void *user_data) { if (strncmp((const char *) a->name, CRM_META "_", sizeof(CRM_META " ") - 1) == 0) { return true; } return pcmk__str_any_of((const char *) a->name, PCMK_XA_ID, PCMK_XA_CRM_FEATURE_SET, PCMK__XA_OP_DIGEST, PCMK__META_ON_NODE, PCMK__META_ON_NODE_UUID, "pcmk_external_ip", NULL); } /*! * \internal * \brief Remove XML attributes not needed for operation digest * * \param[in,out] param_set XML with operation parameters */ void pcmk__filter_op_for_digest(xmlNode *param_set) { char *key = NULL; char *timeout = NULL; guint interval_ms = 0; if (param_set == NULL) { return; } /* Timeout is useful for recurring operation digests, so grab it before * removing meta-attributes */ key = crm_meta_name(PCMK_META_INTERVAL); if (crm_element_value_ms(param_set, key, &interval_ms) != pcmk_ok) { interval_ms = 0; } free(key); key = NULL; if (interval_ms != 0) { key = crm_meta_name(PCMK_META_TIMEOUT); timeout = crm_element_value_copy(param_set, key); } // Remove all CRM_meta_* attributes and certain other attributes pcmk__xe_remove_matching_attrs(param_set, should_filter_for_digest, NULL); // Add timeout back for recurring operation digests if (timeout != NULL) { crm_xml_add(param_set, key, timeout); } free(timeout); free(key); } diff --git a/lib/common/ipc_server.c b/lib/common/ipc_server.c index c62e663c33..237e15fac1 100644 --- a/lib/common/ipc_server.c +++ b/lib/common/ipc_server.c @@ -1,1017 +1,1023 @@ /* * 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 "crmcommon_private.h" /* Evict clients whose event queue grows this large (by default) */ #define PCMK_IPC_DEFAULT_QUEUE_MAX 500 static GHashTable *client_connections = NULL; /*! * \internal * \brief Count IPC clients * * \return Number of active IPC client connections */ guint pcmk__ipc_client_count(void) { return client_connections? g_hash_table_size(client_connections) : 0; } /*! * \internal * \brief Execute a function for each active IPC client connection * * \param[in] func Function to call * \param[in,out] user_data Pointer to pass to function * * \note The parameters are the same as for g_hash_table_foreach(). */ void pcmk__foreach_ipc_client(GHFunc func, gpointer user_data) { if ((func != NULL) && (client_connections != NULL)) { g_hash_table_foreach(client_connections, func, user_data); } } pcmk__client_t * pcmk__find_client(const qb_ipcs_connection_t *c) { if (client_connections) { return g_hash_table_lookup(client_connections, c); } crm_trace("No client found for %p", c); return NULL; } pcmk__client_t * pcmk__find_client_by_id(const char *id) { if ((client_connections != NULL) && (id != NULL)) { gpointer key; pcmk__client_t *client = NULL; GHashTableIter iter; g_hash_table_iter_init(&iter, client_connections); while (g_hash_table_iter_next(&iter, &key, (gpointer *) & client)) { if (strcmp(client->id, id) == 0) { return client; } } } crm_trace("No client found with id='%s'", pcmk__s(id, "")); return NULL; } /*! * \internal * \brief Get a client identifier for use in log messages * * \param[in] c Client * * \return Client's name, client's ID, or a string literal, as available * \note This is intended to be used in format strings like "client %s". */ const char * pcmk__client_name(const pcmk__client_t *c) { if (c == NULL) { return "(unspecified)"; } else if (c->name != NULL) { return c->name; } else if (c->id != NULL) { return c->id; } else { return "(unidentified)"; } } void pcmk__client_cleanup(void) { if (client_connections != NULL) { int active = g_hash_table_size(client_connections); if (active > 0) { crm_warn("Exiting with %d active IPC client%s", active, pcmk__plural_s(active)); } g_hash_table_destroy(client_connections); client_connections = NULL; } } void pcmk__drop_all_clients(qb_ipcs_service_t *service) { qb_ipcs_connection_t *c = NULL; if (service == NULL) { return; } c = qb_ipcs_connection_first_get(service); while (c != NULL) { qb_ipcs_connection_t *last = c; c = qb_ipcs_connection_next_get(service, last); /* There really shouldn't be anyone connected at this point */ crm_notice("Disconnecting client %p, pid=%d...", last, pcmk__client_pid(last)); qb_ipcs_disconnect(last); qb_ipcs_connection_unref(last); } } /*! * \internal * \brief Allocate a new pcmk__client_t object based on an IPC connection * * \param[in] c IPC connection (NULL to allocate generic client) * \param[in] key Connection table key (NULL to use sane default) * \param[in] uid_client UID corresponding to c (ignored if c is NULL) * * \return Pointer to new pcmk__client_t (or NULL on error) */ static pcmk__client_t * client_from_connection(qb_ipcs_connection_t *c, void *key, uid_t uid_client) { pcmk__client_t *client = calloc(1, sizeof(pcmk__client_t)); if (client == NULL) { crm_perror(LOG_ERR, "Allocating client"); return NULL; } if (c) { client->user = pcmk__uid2username(uid_client); if (client->user == NULL) { client->user = strdup("#unprivileged"); CRM_CHECK(client->user != NULL, free(client); return NULL); crm_err("Unable to enforce ACLs for user ID %d, assuming unprivileged", uid_client); } client->ipcs = c; pcmk__set_client_flags(client, pcmk__client_ipc); client->pid = pcmk__client_pid(c); if (key == NULL) { key = c; } } client->id = crm_generate_uuid(); if (key == NULL) { key = client->id; } if (client_connections == NULL) { crm_trace("Creating IPC client table"); client_connections = g_hash_table_new(g_direct_hash, g_direct_equal); } g_hash_table_insert(client_connections, key, client); return client; } /*! * \brief Allocate a new pcmk__client_t object and generate its ID * * \param[in] key What to use as connections hash table key (NULL to use ID) * * \return Pointer to new pcmk__client_t (asserts on failure) */ pcmk__client_t * pcmk__new_unauth_client(void *key) { pcmk__client_t *client = client_from_connection(NULL, key, 0); CRM_ASSERT(client != NULL); return client; } pcmk__client_t * pcmk__new_client(qb_ipcs_connection_t *c, uid_t uid_client, gid_t gid_client) { gid_t uid_cluster = 0; gid_t gid_cluster = 0; pcmk__client_t *client = NULL; CRM_CHECK(c != NULL, return NULL); if (pcmk_daemon_user(&uid_cluster, &gid_cluster) < 0) { static bool need_log = TRUE; if (need_log) { crm_warn("Could not find user and group IDs for user %s", CRM_DAEMON_USER); need_log = FALSE; } } if (uid_client != 0) { crm_trace("Giving group %u access to new IPC connection", gid_cluster); /* Passing -1 to chown(2) means don't change */ qb_ipcs_connection_auth_set(c, -1, gid_cluster, S_IRUSR | S_IWUSR | S_IRGRP | S_IWGRP); } /* TODO: Do our own auth checking, return NULL if unauthorized */ client = client_from_connection(c, NULL, uid_client); if (client == NULL) { return NULL; } if ((uid_client == 0) || (uid_client == uid_cluster)) { /* Remember when a connection came from root or hacluster */ pcmk__set_client_flags(client, pcmk__client_privileged); } crm_debug("New IPC client %s for PID %u with uid %d and gid %d", client->id, client->pid, uid_client, gid_client); return client; } static struct iovec * pcmk__new_ipc_event(void) { struct iovec *iov = calloc(2, sizeof(struct iovec)); CRM_ASSERT(iov != NULL); return iov; } /*! * \brief Free an I/O vector created by pcmk__ipc_prepare_iov() * * \param[in,out] event I/O vector to free */ void pcmk_free_ipc_event(struct iovec *event) { if (event != NULL) { free(event[0].iov_base); free(event[1].iov_base); free(event); } } static void free_event(gpointer data) { pcmk_free_ipc_event((struct iovec *) data); } static void add_event(pcmk__client_t *c, struct iovec *iov) { if (c->event_queue == NULL) { c->event_queue = g_queue_new(); } g_queue_push_tail(c->event_queue, iov); } void pcmk__free_client(pcmk__client_t *c) { if (c == NULL) { return; } if (client_connections) { if (c->ipcs) { crm_trace("Destroying %p/%p (%d remaining)", c, c->ipcs, g_hash_table_size(client_connections) - 1); g_hash_table_remove(client_connections, c->ipcs); } else { crm_trace("Destroying remote connection %p (%d remaining)", c, g_hash_table_size(client_connections) - 1); g_hash_table_remove(client_connections, c->id); } } if (c->event_timer) { g_source_remove(c->event_timer); } if (c->event_queue) { crm_debug("Destroying %d events", g_queue_get_length(c->event_queue)); g_queue_free_full(c->event_queue, free_event); } free(c->id); free(c->name); free(c->user); if (c->remote) { if (c->remote->auth_timeout) { g_source_remove(c->remote->auth_timeout); } free(c->remote->buffer); free(c->remote); } free(c); } /*! * \internal * \brief Raise IPC eviction threshold for a client, if allowed * * \param[in,out] client Client to modify * \param[in] qmax New threshold (as non-NULL string) * * \return true if change was allowed, false otherwise */ bool pcmk__set_client_queue_max(pcmk__client_t *client, const char *qmax) { if (pcmk_is_set(client->flags, pcmk__client_privileged)) { long long qmax_ll; if ((pcmk__scan_ll(qmax, &qmax_ll, 0LL) == pcmk_rc_ok) && (qmax_ll > 0LL) && (qmax_ll <= UINT_MAX)) { client->queue_max = (unsigned int) qmax_ll; return true; } } return false; } int pcmk__client_pid(qb_ipcs_connection_t *c) { struct qb_ipcs_connection_stats stats; stats.client_pid = 0; qb_ipcs_connection_stats_get(c, &stats, 0); return stats.client_pid; } /*! * \internal * \brief Retrieve message XML from data read from client IPC * * \param[in,out] c IPC client connection * \param[in] data Data read from client connection * \param[out] id Where to store message ID from libqb header * \param[out] flags Where to store flags from libqb header * * \return Message XML on success, NULL otherwise */ xmlNode * pcmk__client_data2xml(pcmk__client_t *c, void *data, uint32_t *id, uint32_t *flags) { xmlNode *xml = NULL; char *uncompressed = NULL; char *text = ((char *)data) + sizeof(pcmk__ipc_header_t); pcmk__ipc_header_t *header = data; if (!pcmk__valid_ipc_header(header)) { return NULL; } if (id) { *id = ((struct qb_ipc_response_header *)data)->id; } if (flags) { *flags = header->flags; } if (pcmk_is_set(header->flags, crm_ipc_proxied)) { /* Mark this client as being the endpoint of a proxy connection. * Proxy connections responses are sent on the event channel, to avoid * blocking the controller serving as proxy. */ pcmk__set_client_flags(c, pcmk__client_proxied); } if (header->size_compressed) { int rc = 0; unsigned int size_u = 1 + header->size_uncompressed; uncompressed = calloc(1, size_u); crm_trace("Decompressing message data %u bytes into %u bytes", header->size_compressed, size_u); rc = BZ2_bzBuffToBuffDecompress(uncompressed, &size_u, text, header->size_compressed, 1, 0); text = uncompressed; rc = pcmk__bzlib2rc(rc); if (rc != pcmk_rc_ok) { crm_err("Decompression failed: %s " CRM_XS " rc=%d", pcmk_rc_str(rc), rc); free(uncompressed); return NULL; } } CRM_ASSERT(text[header->size_uncompressed - 1] == 0); xml = pcmk__xml_parse(text); crm_log_xml_trace(xml, "[IPC received]"); free(uncompressed); return xml; } static int crm_ipcs_flush_events(pcmk__client_t *c); static gboolean crm_ipcs_flush_events_cb(gpointer data) { pcmk__client_t *c = data; c->event_timer = 0; crm_ipcs_flush_events(c); return FALSE; } /*! * \internal * \brief Add progressive delay before next event queue flush * * \param[in,out] c Client connection to add delay to * \param[in] queue_len Current event queue length */ static inline void delay_next_flush(pcmk__client_t *c, unsigned int queue_len) { /* Delay a maximum of 1.5 seconds */ guint delay = (queue_len < 5)? (1000 + 100 * queue_len) : 1500; c->event_timer = g_timeout_add(delay, crm_ipcs_flush_events_cb, c); } /*! * \internal * \brief Send client any messages in its queue * * \param[in,out] c Client to flush * * \return Standard Pacemaker return value */ static int crm_ipcs_flush_events(pcmk__client_t *c) { int rc = pcmk_rc_ok; ssize_t qb_rc = 0; unsigned int sent = 0; unsigned int queue_len = 0; if (c == NULL) { return rc; } else if (c->event_timer) { /* There is already a timer, wait until it goes off */ crm_trace("Timer active for %p - %d", c->ipcs, c->event_timer); return rc; } if (c->event_queue) { queue_len = g_queue_get_length(c->event_queue); } while (sent < 100) { pcmk__ipc_header_t *header = NULL; struct iovec *event = NULL; if (c->event_queue) { // We don't pop unless send is successful event = g_queue_peek_head(c->event_queue); } if (event == NULL) { // Queue is empty break; } qb_rc = qb_ipcs_event_sendv(c->ipcs, event, 2); if (qb_rc < 0) { rc = (int) -qb_rc; break; } event = g_queue_pop_head(c->event_queue); sent++; header = event[0].iov_base; if (header->size_compressed) { crm_trace("Event %d to %p[%d] (%lld compressed bytes) sent", header->qb.id, c->ipcs, c->pid, (long long) qb_rc); } else { crm_trace("Event %d to %p[%d] (%lld bytes) sent: %.120s", header->qb.id, c->ipcs, c->pid, (long long) qb_rc, (char *) (event[1].iov_base)); } pcmk_free_ipc_event(event); } queue_len -= sent; if (sent > 0 || queue_len) { crm_trace("Sent %d events (%d remaining) for %p[%d]: %s (%lld)", sent, queue_len, c->ipcs, c->pid, pcmk_rc_str(rc), (long long) qb_rc); } if (queue_len) { /* Allow clients to briefly fall behind on processing incoming messages, * but drop completely unresponsive clients so the connection doesn't * consume resources indefinitely. */ if (queue_len > QB_MAX(c->queue_max, PCMK_IPC_DEFAULT_QUEUE_MAX)) { if ((c->queue_backlog <= 1) || (queue_len < c->queue_backlog)) { /* Don't evict for a new or shrinking backlog */ crm_warn("Client with process ID %u has a backlog of %u messages " CRM_XS " %p", c->pid, queue_len, c->ipcs); } else { crm_err("Evicting client with process ID %u due to backlog of %u messages " CRM_XS " %p", c->pid, queue_len, c->ipcs); c->queue_backlog = 0; qb_ipcs_disconnect(c->ipcs); return rc; } } c->queue_backlog = queue_len; delay_next_flush(c, queue_len); } else { /* Event queue is empty, there is no backlog */ c->queue_backlog = 0; } return rc; } /*! * \internal * \brief Create an I/O vector for sending an IPC XML message * * \param[in] request Identifier for libqb response header * \param[in] message XML message to send * \param[in] max_send_size If 0, default IPC buffer size is used * \param[out] result Where to store prepared I/O vector * \param[out] bytes Size of prepared data in bytes * * \return Standard Pacemaker return code */ int pcmk__ipc_prepare_iov(uint32_t request, const xmlNode *message, uint32_t max_send_size, struct iovec **result, ssize_t *bytes) { - static unsigned int biggest = 0; struct iovec *iov; unsigned int total = 0; - char *compressed = NULL; - char *buffer = NULL; - gchar *g_buffer = NULL; + GString *buffer = NULL; pcmk__ipc_header_t *header = NULL; + int rc = pcmk_rc_ok; if ((message == NULL) || (result == NULL)) { - return EINVAL; + rc = EINVAL; + goto done; } header = calloc(1, sizeof(pcmk__ipc_header_t)); if (header == NULL) { - return ENOMEM; /* errno mightn't be set by allocator */ + rc = ENOMEM; + goto done; } - g_buffer = pcmk__xml_dump(message, 0); - pcmk__str_update(&buffer, g_buffer); - g_free(g_buffer); + buffer = g_string_sized_new(1024); + pcmk__xml_string(message, 0, buffer, 0); if (max_send_size == 0) { max_send_size = crm_ipc_default_buffer_size(); } CRM_LOG_ASSERT(max_send_size != 0); *result = NULL; iov = pcmk__new_ipc_event(); iov[0].iov_len = sizeof(pcmk__ipc_header_t); iov[0].iov_base = header; header->version = PCMK__IPC_VERSION; - header->size_uncompressed = 1 + strlen(buffer); + header->size_uncompressed = 1 + buffer->len; total = iov[0].iov_len + header->size_uncompressed; if (total < max_send_size) { - iov[1].iov_base = buffer; + pcmk__str_update((char **) &(iov[1].iov_base), buffer->str); iov[1].iov_len = header->size_uncompressed; } else { + static unsigned int biggest = 0; + + char *compressed = NULL; unsigned int new_size = 0; - if (pcmk__compress(buffer, (unsigned int) header->size_uncompressed, + if (pcmk__compress(buffer->str, + (unsigned int) header->size_uncompressed, (unsigned int) max_send_size, &compressed, &new_size) == pcmk_rc_ok) { pcmk__set_ipc_flags(header->flags, "send data", crm_ipc_compressed); header->size_compressed = new_size; iov[1].iov_len = header->size_compressed; iov[1].iov_base = compressed; - free(buffer); - biggest = QB_MAX(header->size_compressed, biggest); } else { crm_log_xml_trace(message, "EMSGSIZE"); biggest = QB_MAX(header->size_uncompressed, biggest); crm_err("Could not compress %u-byte message into less than IPC " "limit of %u bytes; set PCMK_ipc_buffer to higher value " "(%u bytes suggested)", header->size_uncompressed, max_send_size, 4 * biggest); free(compressed); - free(buffer); pcmk_free_ipc_event(iov); - return EMSGSIZE; + rc = EMSGSIZE; + goto done; } } header->qb.size = iov[0].iov_len + iov[1].iov_len; header->qb.id = (int32_t)request; /* Replying to a specific request */ *result = iov; CRM_ASSERT(header->qb.size > 0); if (bytes != NULL) { *bytes = header->qb.size; } - return pcmk_rc_ok; + +done: + if (buffer != NULL) { + g_string_free(buffer, TRUE); + } + return rc; } int pcmk__ipc_send_iov(pcmk__client_t *c, struct iovec *iov, uint32_t flags) { int rc = pcmk_rc_ok; static uint32_t id = 1; pcmk__ipc_header_t *header = iov[0].iov_base; if (c->flags & pcmk__client_proxied) { /* _ALL_ replies to proxied connections need to be sent as events */ if (!pcmk_is_set(flags, crm_ipc_server_event)) { /* The proxied flag lets us know this was originally meant to be a * response, even though we're sending it over the event channel. */ pcmk__set_ipc_flags(flags, "server event", crm_ipc_server_event |crm_ipc_proxied_relay_response); } } pcmk__set_ipc_flags(header->flags, "server event", flags); if (flags & crm_ipc_server_event) { header->qb.id = id++; /* We don't really use it, but doesn't hurt to set one */ if (flags & crm_ipc_server_free) { crm_trace("Sending the original to %p[%d]", c->ipcs, c->pid); add_event(c, iov); } else { struct iovec *iov_copy = pcmk__new_ipc_event(); crm_trace("Sending a copy to %p[%d]", c->ipcs, c->pid); iov_copy[0].iov_len = iov[0].iov_len; iov_copy[0].iov_base = malloc(iov[0].iov_len); memcpy(iov_copy[0].iov_base, iov[0].iov_base, iov[0].iov_len); iov_copy[1].iov_len = iov[1].iov_len; iov_copy[1].iov_base = malloc(iov[1].iov_len); memcpy(iov_copy[1].iov_base, iov[1].iov_base, iov[1].iov_len); add_event(c, iov_copy); } } else { ssize_t qb_rc; CRM_LOG_ASSERT(header->qb.id != 0); /* Replying to a specific request */ qb_rc = qb_ipcs_response_sendv(c->ipcs, iov, 2); if (qb_rc < header->qb.size) { if (qb_rc < 0) { rc = (int) -qb_rc; } crm_notice("Response %d to pid %d failed: %s " CRM_XS " bytes=%u rc=%lld ipcs=%p", header->qb.id, c->pid, pcmk_rc_str(rc), header->qb.size, (long long) qb_rc, c->ipcs); } else { crm_trace("Response %d sent, %lld bytes to %p[%d]", header->qb.id, (long long) qb_rc, c->ipcs, c->pid); } if (flags & crm_ipc_server_free) { pcmk_free_ipc_event(iov); } } if (flags & crm_ipc_server_event) { rc = crm_ipcs_flush_events(c); } else { crm_ipcs_flush_events(c); } if ((rc == EPIPE) || (rc == ENOTCONN)) { crm_trace("Client %p disconnected", c->ipcs); } return rc; } int pcmk__ipc_send_xml(pcmk__client_t *c, uint32_t request, const xmlNode *message, uint32_t flags) { struct iovec *iov = NULL; int rc = pcmk_rc_ok; if (c == NULL) { return EINVAL; } rc = pcmk__ipc_prepare_iov(request, message, crm_ipc_default_buffer_size(), &iov, NULL); if (rc == pcmk_rc_ok) { pcmk__set_ipc_flags(flags, "send data", crm_ipc_server_free); rc = pcmk__ipc_send_iov(c, iov, flags); } else { pcmk_free_ipc_event(iov); crm_notice("IPC message to pid %d failed: %s " CRM_XS " rc=%d", c->pid, pcmk_rc_str(rc), rc); } return rc; } /*! * \internal * \brief Create an acknowledgement with a status code to send to a client * * \param[in] function Calling function * \param[in] line Source file line within calling function * \param[in] flags IPC flags to use when sending * \param[in] tag Element name to use for acknowledgement * \param[in] ver IPC protocol version (can be NULL) * \param[in] status Exit status code to add to ack * * \return Newly created XML for ack * \note The caller is responsible for freeing the return value with free_xml(). */ xmlNode * pcmk__ipc_create_ack_as(const char *function, int line, uint32_t flags, const char *tag, const char *ver, crm_exit_t status) { xmlNode *ack = NULL; if (pcmk_is_set(flags, crm_ipc_client_response)) { ack = create_xml_node(NULL, tag); crm_xml_add(ack, PCMK_XA_FUNCTION, function); crm_xml_add_int(ack, PCMK__XA_LINE, line); crm_xml_add_int(ack, PCMK_XA_STATUS, (int) status); crm_xml_add(ack, PCMK__XA_IPC_PROTO_VERSION, ver); } return ack; } /*! * \internal * \brief Send an acknowledgement with a status code to a client * * \param[in] function Calling function * \param[in] line Source file line within calling function * \param[in] c Client to send ack to * \param[in] request Request ID being replied to * \param[in] flags IPC flags to use when sending * \param[in] tag Element name to use for acknowledgement * \param[in] ver IPC protocol version (can be NULL) * \param[in] status Status code to send with acknowledgement * * \return Standard Pacemaker return code */ int pcmk__ipc_send_ack_as(const char *function, int line, pcmk__client_t *c, uint32_t request, uint32_t flags, const char *tag, const char *ver, crm_exit_t status) { int rc = pcmk_rc_ok; xmlNode *ack = pcmk__ipc_create_ack_as(function, line, flags, tag, ver, status); if (ack != NULL) { crm_trace("Ack'ing IPC message from client %s as <%s status=%d>", pcmk__client_name(c), tag, status); crm_log_xml_trace(ack, "sent-ack"); c->request_id = 0; rc = pcmk__ipc_send_xml(c, request, ack, flags); free_xml(ack); } return rc; } /*! * \internal * \brief Add an IPC server to the main loop for the pacemaker-based API * * \param[out] ipcs_ro New IPC server for read-only pacemaker-based API * \param[out] ipcs_rw New IPC server for read/write pacemaker-based API * \param[out] ipcs_shm New IPC server for shared-memory pacemaker-based API * \param[in] ro_cb IPC callbacks for read-only API * \param[in] rw_cb IPC callbacks for read/write and shared-memory APIs * * \note This function exits fatally if unable to create the servers. */ void pcmk__serve_based_ipc(qb_ipcs_service_t **ipcs_ro, qb_ipcs_service_t **ipcs_rw, qb_ipcs_service_t **ipcs_shm, struct qb_ipcs_service_handlers *ro_cb, struct qb_ipcs_service_handlers *rw_cb) { *ipcs_ro = mainloop_add_ipc_server(PCMK__SERVER_BASED_RO, QB_IPC_NATIVE, ro_cb); *ipcs_rw = mainloop_add_ipc_server(PCMK__SERVER_BASED_RW, QB_IPC_NATIVE, rw_cb); *ipcs_shm = mainloop_add_ipc_server(PCMK__SERVER_BASED_SHM, QB_IPC_SHM, rw_cb); if (*ipcs_ro == NULL || *ipcs_rw == NULL || *ipcs_shm == NULL) { crm_err("Failed to create the CIB manager: exiting and inhibiting respawn"); crm_warn("Verify pacemaker and pacemaker_remote are not both enabled"); crm_exit(CRM_EX_FATAL); } } /*! * \internal * \brief Destroy IPC servers for pacemaker-based API * * \param[out] ipcs_ro IPC server for read-only pacemaker-based API * \param[out] ipcs_rw IPC server for read/write pacemaker-based API * \param[out] ipcs_shm IPC server for shared-memory pacemaker-based API * * \note This is a convenience function for calling qb_ipcs_destroy() for each * argument. */ void pcmk__stop_based_ipc(qb_ipcs_service_t *ipcs_ro, qb_ipcs_service_t *ipcs_rw, qb_ipcs_service_t *ipcs_shm) { qb_ipcs_destroy(ipcs_ro); qb_ipcs_destroy(ipcs_rw); qb_ipcs_destroy(ipcs_shm); } /*! * \internal * \brief Add an IPC server to the main loop for the pacemaker-controld API * * \param[in] cb IPC callbacks * * \return Newly created IPC server */ qb_ipcs_service_t * pcmk__serve_controld_ipc(struct qb_ipcs_service_handlers *cb) { return mainloop_add_ipc_server(CRM_SYSTEM_CRMD, QB_IPC_NATIVE, cb); } /*! * \internal * \brief Add an IPC server to the main loop for the pacemaker-attrd API * * \param[out] ipcs Where to store newly created IPC server * \param[in] cb IPC callbacks * * \note This function exits fatally if unable to create the servers. */ void pcmk__serve_attrd_ipc(qb_ipcs_service_t **ipcs, struct qb_ipcs_service_handlers *cb) { *ipcs = mainloop_add_ipc_server(PCMK__VALUE_ATTRD, QB_IPC_NATIVE, cb); if (*ipcs == NULL) { crm_err("Failed to create pacemaker-attrd server: exiting and inhibiting respawn"); crm_warn("Verify pacemaker and pacemaker_remote are not both enabled."); crm_exit(CRM_EX_FATAL); } } /*! * \internal * \brief Add an IPC server to the main loop for the pacemaker-fenced API * * \param[out] ipcs Where to store newly created IPC server * \param[in] cb IPC callbacks * * \note This function exits fatally if unable to create the servers. */ void pcmk__serve_fenced_ipc(qb_ipcs_service_t **ipcs, struct qb_ipcs_service_handlers *cb) { *ipcs = mainloop_add_ipc_server_with_prio("stonith-ng", QB_IPC_NATIVE, cb, QB_LOOP_HIGH); if (*ipcs == NULL) { crm_err("Failed to create fencer: exiting and inhibiting respawn."); crm_warn("Verify pacemaker and pacemaker_remote are not both enabled."); crm_exit(CRM_EX_FATAL); } } /*! * \internal * \brief Add an IPC server to the main loop for the pacemakerd API * * \param[out] ipcs Where to store newly created IPC server * \param[in] cb IPC callbacks * * \note This function exits with CRM_EX_OSERR if unable to create the servers. */ void pcmk__serve_pacemakerd_ipc(qb_ipcs_service_t **ipcs, struct qb_ipcs_service_handlers *cb) { *ipcs = mainloop_add_ipc_server(CRM_SYSTEM_MCP, QB_IPC_NATIVE, cb); if (*ipcs == NULL) { crm_err("Couldn't start pacemakerd IPC server"); crm_warn("Verify pacemaker and pacemaker_remote are not both enabled."); /* sub-daemons are observed by pacemakerd. Thus we exit CRM_EX_FATAL * if we want to prevent pacemakerd from restarting them. * With pacemakerd we leave the exit-code shown to e.g. systemd * to what it was prior to moving the code here from pacemakerd.c */ crm_exit(CRM_EX_OSERR); } } /*! * \internal * \brief Add an IPC server to the main loop for the pacemaker-schedulerd API * * \param[in] cb IPC callbacks * * \return Newly created IPC server * \note This function exits fatally if unable to create the servers. */ qb_ipcs_service_t * pcmk__serve_schedulerd_ipc(struct qb_ipcs_service_handlers *cb) { return mainloop_add_ipc_server(CRM_SYSTEM_PENGINE, QB_IPC_NATIVE, cb); } /*! * \brief Check whether string represents a client name used by cluster daemons * * \param[in] name String to check * * \return true if name is standard client name used by daemons, false otherwise * * \note This is provided by the client, and so cannot be used by itself as a * secure means of authentication. */ bool crm_is_daemon_name(const char *name) { return pcmk__str_any_of(pcmk__message_name(name), "attrd", CRM_SYSTEM_CIB, CRM_SYSTEM_CRMD, CRM_SYSTEM_DC, CRM_SYSTEM_LRMD, CRM_SYSTEM_MCP, CRM_SYSTEM_PENGINE, CRM_SYSTEM_STONITHD, CRM_SYSTEM_TENGINE, "pacemaker-remoted", "stonith-ng", NULL); } diff --git a/lib/common/options.c b/lib/common/options.c index 2b3adbe2bc..24285362bb 100644 --- a/lib/common/options.c +++ b/lib/common/options.c @@ -1,1012 +1,1014 @@ /* * 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 _GNU_SOURCE # define _GNU_SOURCE #endif #include #include #include #include #include #include #include #include void pcmk__cli_help(char cmd) { if (cmd == 'v' || cmd == '$') { printf("Pacemaker %s\n", PACEMAKER_VERSION); printf("Written by Andrew Beekhof and " "the Pacemaker project contributors\n"); } else if (cmd == '!') { printf("Pacemaker %s (Build: %s): %s\n", PACEMAKER_VERSION, BUILD_VERSION, CRM_FEATURES); } crm_exit(CRM_EX_OK); while(1); // above does not return } /* * Option metadata */ static pcmk__cluster_option_t cluster_options[] = { /* name, old name, type, allowed values, * default value, validator, * flags, * short description, * long description */ { PCMK_OPT_DC_VERSION, NULL, "string", NULL, NULL, NULL, pcmk__opt_controld|pcmk__opt_generated, N_("Pacemaker version on cluster node elected Designated Controller " "(DC)"), N_("Includes a hash which identifies the exact revision the code was " "built from. Used for diagnostic purposes."), }, { PCMK_OPT_CLUSTER_INFRASTRUCTURE, NULL, "string", NULL, NULL, NULL, pcmk__opt_controld|pcmk__opt_generated, N_("The messaging layer on which Pacemaker is currently running"), N_("Used for informational and diagnostic purposes."), }, { PCMK_OPT_CLUSTER_NAME, NULL, "string", NULL, NULL, NULL, pcmk__opt_controld, N_("An arbitrary name for the cluster"), N_("This optional value is mostly for users' convenience as desired " "in administration, but may also be used in Pacemaker " "configuration rules via the #cluster-name node attribute, and " "by higher-level tools and resource agents."), }, { PCMK_OPT_DC_DEADTIME, NULL, "time", NULL, "20s", pcmk__valid_interval_spec, pcmk__opt_controld, N_("How long to wait for a response from other nodes during start-up"), N_("The optimal value will depend on the speed and load of your " "network and the type of switches used."), }, { PCMK_OPT_CLUSTER_RECHECK_INTERVAL, NULL, "time", NULL, "15min", pcmk__valid_interval_spec, pcmk__opt_controld, N_("Polling interval to recheck cluster state and evaluate rules " "with date specifications"), N_("Pacemaker is primarily event-driven, and looks ahead to know when " "to recheck cluster state for failure-timeout settings and most " "time-based rules. However, it will also recheck the cluster after " "this amount of inactivity, to evaluate rules with date " "specifications and serve as a fail-safe for certain types of " "scheduler bugs. A value of 0 disables polling. A positive value " "sets an interval in seconds, unless other units are specified " "(for example, \"5min\")."), }, { PCMK_OPT_FENCE_REACTION, NULL, "select", PCMK_VALUE_STOP ", " PCMK_VALUE_PANIC, PCMK_VALUE_STOP, NULL, pcmk__opt_controld, N_("How a cluster node should react if notified of its own fencing"), N_("A cluster node may receive notification of a \"succeeded\" " "fencing that targeted it if fencing is misconfigured, or if " "fabric fencing is in use that doesn't cut cluster communication. " "Use \"stop\" to attempt to immediately stop Pacemaker and stay " "stopped, or \"panic\" to attempt to immediately reboot the local " "node, falling back to stop on failure."), }, { PCMK_OPT_ELECTION_TIMEOUT, NULL, "time", NULL, "2min", pcmk__valid_interval_spec, pcmk__opt_controld|pcmk__opt_advanced, N_("Declare an election failed if it is not decided within this much " "time. If you need to adjust this value, it probably indicates " "the presence of a bug."), NULL, }, { PCMK_OPT_SHUTDOWN_ESCALATION, NULL, "time", NULL, "20min", pcmk__valid_interval_spec, pcmk__opt_controld|pcmk__opt_advanced, N_("Exit immediately if shutdown does not complete within this much " "time. If you need to adjust this value, it probably indicates " "the presence of a bug."), NULL, }, { PCMK_OPT_JOIN_INTEGRATION_TIMEOUT, "crmd-integration-timeout", "time", NULL, "3min", pcmk__valid_interval_spec, pcmk__opt_controld|pcmk__opt_advanced, N_("If you need to adjust this value, it probably indicates " "the presence of a bug."), NULL, }, { PCMK_OPT_JOIN_FINALIZATION_TIMEOUT, "crmd-finalization-timeout", "time", NULL, "30min", pcmk__valid_interval_spec, pcmk__opt_controld|pcmk__opt_advanced, N_("If you need to adjust this value, it probably indicates " "the presence of a bug."), NULL, }, { PCMK_OPT_TRANSITION_DELAY, "crmd-transition-delay", "time", NULL, "0s", pcmk__valid_interval_spec, pcmk__opt_controld|pcmk__opt_advanced, N_("Enabling this option will slow down cluster recovery under all " "conditions"), N_("Delay cluster recovery for this much time to allow for additional " "events to occur. Useful if your configuration is sensitive to " "the order in which ping updates arrive."), }, { PCMK_OPT_NO_QUORUM_POLICY, NULL, "select", PCMK_VALUE_STOP ", " PCMK_VALUE_FREEZE ", " PCMK_VALUE_IGNORE ", " PCMK_VALUE_DEMOTE ", " PCMK_VALUE_FENCE_LEGACY, PCMK_VALUE_STOP, pcmk__valid_no_quorum_policy, pcmk__opt_schedulerd, N_("What to do when the cluster does not have quorum"), NULL, }, { PCMK_OPT_SHUTDOWN_LOCK, NULL, "boolean", NULL, PCMK_VALUE_FALSE, pcmk__valid_boolean, pcmk__opt_schedulerd, N_("Whether to lock resources to a cleanly shut down node"), N_("When true, resources active on a node when it is cleanly shut down " "are kept \"locked\" to that node (not allowed to run elsewhere) " "until they start again on that node after it rejoins (or for at " "most shutdown-lock-limit, if set). Stonith resources and " "Pacemaker Remote connections are never locked. Clone and bundle " "instances and the promoted role of promotable clones are " "currently never locked, though support could be added in a future " "release."), }, { PCMK_OPT_SHUTDOWN_LOCK_LIMIT, NULL, "time", NULL, "0", pcmk__valid_interval_spec, pcmk__opt_schedulerd, N_("Do not lock resources to a cleanly shut down node longer than " "this"), N_("If shutdown-lock is true and this is set to a nonzero time " "duration, shutdown locks will expire after this much time has " "passed since the shutdown was initiated, even if the node has not " "rejoined."), }, { PCMK_OPT_ENABLE_ACL, NULL, "boolean", NULL, PCMK_VALUE_FALSE, pcmk__valid_boolean, pcmk__opt_based, N_("Enable Access Control Lists (ACLs) for the CIB"), NULL, }, { PCMK_OPT_SYMMETRIC_CLUSTER, NULL, "boolean", NULL, PCMK_VALUE_TRUE, pcmk__valid_boolean, pcmk__opt_schedulerd, N_("Whether resources can run on any node by default"), NULL, }, { PCMK_OPT_MAINTENANCE_MODE, NULL, "boolean", NULL, PCMK_VALUE_FALSE, pcmk__valid_boolean, pcmk__opt_schedulerd, N_("Whether the cluster should refrain from monitoring, starting, and " "stopping resources"), NULL, }, { PCMK_OPT_START_FAILURE_IS_FATAL, NULL, "boolean", NULL, PCMK_VALUE_TRUE, pcmk__valid_boolean, pcmk__opt_schedulerd, N_("Whether a start failure should prevent a resource from being " "recovered on the same node"), N_("When true, the cluster will immediately ban a resource from a node " "if it fails to start there. When false, the cluster will instead " "check the resource's fail count against its migration-threshold.") }, { PCMK_OPT_ENABLE_STARTUP_PROBES, NULL, "boolean", NULL, PCMK_VALUE_TRUE, pcmk__valid_boolean, pcmk__opt_schedulerd, N_("Whether the cluster should check for active resources during " "start-up"), NULL, }, // Fencing-related options { PCMK_OPT_STONITH_ENABLED, NULL, "boolean", NULL, PCMK_VALUE_TRUE, pcmk__valid_boolean, pcmk__opt_schedulerd|pcmk__opt_advanced, N_("Whether nodes may be fenced as part of recovery"), N_("If false, unresponsive nodes are immediately assumed to be " "harmless, and resources that were active on them may be recovered " "elsewhere. This can result in a \"split-brain\" situation, " "potentially leading to data loss and/or service unavailability."), }, { PCMK_OPT_STONITH_ACTION, NULL, "select", "reboot, off, poweroff", PCMK_ACTION_REBOOT, pcmk__is_fencing_action, pcmk__opt_schedulerd, N_("Action to send to fence device when a node needs to be fenced " "(\"poweroff\" is a deprecated alias for \"off\")"), NULL, }, { PCMK_OPT_STONITH_TIMEOUT, NULL, "time", NULL, "60s", pcmk__valid_interval_spec, pcmk__opt_schedulerd, N_("How long to wait for on, off, and reboot fence actions to complete " "by default"), NULL, }, { PCMK_OPT_HAVE_WATCHDOG, NULL, "boolean", NULL, PCMK_VALUE_FALSE, pcmk__valid_boolean, pcmk__opt_schedulerd|pcmk__opt_generated, N_("Whether watchdog integration is enabled"), N_("This is set automatically by the cluster according to whether SBD " "is detected to be in use. User-configured values are ignored. " "The value `true` is meaningful if diskless SBD is used and " "`stonith-watchdog-timeout` is nonzero. In that case, if fencing " "is required, watchdog-based self-fencing will be performed via " "SBD without requiring a fencing resource explicitly configured."), }, { /* @COMPAT Currently, unparsable values default to -1 (auto-calculate), * while missing values default to 0 (disable). All values are accepted * (unless the controller finds that the value conflicts with the * SBD_WATCHDOG_TIMEOUT). * * At a compatibility break: properly validate as a timeout, let * either negative values or a particular string like "auto" mean auto- * calculate, and use 0 as the single default for when the option either * is unset or fails to validate. */ PCMK_OPT_STONITH_WATCHDOG_TIMEOUT, NULL, "time", NULL, "0", NULL, pcmk__opt_controld, N_("How long before nodes can be assumed to be safely down when " "watchdog-based self-fencing via SBD is in use"), N_("If this is set to a positive value, lost nodes are assumed to " "achieve self-fencing using watchdog-based SBD within this much " "time. This does not require a fencing resource to be explicitly " "configured, though a fence_watchdog resource can be configured, to " "limit use to specific nodes. If this is set to 0 (the default), " "the cluster will never assume watchdog-based self-fencing. If this " "is set to a negative value, the cluster will use twice the local " "value of the `SBD_WATCHDOG_TIMEOUT` environment variable if that " "is positive, or otherwise treat this as 0. WARNING: When used, " "this timeout must be larger than `SBD_WATCHDOG_TIMEOUT` on all " "nodes that use watchdog-based SBD, and Pacemaker will refuse to " "start on any of those nodes where this is not true for the local " "value or SBD is not active. When this is set to a negative value, " "`SBD_WATCHDOG_TIMEOUT` must be set to the same value on all nodes " "that use SBD, otherwise data corruption or loss could occur."), }, { PCMK_OPT_STONITH_MAX_ATTEMPTS, NULL, "integer", NULL, "10", pcmk__valid_positive_int, pcmk__opt_controld, N_("How many times fencing can fail before it will no longer be " "immediately re-attempted on a target"), NULL, }, { PCMK_OPT_CONCURRENT_FENCING, NULL, "boolean", NULL, PCMK__CONCURRENT_FENCING_DEFAULT, pcmk__valid_boolean, pcmk__opt_schedulerd, N_("Allow performing fencing operations in parallel"), NULL, }, { PCMK_OPT_STARTUP_FENCING, NULL, "boolean", NULL, PCMK_VALUE_TRUE, pcmk__valid_boolean, pcmk__opt_schedulerd|pcmk__opt_advanced, N_("Whether to fence unseen nodes at start-up"), N_("Setting this to false may lead to a \"split-brain\" situation, " "potentially leading to data loss and/or service unavailability."), }, { PCMK_OPT_PRIORITY_FENCING_DELAY, NULL, "time", NULL, "0", pcmk__valid_interval_spec, pcmk__opt_schedulerd, N_("Apply fencing delay targeting the lost nodes with the highest " "total resource priority"), N_("Apply specified delay for the fencings that are targeting the lost " "nodes with the highest total resource priority in case we don't " "have the majority of the nodes in our cluster partition, so that " "the more significant nodes potentially win any fencing match, " "which is especially meaningful under split-brain of 2-node " "cluster. A promoted resource instance takes the base priority + 1 " "on calculation if the base priority is not 0. Any static/random " "delays that are introduced by `pcmk_delay_base/max` configured " "for the corresponding fencing resources will be added to this " "delay. This delay should be significantly greater than, safely " "twice, the maximum `pcmk_delay_base/max`. By default, priority " "fencing delay is disabled."), }, { PCMK_OPT_NODE_PENDING_TIMEOUT, NULL, "time", NULL, "0", pcmk__valid_interval_spec, pcmk__opt_schedulerd, N_("How long to wait for a node that has joined the cluster to join " "the controller process group"), N_("Fence nodes that do not join the controller process group within " "this much time after joining the cluster, to allow the cluster " "to continue managing resources. A value of 0 means never fence " "pending nodes. Setting the value to 2h means fence nodes after " "2 hours."), }, { PCMK_OPT_CLUSTER_DELAY, NULL, "time", NULL, "60s", pcmk__valid_interval_spec, pcmk__opt_schedulerd, N_("Maximum time for node-to-node communication"), N_("The node elected Designated Controller (DC) will consider an action " "failed if it does not get a response from the node executing the " "action within this time (after considering the action's own " "timeout). The \"correct\" value will depend on the speed and " "load of your network and cluster nodes.") }, // Limits { PCMK_OPT_LOAD_THRESHOLD, NULL, "percentage", NULL, "80%", pcmk__valid_percentage, pcmk__opt_controld, N_("Maximum amount of system load that should be used by cluster " "nodes"), N_("The cluster will slow down its recovery process when the amount of " "system resources used (currently CPU) approaches this limit"), }, { PCMK_OPT_NODE_ACTION_LIMIT, NULL, "integer", NULL, "0", pcmk__valid_int, pcmk__opt_controld, N_("Maximum number of jobs that can be scheduled per node (defaults to " "2x cores)"), NULL, }, { PCMK_OPT_BATCH_LIMIT, NULL, "integer", NULL, "0", pcmk__valid_int, pcmk__opt_schedulerd, N_("Maximum number of jobs that the cluster may execute in parallel " "across all nodes"), N_("The \"correct\" value will depend on the speed and load of your " "network and cluster nodes. If set to 0, the cluster will " "impose a dynamically calculated limit when any node has a " "high load."), }, { PCMK_OPT_MIGRATION_LIMIT, NULL, "integer", NULL, "-1", pcmk__valid_int, pcmk__opt_schedulerd, N_("The number of live migration actions that the cluster is allowed " "to execute in parallel on a node (-1 means no limit)"), NULL, }, { PCMK_OPT_CLUSTER_IPC_LIMIT, NULL, "integer", NULL, "500", pcmk__valid_positive_int, pcmk__opt_based, N_("Maximum IPC message backlog before disconnecting a cluster daemon"), N_("Raise this if log has \"Evicting client\" messages for cluster " "daemon PIDs (a good value is the number of resources in the " "cluster multiplied by the number of nodes)."), }, // Orphans and stopping { PCMK_OPT_STOP_ALL_RESOURCES, NULL, "boolean", NULL, PCMK_VALUE_FALSE, pcmk__valid_boolean, pcmk__opt_schedulerd, N_("Whether the cluster should stop all active resources"), NULL, }, { PCMK_OPT_STOP_ORPHAN_RESOURCES, NULL, "boolean", NULL, PCMK_VALUE_TRUE, pcmk__valid_boolean, pcmk__opt_schedulerd, N_("Whether to stop resources that were removed from the " "configuration"), NULL, }, { PCMK_OPT_STOP_ORPHAN_ACTIONS, NULL, "boolean", NULL, PCMK_VALUE_TRUE, pcmk__valid_boolean, pcmk__opt_schedulerd, N_("Whether to cancel recurring actions removed from the " "configuration"), NULL, }, { PCMK__OPT_REMOVE_AFTER_STOP, NULL, "boolean", NULL, PCMK_VALUE_FALSE, pcmk__valid_boolean, pcmk__opt_schedulerd|pcmk__opt_deprecated, N_("Whether to remove stopped resources from the executor"), N_("Values other than default are poorly tested and potentially " "dangerous."), }, // Storing inputs { PCMK_OPT_PE_ERROR_SERIES_MAX, NULL, "integer", NULL, "-1", pcmk__valid_int, pcmk__opt_schedulerd, N_("The number of scheduler inputs resulting in errors to save"), N_("Zero to disable, -1 to store unlimited."), }, { PCMK_OPT_PE_WARN_SERIES_MAX, NULL, "integer", NULL, "5000", pcmk__valid_int, pcmk__opt_schedulerd, N_("The number of scheduler inputs resulting in warnings to save"), N_("Zero to disable, -1 to store unlimited."), }, { PCMK_OPT_PE_INPUT_SERIES_MAX, NULL, "integer", NULL, "4000", pcmk__valid_int, pcmk__opt_schedulerd, N_("The number of scheduler inputs without errors or warnings to save"), N_("Zero to disable, -1 to store unlimited."), }, // Node health { PCMK_OPT_NODE_HEALTH_STRATEGY, NULL, "select", PCMK_VALUE_NONE ", " PCMK_VALUE_MIGRATE_ON_RED ", " PCMK_VALUE_ONLY_GREEN ", " PCMK_VALUE_PROGRESSIVE ", " PCMK_VALUE_CUSTOM, PCMK_VALUE_NONE, pcmk__validate_health_strategy, pcmk__opt_schedulerd, N_("How cluster should react to node health attributes"), N_("Requires external entities to create node attributes (named with " "the prefix \"#health\") with values \"red\", \"yellow\", or " "\"green\".") }, { PCMK_OPT_NODE_HEALTH_BASE, NULL, "integer", NULL, "0", pcmk__valid_int, pcmk__opt_schedulerd, N_("Base health score assigned to a node"), N_("Only used when \"node-health-strategy\" is set to " "\"progressive\"."), }, { PCMK_OPT_NODE_HEALTH_GREEN, NULL, "integer", NULL, "0", pcmk__valid_int, pcmk__opt_schedulerd, N_("The score to use for a node health attribute whose value is " "\"green\""), N_("Only used when \"node-health-strategy\" is set to \"custom\" or " "\"progressive\"."), }, { PCMK_OPT_NODE_HEALTH_YELLOW, NULL, "integer", NULL, "0", pcmk__valid_int, pcmk__opt_schedulerd, N_("The score to use for a node health attribute whose value is " "\"yellow\""), N_("Only used when \"node-health-strategy\" is set to \"custom\" or " "\"progressive\"."), }, { PCMK_OPT_NODE_HEALTH_RED, NULL, "integer", NULL, "-INFINITY", pcmk__valid_int, pcmk__opt_schedulerd, N_("The score to use for a node health attribute whose value is " "\"red\""), N_("Only used when \"node-health-strategy\" is set to \"custom\" or " "\"progressive\".") }, // Placement strategy { PCMK_OPT_PLACEMENT_STRATEGY, NULL, "select", PCMK_VALUE_DEFAULT ", " PCMK_VALUE_UTILIZATION ", " PCMK_VALUE_MINIMAL ", " PCMK_VALUE_BALANCED, PCMK_VALUE_DEFAULT, pcmk__valid_placement_strategy, pcmk__opt_schedulerd, N_("How the cluster should allocate resources to nodes"), NULL, }, { NULL, }, }; /* * Environment variable option handling */ /*! * \internal * \brief Get the value of a Pacemaker environment variable option * * If an environment variable option is set, with either a PCMK_ or (for * backward compatibility) HA_ prefix, log and return the value. * * \param[in] option Environment variable name (without prefix) * * \return Value of environment variable option, or NULL in case of * option name too long or value not found */ const char * pcmk__env_option(const char *option) { const char *const prefixes[] = {"PCMK_", "HA_"}; char env_name[NAME_MAX]; const char *value = NULL; CRM_CHECK(!pcmk__str_empty(option), return NULL); for (int i = 0; i < PCMK__NELEM(prefixes); i++) { int rv = snprintf(env_name, NAME_MAX, "%s%s", prefixes[i], option); if (rv < 0) { crm_err("Failed to write %s%s to buffer: %s", prefixes[i], option, strerror(errno)); return NULL; } if (rv >= sizeof(env_name)) { crm_trace("\"%s%s\" is too long", prefixes[i], option); continue; } value = getenv(env_name); if (value != NULL) { crm_trace("Found %s = %s", env_name, value); return value; } } crm_trace("Nothing found for %s", option); return NULL; } /*! * \brief Set or unset a Pacemaker environment variable option * * Set an environment variable option with a \c "PCMK_" prefix and optionally * an \c "HA_" prefix for backward compatibility. * * \param[in] option Environment variable name (without prefix) * \param[in] value New value (or NULL to unset) * \param[in] compat If false and \p value is not \c NULL, set only * \c "PCMK_