diff --git a/daemons/controld/controld_callbacks.c b/daemons/controld/controld_callbacks.c index 06ffb9d1b9..3ce7470442 100644 --- a/daemons/controld/controld_callbacks.c +++ b/daemons/controld/controld_callbacks.c @@ -1,340 +1,340 @@ /* * Copyright 2004-2019 the Pacemaker project contributors * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* From join_dc... */ extern gboolean check_join_state(enum crmd_fsa_state cur_state, const char *source); void crmd_ha_msg_filter(xmlNode * msg) { if (AM_I_DC) { const char *sys_from = crm_element_value(msg, F_CRM_SYS_FROM); if (safe_str_eq(sys_from, CRM_SYSTEM_DC)) { const char *from = crm_element_value(msg, F_ORIG); if (safe_str_neq(from, fsa_our_uname)) { int level = LOG_INFO; const char *op = crm_element_value(msg, F_CRM_TASK); /* make sure the election happens NOW */ if (fsa_state != S_ELECTION) { ha_msg_input_t new_input; level = LOG_WARNING; new_input.msg = msg; register_fsa_error_adv(C_FSA_INTERNAL, I_ELECTION, NULL, &new_input, __FUNCTION__); } do_crm_log(level, "Another DC detected: %s (op=%s)", from, op); goto done; } } } else { const char *sys_to = crm_element_value(msg, F_CRM_SYS_TO); if (safe_str_eq(sys_to, CRM_SYSTEM_DC)) { return; } } /* crm_log_xml_trace("HA[inbound]", msg); */ route_message(C_HA_MESSAGE, msg); done: trigger_fsa(fsa_source); } /*! * \internal * \brief Check whether a node is online * * \param[in] node Node to check * * \retval -1 if completely dead * \retval 0 if partially alive * \retval 1 if completely alive */ static int node_alive(const crm_node_t *node) { if (is_set(node->flags, crm_remote_node)) { // Pacemaker Remote nodes can't be partially alive return safe_str_eq(node->state, CRM_NODE_MEMBER)? 1: -1; } else if (crm_is_peer_active(node)) { // Completely up cluster node: both cluster member and peer return 1; } else if (is_not_set(node->processes, crm_get_cluster_proc()) && safe_str_neq(node->state, CRM_NODE_MEMBER)) { // Completely down cluster node: neither cluster member nor peer return -1; } // Partially up cluster node: only cluster member or only peer return 0; } #define state_text(state) ((state)? (const char *)(state) : "in unknown state") void peer_update_callback(enum crm_status_type type, crm_node_t * node, const void *data) { uint32_t old = 0; bool appeared = FALSE; bool is_remote = is_set(node->flags, crm_remote_node); /* The controller waits to receive some information from the membership * layer before declaring itself operational. If this is being called for a * cluster node, indicate that we have it. */ if (!is_remote) { set_bit(fsa_input_register, R_PEER_DATA); } if (node->uname == NULL) { return; } switch (type) { case crm_status_uname: /* If we've never seen the node, then it also won't be in the status section */ crm_info("%s node %s is now %s", (is_remote? "Remote" : "Cluster"), node->uname, state_text(node->state)); return; case crm_status_nstate: /* This callback should not be called unless the state actually * changed, but here's a failsafe just in case. */ CRM_CHECK(safe_str_neq(data, node->state), return); crm_info("%s node %s is now %s (was %s)", (is_remote? "Remote" : "Cluster"), node->uname, state_text(node->state), state_text(data)); if (safe_str_eq(CRM_NODE_MEMBER, node->state)) { appeared = TRUE; if (!is_remote) { remove_stonith_cleanup(node->uname); } } else { controld_remove_voter(node->uname); } crmd_alert_node_event(node); break; case crm_status_processes: CRM_CHECK(data != NULL, return); old = *(const uint32_t *)data; appeared = is_set(node->processes, crm_get_cluster_proc()); crm_info("Node %s is %s a peer " CRM_XS " DC=%s old=0x%07x new=0x%07x", node->uname, (appeared? "now" : "no longer"), (AM_I_DC? "true" : (fsa_our_dc? fsa_our_dc : "")), old, node->processes); if (is_not_set((node->processes ^ old), crm_get_cluster_proc())) { /* Peer status did not change. This should not be possible, * since we don't track process flags other than peer status. */ crm_trace("Process flag 0x%7x did not change from 0x%7x to 0x%7x", crm_get_cluster_proc(), old, node->processes); return; } if (!appeared) { controld_remove_voter(node->uname); } if (is_not_set(fsa_input_register, R_CIB_CONNECTED)) { crm_trace("Ignoring peer status change because not connected to CIB"); return; } else if (fsa_state == S_STOPPING) { crm_trace("Ignoring peer status change because stopping"); return; } if (safe_str_eq(node->uname, fsa_our_uname) && !appeared) { /* Did we get evicted? */ crm_notice("Our peer connection failed"); register_fsa_input(C_CRMD_STATUS_CALLBACK, I_ERROR, NULL); } else if (safe_str_eq(node->uname, fsa_our_dc) && crm_is_peer_active(node) == FALSE) { /* Did the DC leave us? */ crm_notice("Our peer on the DC (%s) is dead", fsa_our_dc); register_fsa_input(C_CRMD_STATUS_CALLBACK, I_ELECTION, NULL); /* @COMPAT DC < 1.1.13: If a DC shuts down normally, we don't * want to fence it. Newer DCs will send their shutdown request * to all peers, who will update the DC's expected state to * down, thus avoiding fencing. We can safely erase the DC's * transient attributes when it leaves in that case. However, * the only way to avoid fencing older DCs is to leave the * transient attributes intact until it rejoins. */ if (compare_version(fsa_our_dc_version, "3.0.9") > 0) { erase_status_tag(node->uname, XML_TAG_TRANSIENT_NODEATTRS, cib_scope_local); } } else if(AM_I_DC) { if (appeared) { te_trigger_stonith_history_sync(); } else { erase_status_tag(node->uname, XML_TAG_TRANSIENT_NODEATTRS, cib_scope_local); } } break; } if (AM_I_DC) { xmlNode *update = NULL; int flags = node_update_peer; int alive = node_alive(node); crm_action_t *down = match_down_event(node->uuid); crm_trace("Alive=%d, appeared=%d, down=%d", alive, appeared, (down? down->id : -1)); - if (appeared && (alive > 0)) { + if (appeared && (alive > 0) && !is_remote) { register_fsa_input_before(C_FSA_INTERNAL, I_NODE_JOIN, NULL); } if (down) { const char *task = crm_element_value(down->xml, XML_LRM_ATTR_TASK); if (safe_str_eq(task, CRM_OP_FENCE)) { /* tengine_stonith_callback() confirms fence actions */ crm_trace("Updating CIB %s fencer reported fencing of %s complete", (down->confirmed? "after" : "before"), node->uname); } else if (!appeared && safe_str_eq(task, CRM_OP_SHUTDOWN)) { // Shutdown actions are immediately confirmed (i.e. no_wait) if (!is_remote) { flags |= node_update_join | node_update_expected; crmd_peer_down(node, FALSE); check_join_state(fsa_state, __FUNCTION__); } if (alive >= 0) { crm_info("%s of peer %s is in progress " CRM_XS " action=%d", task, node->uname, down->id); } else { crm_notice("%s of peer %s is complete " CRM_XS " action=%d", task, node->uname, down->id); update_graph(transition_graph, down); trigger_graph(); } } else { crm_trace("Node %s is %s, was expected to %s (op %d)", node->uname, ((alive > 0)? "alive" : ((alive < 0)? "dead" : "partially alive")), task, down->id); } } else if (appeared == FALSE) { crm_warn("Stonith/shutdown of node %s was not expected", node->uname); if (!is_remote) { crm_update_peer_join(__FUNCTION__, node, crm_join_none); check_join_state(fsa_state, __FUNCTION__); } abort_transition(INFINITY, tg_restart, "Node failure", NULL); fail_incompletable_actions(transition_graph, node->uuid); } else { crm_trace("Node %s came up, was not expected to be down", node->uname); } if (is_remote) { /* A pacemaker_remote node won't have its cluster status updated * in the CIB by membership-layer callbacks, so do it here. */ flags |= node_update_cluster; /* Trigger resource placement on newly integrated nodes */ if (appeared) { abort_transition(INFINITY, tg_restart, "pacemaker_remote node integrated", NULL); } } /* Update the CIB node state */ update = create_node_state_update(node, flags, NULL, __FUNCTION__); if (update == NULL) { crm_debug("Node state update not yet possible for %s", node->uname); } else { fsa_cib_anon_update(XML_CIB_TAG_STATUS, update); } free_xml(update); } trigger_fsa(fsa_source); } void crmd_cib_connection_destroy(gpointer user_data) { CRM_CHECK(user_data == fsa_cib_conn,;); crm_trace("Invoked"); trigger_fsa(fsa_source); fsa_cib_conn->state = cib_disconnected; if (is_set(fsa_input_register, R_CIB_CONNECTED) == FALSE) { crm_info("Connection to the CIB manager terminated"); return; } // @TODO This should trigger a reconnect, not a shutdown crm_crit("Lost connection to the CIB manager, shutting down"); register_fsa_input(C_FSA_INTERNAL, I_ERROR, NULL); clear_bit(fsa_input_register, R_CIB_CONNECTED); return; } gboolean crm_fsa_trigger(gpointer user_data) { crm_trace("Invoked (queue len: %d)", g_list_length(fsa_message_queue)); s_crmd_fsa(C_FSA_INTERNAL); crm_trace("Exited (queue len: %d)", g_list_length(fsa_message_queue)); return TRUE; } diff --git a/daemons/execd/pacemaker-execd.c b/daemons/execd/pacemaker-execd.c index e2fdfcad71..cfa5500f3a 100644 --- a/daemons/execd/pacemaker-execd.c +++ b/daemons/execd/pacemaker-execd.c @@ -1,622 +1,630 @@ /* * Copyright 2012-2019 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 "pacemaker-execd.h" #if defined(HAVE_GNUTLS_GNUTLS_H) && defined(SUPPORT_REMOTE) # define ENABLE_PCMK_REMOTE // Hidden in liblrmd extern int lrmd_tls_send_msg(crm_remote_t *session, xmlNode *msg, uint32_t id, const char *msg_type); #endif static GMainLoop *mainloop = NULL; static qb_ipcs_service_t *ipcs = NULL; static stonith_t *stonith_api = NULL; int lrmd_call_id = 0; #ifdef ENABLE_PCMK_REMOTE /* whether shutdown request has been sent */ static sig_atomic_t shutting_down = FALSE; /* timer for waiting for acknowledgment of shutdown request */ static guint shutdown_ack_timer = 0; static gboolean lrmd_exit(gpointer data); #endif static void stonith_connection_destroy_cb(stonith_t * st, stonith_event_t * e) { stonith_api->state = stonith_disconnected; crm_err("Connection to fencer lost"); stonith_connection_failed(); } stonith_t * get_stonith_connection(void) { if (stonith_api && stonith_api->state == stonith_disconnected) { stonith_api_delete(stonith_api); stonith_api = NULL; } if (stonith_api == NULL) { int rc = pcmk_ok; stonith_api = stonith_api_new(); rc = stonith_api_connect_retry(stonith_api, crm_system_name, 10); if (rc != pcmk_ok) { crm_err("Could not connect to fencer in 10 attempts: %s " CRM_XS " rc=%d", pcmk_strerror(rc), rc); stonith_api_delete(stonith_api); stonith_api = NULL; } else { stonith_api->cmds->register_notification(stonith_api, T_STONITH_NOTIFY_DISCONNECT, stonith_connection_destroy_cb); } } return stonith_api; } static int32_t lrmd_ipc_accept(qb_ipcs_connection_t * c, uid_t uid, gid_t gid) { crm_trace("Connection %p", c); if (crm_client_new(c, uid, gid) == NULL) { return -EIO; } return 0; } static void lrmd_ipc_created(qb_ipcs_connection_t * c) { crm_client_t *new_client = crm_client_get(c); crm_trace("Connection %p", c); CRM_ASSERT(new_client != NULL); /* Now that the connection is offically established, alert * the other clients a new connection exists. */ notify_of_new_client(new_client); } static int32_t lrmd_ipc_dispatch(qb_ipcs_connection_t * c, void *data, size_t size) { uint32_t id = 0; uint32_t flags = 0; crm_client_t *client = crm_client_get(c); xmlNode *request = crm_ipcs_recv(client, data, size, &id, &flags); CRM_CHECK(client != NULL, crm_err("Invalid client"); return FALSE); CRM_CHECK(client->id != NULL, crm_err("Invalid client: %p", client); return FALSE); CRM_CHECK(flags & crm_ipc_client_response, crm_err("Invalid client request: %p", client); return FALSE); if (!request) { return 0; } if (!client->name) { const char *value = crm_element_value(request, F_LRMD_CLIENTNAME); if (value == NULL) { client->name = crm_itoa(crm_ipcs_client_pid(c)); } else { client->name = strdup(value); } } lrmd_call_id++; if (lrmd_call_id < 1) { lrmd_call_id = 1; } crm_xml_add(request, F_LRMD_CLIENTID, client->id); crm_xml_add(request, F_LRMD_CLIENTNAME, client->name); crm_xml_add_int(request, F_LRMD_CALLID, lrmd_call_id); process_lrmd_message(client, id, request); free_xml(request); return 0; } /*! * \internal * \brief Free a client connection, and exit if appropriate * * \param[in] client Client connection to free */ void lrmd_client_destroy(crm_client_t *client) { crm_client_destroy(client); #ifdef ENABLE_PCMK_REMOTE /* If we were waiting to shut down, we can now safely do so * if there are no more proxied IPC providers */ if (shutting_down && (ipc_proxy_get_provider() == NULL)) { lrmd_exit(NULL); } #endif } static int32_t lrmd_ipc_closed(qb_ipcs_connection_t * c) { crm_client_t *client = crm_client_get(c); if (client == NULL) { return 0; } crm_trace("Connection %p", c); client_disconnect_cleanup(client->id); #ifdef ENABLE_PCMK_REMOTE ipc_proxy_remove_provider(client); #endif lrmd_client_destroy(client); return 0; } static void lrmd_ipc_destroy(qb_ipcs_connection_t * c) { lrmd_ipc_closed(c); crm_trace("Connection %p", c); } static struct qb_ipcs_service_handlers lrmd_ipc_callbacks = { .connection_accept = lrmd_ipc_accept, .connection_created = lrmd_ipc_created, .msg_process = lrmd_ipc_dispatch, .connection_closed = lrmd_ipc_closed, .connection_destroyed = lrmd_ipc_destroy }; int lrmd_server_send_reply(crm_client_t * client, uint32_t id, xmlNode * reply) { crm_trace("Sending reply (%d) to client (%s)", id, client->id); switch (client->kind) { case CRM_CLIENT_IPC: return crm_ipcs_send(client, id, reply, FALSE); #ifdef ENABLE_PCMK_REMOTE case CRM_CLIENT_TLS: return lrmd_tls_send_msg(client->remote, reply, id, "reply"); #endif default: crm_err("Could not send reply: unknown client type %d", client->kind); } return -ENOTCONN; } int lrmd_server_send_notify(crm_client_t * client, xmlNode * msg) { crm_trace("Sending notification to client (%s)", client->id); switch (client->kind) { case CRM_CLIENT_IPC: if (client->ipcs == NULL) { crm_trace("Could not notify local client: disconnected"); return -ENOTCONN; } return crm_ipcs_send(client, 0, msg, crm_ipc_server_event); #ifdef ENABLE_PCMK_REMOTE case CRM_CLIENT_TLS: if (client->remote == NULL) { crm_trace("Could not notify remote client: disconnected"); return -ENOTCONN; } return lrmd_tls_send_msg(client->remote, msg, 0, "notify"); #endif default: crm_err("Could not notify client: unknown type %d", client->kind); } return -ENOTCONN; } /*! * \internal * \brief Clean up and exit immediately * * \param[in] data Ignored * * \return Doesn't return * \note This can be used as a timer callback. */ static gboolean lrmd_exit(gpointer data) { crm_info("Terminating with %d clients", crm_hash_table_size(client_connections)); if (stonith_api) { stonith_api->cmds->remove_notification(stonith_api, T_STONITH_NOTIFY_DISCONNECT); stonith_api->cmds->disconnect(stonith_api); stonith_api_delete(stonith_api); } if (ipcs) { mainloop_del_ipc_server(ipcs); } #ifdef ENABLE_PCMK_REMOTE lrmd_tls_server_destroy(); ipc_proxy_cleanup(); #endif crm_client_cleanup(); g_hash_table_destroy(rsc_list); if (mainloop) { lrmd_drain_alerts(mainloop); } crm_exit(CRM_EX_OK); return FALSE; } /*! * \internal * \brief Request cluster shutdown if appropriate, otherwise exit immediately * * \param[in] nsig Signal that caused invocation (ignored) */ static void lrmd_shutdown(int nsig) { #ifdef ENABLE_PCMK_REMOTE crm_client_t *ipc_proxy = ipc_proxy_get_provider(); /* If there are active proxied IPC providers, then we may be running * resources, so notify the cluster that we wish to shut down. */ if (ipc_proxy) { if (shutting_down) { crm_notice("Waiting for cluster to stop resources before exiting"); return; } crm_info("Sending shutdown request to cluster"); if (ipc_proxy_shutdown_req(ipc_proxy) < 0) { crm_crit("Shutdown request failed, exiting immediately"); } else { /* We requested a shutdown. Now, we need to wait for an * acknowledgement from the proxy host (which ensures the proxy host * supports shutdown requests), then wait for all proxy hosts to * disconnect (which ensures that all resources have been stopped). */ shutting_down = TRUE; /* Stop accepting new proxy connections */ lrmd_tls_server_destroy(); /* Older controller versions will never acknowledge our request, so * set a fairly short timeout to exit quickly in that case. If we * get the ack, we'll defuse this timer. */ shutdown_ack_timer = g_timeout_add_seconds(20, lrmd_exit, NULL); /* Currently, we let the OS kill us if the clients don't disconnect * in a reasonable time. We could instead set a long timer here * (shorter than what the OS is likely to use) and exit immediately * if it pops. */ return; } } #endif lrmd_exit(NULL); } /*! * \internal * \brief Defuse short exit timer if shutting down */ void handle_shutdown_ack() { #ifdef ENABLE_PCMK_REMOTE if (shutting_down) { crm_info("Received shutdown ack"); if (shutdown_ack_timer > 0) { g_source_remove(shutdown_ack_timer); shutdown_ack_timer = 0; } return; } #endif crm_debug("Ignoring unexpected shutdown ack"); } /*! * \internal * \brief Make short exit timer fire immediately */ void handle_shutdown_nack() { #ifdef ENABLE_PCMK_REMOTE if (shutting_down) { crm_info("Received shutdown nack"); if (shutdown_ack_timer > 0) { g_source_remove(shutdown_ack_timer); shutdown_ack_timer = g_timeout_add(0, lrmd_exit, NULL); } return; } #endif crm_debug("Ignoring unexpected shutdown nack"); } static pid_t main_pid = 0; static void sigdone(void) { exit(CRM_EX_OK); } static void sigreap(void) { pid_t pid = 0; int status; do { /* * Opinions seem to differ as to what to put here: * -1, any child process * 0, any child process whose process group ID is equal to that of the calling process */ pid = waitpid(-1, &status, WNOHANG); if(pid == main_pid) { /* Exit when pacemaker-remote exits and use the same return code */ if (WIFEXITED(status)) { exit(WEXITSTATUS(status)); } exit(CRM_EX_ERROR); } } while (pid > 0); } static struct { int sig; void (*handler)(void); } sigmap[] = { { SIGCHLD, sigreap }, { SIGINT, sigdone }, }; static void spawn_pidone(int argc, char **argv, char **envp) { sigset_t set; if (getpid() != 1) { return; } + /* Containers can be expected to have /var/log, but they may not have + * /var/log/pacemaker, so use a different default if no value has been + * explicitly configured in the container's environment. + */ + if (daemon_option("logfile") == NULL) { + set_daemon_option("logfile", "/var/log/pcmk-init.log"); + } + sigfillset(&set); sigprocmask(SIG_BLOCK, &set, 0); main_pid = fork(); switch (main_pid) { case 0: sigprocmask(SIG_UNBLOCK, &set, NULL); setsid(); setpgid(0, 0); /* Child remains as pacemaker-remoted */ return; case -1: perror("fork"); } /* Parent becomes the reaper of zombie processes */ /* Safe to initialize logging now if needed */ #ifdef HAVE___PROGNAME /* Differentiate ourselves in the 'ps' output */ { char *p; int i, maxlen; char *LastArgv = NULL; const char *name = "pcmk-init"; for(i = 0; i < argc; i++) { if(!i || (LastArgv + 1 == argv[i])) LastArgv = argv[i] + strlen(argv[i]); } for(i = 0; envp[i] != NULL; i++) { if((LastArgv + 1) == envp[i]) { LastArgv = envp[i] + strlen(envp[i]); } } maxlen = (LastArgv - argv[0]) - 2; i = strlen(name); /* We can overwrite individual argv[] arguments */ snprintf(argv[0], maxlen, "%s", name); /* Now zero out everything else */ p = &argv[0][i]; while(p < LastArgv) *p++ = '\0'; argv[1] = NULL; } #endif /* HAVE___PROGNAME */ while (1) { int sig; size_t i; sigwait(&set, &sig); for (i = 0; i < DIMOF(sigmap); i++) { if (sigmap[i].sig == sig) { sigmap[i].handler(); break; } } } } /* *INDENT-OFF* */ static struct crm_option long_options[] = { /* Top-level Options */ {"help", 0, 0, '?', "\tThis text"}, {"version", 0, 0, '$', "\tVersion information" }, {"verbose", 0, 0, 'V', "\tIncrease debug output"}, {"logfile", 1, 0, 'l', "\tSend logs to the additional named logfile"}, #ifdef ENABLE_PCMK_REMOTE {"port", 1, 0, 'p', "\tPort to listen on"}, #endif {0, 0, 0, 0} }; /* *INDENT-ON* */ #ifdef ENABLE_PCMK_REMOTE # define EXECD_TYPE "remote" #else # define EXECD_TYPE "local" #endif int main(int argc, char **argv, char **envp) { int flag = 0; int index = 0; int bump_log_num = 0; const char *option = NULL; /* If necessary, create PID1 now before any FDs are opened */ spawn_pidone(argc, argv, envp); #ifndef ENABLE_PCMK_REMOTE crm_log_preinit("pacemaker-execd", argc, argv); crm_set_options(NULL, "[options]", long_options, "Resource agent executor daemon for cluster nodes"); #else crm_log_preinit("pacemaker-remoted", argc, argv); crm_set_options(NULL, "[options]", long_options, "Resource agent executor daemon for Pacemaker Remote nodes"); #endif while (1) { flag = crm_get_option(argc, argv, &index); if (flag == -1) { break; } switch (flag) { case 'l': crm_add_logfile(optarg); break; case 'p': setenv("PCMK_remote_port", optarg, 1); break; case 'V': bump_log_num++; break; case '?': case '$': crm_help(flag, CRM_EX_OK); break; default: crm_help('?', CRM_EX_USAGE); break; } } crm_log_init(NULL, LOG_INFO, TRUE, FALSE, argc, argv, FALSE); while (bump_log_num > 0) { crm_bump_log_level(argc, argv); bump_log_num--; } option = daemon_option("logfacility"); if(option && safe_str_neq(option, "none")) { setenv("HA_LOGFACILITY", option, 1); /* Used by the ocf_log/ha_log OCF macro */ } option = daemon_option("logfile"); if(option && safe_str_neq(option, "none")) { setenv("HA_LOGFILE", option, 1); /* Used by the ocf_log/ha_log OCF macro */ if (daemon_option_enabled(crm_system_name, "debug")) { setenv("HA_DEBUGLOG", option, 1); /* Used by the ocf_log/ha_debug OCF macro */ } } crm_notice("Starting Pacemaker " EXECD_TYPE " executor"); /* The presence of this variable allegedly controls whether child * processes like httpd will try and use Systemd's sd_notify * API */ unsetenv("NOTIFY_SOCKET"); /* Used by RAs - Leave owned by root */ crm_build_path(CRM_RSCTMP_DIR, 0755); rsc_list = g_hash_table_new_full(crm_str_hash, g_str_equal, NULL, free_rsc); ipcs = mainloop_add_ipc_server(CRM_SYSTEM_LRMD, QB_IPC_SHM, &lrmd_ipc_callbacks); if (ipcs == NULL) { crm_err("Failed to create IPC server: shutting down and inhibiting respawn"); crm_exit(CRM_EX_FATAL); } #ifdef ENABLE_PCMK_REMOTE if (lrmd_init_remote_tls_server() < 0) { crm_err("Failed to create TLS listener: shutting down and staying down"); crm_exit(CRM_EX_FATAL); } ipc_proxy_init(); #endif mainloop_add_signal(SIGTERM, lrmd_shutdown); mainloop = g_main_loop_new(NULL, FALSE); crm_notice("Pacemaker " EXECD_TYPE " executor successfully started and accepting connections"); g_main_loop_run(mainloop); /* should never get here */ lrmd_exit(NULL); return CRM_EX_OK; } diff --git a/doc/Pacemaker_Explained/en-US/Ch-Advanced-Resources.txt b/doc/Pacemaker_Explained/en-US/Ch-Advanced-Resources.txt index e431626f0b..4a181dfb32 100644 --- a/doc/Pacemaker_Explained/en-US/Ch-Advanced-Resources.txt +++ b/doc/Pacemaker_Explained/en-US/Ch-Advanced-Resources.txt @@ -1,1414 +1,1423 @@ :compat-mode: legacy = Advanced Resource Types = [[group-resources]] == Groups - A Syntactic Shortcut == indexterm:[Group Resources] indexterm:[Resource,Groups] One of the most common elements of a cluster is a set of resources that need to be located together, start sequentially, and stop in the reverse order. To simplify this configuration, we support the concept of groups. .A group of two primitive resources ====== [source,XML] ------- ------- ====== Although the example above contains only two resources, there is no limit to the number of resources a group can contain. The example is also sufficient to explain the fundamental properties of a group: * Resources are started in the order they appear in (+Public-IP+ first, then +Email+) * Resources are stopped in the reverse order to which they appear in (+Email+ first, then +Public-IP+) If a resource in the group can't run anywhere, then nothing after that is allowed to run, too. * If +Public-IP+ can't run anywhere, neither can +Email+; * but if +Email+ can't run anywhere, this does not affect +Public-IP+ in any way The group above is logically equivalent to writing: .How the cluster sees a group resource ====== [source,XML] ------- ------- ====== Obviously as the group grows bigger, the reduced configuration effort can become significant. Another (typical) example of a group is a DRBD volume, the filesystem mount, an IP address, and an application that uses them. === Group Properties === .Properties of a Group Resource [width="95%",cols="3m,<5",options="header",align="center"] |========================================================= |Field |Description |id |A unique name for the group indexterm:[id,Group Resource Property] indexterm:[Resource,Group Property,id] |========================================================= === Group Options === Groups inherit the +priority+, +target-role+, and +is-managed+ properties from primitive resources. See <> for information about those properties. === Group Instance Attributes === Groups have no instance attributes. However, any that are set for the group object will be inherited by the group's children. === Group Contents === Groups may only contain a collection of cluster resources (see <>). To refer to a child of a group resource, just use the child's +id+ instead of the group's. === Group Constraints === Although it is possible to reference a group's children in constraints, it is usually preferable to reference the group itself. .Some constraints involving groups ====== [source,XML] ------- ------- ====== === Group Stickiness === indexterm:[resource-stickiness,Groups] Stickiness, the measure of how much a resource wants to stay where it is, is additive in groups. Every active resource of the group will contribute its stickiness value to the group's total. So if the default +resource-stickiness+ is 100, and a group has seven members, five of which are active, then the group as a whole will prefer its current location with a score of 500. [[s-resource-clone]] == Clones - Resources That Can Have Multiple Active Instances == indexterm:[Clone Resources] indexterm:[Resource,Clones] 'Clone' resources are resources that can have more than one copy active at the same time. This allows you, for example, to run a copy of a daemon on every node. You can clone any primitive or group resource. footnote:[ Of course, the service must support running multiple instances. ] === Anonymous versus Unique Clones === A clone resource is configured to be either 'anonymous' or 'globally unique'. Anonymous clones are the simplest. These behave completely identically everywhere they are running. Because of this, there can be only one instance of an anonymous clone active per node. The instances of globally unique clones are distinct entities. All instances are launched identically, but one instance of the clone is not identical to any other instance, whether running on the same node or a different node. As an example, a cloned IP address can use special kernel functionality such that each instance handles a subset of requests for the same IP address. [[s-resource-promotable]] === Promotable clones === indexterm:[Promotable Clone Resources] indexterm:[Resource,Promotable] If a clone is 'promotable', its instances can perform a special role that Pacemaker will manage via the +promote+ and +demote+ actions of the resource agent. Services that support such a special role have various terms for the special role and the default role: primary and secondary, master and replica, controller and worker, etc. Pacemaker uses the terms 'master' and 'slave', footnote:[ These are historical terms that will eventually be replaced, but the extensive use of them and the need for backward compatibility makes it a long process. You may see examples using a +master+ tag instead of a +clone+ tag with the +promotable+ meta-attribute set to +true+; the +master+ tag is supported, but deprecated, and will be removed in a future version. You may also see such services referred to as 'multi-state' or 'stateful'; these means the same thing as 'promotable'. ] but is agnostic to what the service calls them or what they do. All that Pacemaker cares about is that an instance comes up in the default role when started, and the resource agent supports the +promote+ and +demote+ actions to manage entering and exiting the special role. === Clone Properties === .Properties of a Clone Resource [width="95%",cols="3m,<5",options="header",align="center"] |========================================================= |Field |Description |id |A unique name for the clone indexterm:[id,Clone Property] indexterm:[Clone,Property,id] |========================================================= === Clone Options === <> inherited from primitive resources: +priority, target-role, is-managed+ .Clone-specific configuration options [width="95%",cols="1m,1,<3",options="header",align="center"] |========================================================= |Field |Default |Description |globally-unique |false |If +true+, each clone instance performs a distinct function indexterm:[globally-unique,Clone Option] indexterm:[Clone,Option,globally-unique] |clone-max |number of nodes in cluster |The maximum number of clone instances that can be started across the entire cluster indexterm:[clone-max,Clone Option] indexterm:[Clone,Option,clone-max] |clone-node-max |1 |If +globally-unique+ is +true+, the maximum number of clone instances that can be started on a single node indexterm:[clone-node-max,Clone Option] indexterm:[Clone,Option,clone-node-max] |clone-min |0 |Require at least this number of clone instances to be runnable before allowing resources depending on the clone to be runnable. A value of 0 means require all clone instances to be runnable. indexterm:[clone-min,Clone Option] indexterm:[Clone,Option,clone-min] |notify |false |Call the resource agent's +notify+ action for all active instances, before and after starting or stopping any clone instance. The resource agent must support this action. Allowed values: +false+, +true+ indexterm:[notify,Clone Option] indexterm:[Clone,Option,notify] |ordered |false |If +true+, clone instances must be started sequentially instead of in parallel Allowed values: +false+, +true+ indexterm:[ordered,Clone Option] indexterm:[Clone,Option,ordered] |interleave |false |When this clone is ordered relative to another clone, if this option is +false+ (the default), the ordering is relative to 'all' instances of the other clone, whereas if this option is +true+, the ordering is relative only to instances on the same node. Allowed values: +false+, +true+ indexterm:[interleave,Clone Option] indexterm:[Clone,Option,interleave] |promotable |false |If +true+, clone instances can perform a special role that Pacemaker will manage via the resource agent's +promote+ and +demote+ actions. The resource agent must support these actions. Allowed values: +false+, +true+ indexterm:[promotable,Clone Option] indexterm:[Clone,Option,promotable] |promoted-max |1 |If +promotable+ is +true+, the number of instances that can be promoted at one time across the entire cluster indexterm:[promoted-max,Clone Option] indexterm:[Clone,Option,promoted-max] |promoted-node-max |1 |If +promotable+ is +true+ and +globally-unique+ is +false+, the number of clone instances can be promoted at one time on a single node indexterm:[promoted-node-max,Clone Option] indexterm:[Clone,Option,promoted-node-max] |========================================================= For backward compatibility, +master-max+ and +master-node-max+ are accepted as aliases for +promoted-max+ and +promoted-node-max+, but are deprecated since 2.0.0, and support for them will be removed in a future version. === Clone Contents === Clones must contain exactly one primitive or group resource. .A clone that runs a web server on all nodes ==== [source,XML] ---- ---- ==== [WARNING] You should never reference the name of a clone's child (the primitive or group resource being cloned). If you think you need to do this, you probably need to re-evaluate your design. === Clone Instance Attributes === Clones have no instance attributes; however, any that are set here will be inherited by the clone's child. === Clone Constraints === In most cases, a clone will have a single instance on each active cluster node. If this is not the case, you can indicate which nodes the cluster should preferentially assign copies to with resource location constraints. These constraints are written no differently from those for primitive resources except that the clone's +id+ is used. .Some constraints involving clones ====== [source,XML] ------- ------- ====== Ordering constraints behave slightly differently for clones. In the example above, +apache-stats+ will wait until all copies of +apache-clone+ that need to be started have done so before being started itself. Only if _no_ copies can be started will +apache-stats+ be prevented from being active. Additionally, the clone will wait for +apache-stats+ to be stopped before stopping itself. Colocation of a primitive or group resource with a clone means that the resource can run on any node with an active instance of the clone. The cluster will choose an instance based on where the clone is running and the resource's own location preferences. Colocation between clones is also possible. If one clone +A+ is colocated with another clone +B+, the set of allowed locations for +A+ is limited to nodes on which +B+ is (or will be) active. Placement is then performed normally. ==== Promotable Clone Constraints ==== For promotable clone resources, the +first-action+ and/or +then-action+ fields for ordering constraints may be set to +promote+ or +demote+ to constrain the master role, and colocation constraints may contain +rsc-role+ and/or +with-rsc-role+ fields. .Additional colocation constraint options for promotable clone resources [width="95%",cols="1m,1,<3",options="header",align="center"] |========================================================= |Field |Default |Description |rsc-role |Started |An additional attribute of colocation constraints that specifies the role that +rsc+ must be in. Allowed values: +Started+, +Master+, +Slave+. indexterm:[rsc-role,Ordering Constraints] indexterm:[Constraints,Ordering,rsc-role] |with-rsc-role |Started |An additional attribute of colocation constraints that specifies the role that +with-rsc+ must be in. Allowed values: +Started+, +Master+, +Slave+. indexterm:[with-rsc-role,Ordering Constraints] indexterm:[Constraints,Ordering,with-rsc-role] |========================================================= .Constraints involving promotable clone resources ====== [source,XML] ------- ------- ====== In the example above, +myApp+ will wait until one of the database copies has been started and promoted to master before being started itself on the same node. Only if no copies can be promoted will +myApp+ be prevented from being active. Additionally, the cluster will wait for +myApp+ to be stopped before demoting the database. Colocation of a primitive or group resource with a promotable clone resource means that it can run on any node with an active instance of the promotable clone resource that has the specified role (+master+ or +slave+). In the example above, the cluster will choose a location based on where database is running as a +master+, and if there are multiple +master+ instances it will also factor in +myApp+'s own location preferences when deciding which location to choose. Colocation with regular clones and other promotable clone resources is also possible. In such cases, the set of allowed locations for the +rsc+ clone is (after role filtering) limited to nodes on which the +with-rsc+ promotable clone resource is (or will be) in the specified role. Placement is then performed as normal. ==== Using Promotable Clone Resources in Colocation Sets ==== .Additional colocation set options relevant to promotable clone resources [width="95%",cols="1m,1,<6",options="header",align="center"] |========================================================= |Field |Default |Description |role |Started |The role that 'all members' of the set must be in. Allowed values: +Started+, +Master+, +Slave+. indexterm:[role,Ordering Constraints] indexterm:[Constraints,Ordering,role] |========================================================= In the following example +B+'s master must be located on the same node as +A+'s master. Additionally resources +C+ and +D+ must be located on the same node as +A+'s and +B+'s masters. .Colocate C and D with A's and B's master instances ====== [source,XML] ------- ------- ====== ==== Using Promotable Clone Resources in Ordered Sets ==== .Additional ordered set options relevant to promotable clone resources [width="95%",cols="1m,1,<3",options="header",align="center"] |========================================================= |Field |Default |Description |action |value of +first-action+ |An additional attribute of ordering constraint sets that specifies the action that applies to 'all members' of the set. Allowed values: +start+, +stop+, +promote+, +demote+. indexterm:[action,Ordering Constraints] indexterm:[Constraints,Ordering,action] |========================================================= .Start C and D after first promoting A and B ====== [source,XML] ------- ------- ====== In the above example, +B+ cannot be promoted to a master role until +A+ has been promoted. Additionally, resources +C+ and +D+ must wait until +A+ and +B+ have been promoted before they can start. [[s-clone-stickiness]] === Clone Stickiness === indexterm:[resource-stickiness,Clones] To achieve a stable allocation pattern, clones are slightly sticky by default. If no value for +resource-stickiness+ is provided, the clone will use a value of 1. Being a small value, it causes minimal disturbance to the score calculations of other resources but is enough to prevent Pacemaker from needlessly moving copies around the cluster. [NOTE] ==== For globally unique clones, this may result in multiple instances of the clone staying on a single node, even after another eligible node becomes active (for example, after being put into standby mode then made active again). If you do not want this behavior, specify a +resource-stickiness+ of 0 for the clone temporarily and let the cluster adjust, then set it back to 1 if you want the default behavior to apply again. ==== [IMPORTANT] ==== If +resource-stickiness+ is set in the +rsc_defaults+ section, it will apply to clone instances as well. This means an explicit +resource-stickiness+ of 0 in +rsc_defaults+ works differently from the implicit default used when +resource-stickiness+ is not specified. ==== === Clone Resource Agent Requirements === Any resource can be used as an anonymous clone, as it requires no additional support from the resource agent. Whether it makes sense to do so depends on your resource and its resource agent. ==== Resource Agent Requirements for Globally Unique Clones ==== Globally unique clones require additional support in the resource agent. In particular, it must only respond with +$\{OCF_SUCCESS}+ if the node has that exact instance active. All other probes for instances of the clone should result in +$\{OCF_NOT_RUNNING}+ (or one of the other OCF error codes if they are failed). Individual instances of a clone are identified by appending a colon and a numerical offset, e.g. +apache:2+. Resource agents can find out how many copies there are by examining the +OCF_RESKEY_CRM_meta_clone_max+ environment variable and which instance it is by examining +OCF_RESKEY_CRM_meta_clone+. The resource agent must not make any assumptions (based on +OCF_RESKEY_CRM_meta_clone+) about which numerical instances are active. In particular, the list of active copies will not always be an unbroken sequence, nor always start at 0. ==== Resource Agent Requirements for Promotable Clones ==== Promotable clone resources require two extra actions, +demote+ and +promote+, which are responsible for changing the state of the resource. Like +start+ and +stop+, they should return +$\{OCF_SUCCESS}+ if they completed successfully or a relevant error code if they did not. The states can mean whatever you wish, but when the resource is started, it must come up in the mode called +slave+. From there the cluster will decide which instances to promote to +master+. In addition to the clone requirements for monitor actions, agents must also _accurately_ report which state they are in. The cluster relies on the agent to report its status (including role) accurately and does not indicate to the agent what role it currently believes it to be in. .Role implications of OCF return codes [width="95%",cols="1,<1",options="header",align="center"] |========================================================= |Monitor Return Code |Description |OCF_NOT_RUNNING |Stopped indexterm:[Return Code,OCF_NOT_RUNNING] |OCF_SUCCESS |Running (Slave) indexterm:[Return Code,OCF_SUCCESS] |OCF_RUNNING_MASTER |Running (Master) indexterm:[Return Code,OCF_RUNNING_MASTER] |OCF_FAILED_MASTER |Failed (Master) indexterm:[Return Code,OCF_FAILED_MASTER] |Other |Failed (Slave) |========================================================= ==== Clone Notifications ==== If the clone has the +notify+ meta-attribute set to +true+, and the resource agent supports the +notify+ action, Pacemaker will call the action when appropriate, passing a number of extra variables which, when combined with additional context, can be used to calculate the current state of the cluster and what is about to happen to it. .Environment variables supplied with Clone notify actions [width="95%",cols="5,<3",options="header",align="center"] |========================================================= |Variable |Description |OCF_RESKEY_CRM_meta_notify_type |Allowed values: +pre+, +post+ indexterm:[Environment Variable,OCF_RESKEY_CRM_meta_notify_,type] indexterm:[type,Notification Environment Variable] |OCF_RESKEY_CRM_meta_notify_operation |Allowed values: +start+, +stop+ indexterm:[Environment Variable,OCF_RESKEY_CRM_meta_notify_,operation] indexterm:[operation,Notification Environment Variable] |OCF_RESKEY_CRM_meta_notify_start_resource |Resources to be started indexterm:[Environment Variable,OCF_RESKEY_CRM_meta_notify_,start_resource] indexterm:[start_resource,Notification Environment Variable] |OCF_RESKEY_CRM_meta_notify_stop_resource |Resources to be stopped indexterm:[Environment Variable,OCF_RESKEY_CRM_meta_notify_,stop_resource] indexterm:[stop_resource,Notification Environment Variable] |OCF_RESKEY_CRM_meta_notify_active_resource |Resources that are running indexterm:[Environment Variable,OCF_RESKEY_CRM_meta_notify_,active_resource] indexterm:[active_resource,Notification Environment Variable] |OCF_RESKEY_CRM_meta_notify_inactive_resource |Resources that are not running indexterm:[Environment Variable,OCF_RESKEY_CRM_meta_notify_,inactive_resource] indexterm:[inactive_resource,Notification Environment Variable] |OCF_RESKEY_CRM_meta_notify_start_uname |Nodes on which resources will be started indexterm:[Environment Variable,OCF_RESKEY_CRM_meta_notify_,start_uname] indexterm:[start_uname,Notification Environment Variable] |OCF_RESKEY_CRM_meta_notify_stop_uname |Nodes on which resources will be stopped indexterm:[Environment Variable,OCF_RESKEY_CRM_meta_notify_,stop_uname] indexterm:[stop_uname,Notification Environment Variable] |OCF_RESKEY_CRM_meta_notify_active_uname |Nodes on which resources are running indexterm:[Environment Variable,OCF_RESKEY_CRM_meta_notify_,active_uname] indexterm:[active_uname,Notification Environment Variable] |========================================================= The variables come in pairs, such as +OCF_RESKEY_CRM_meta_notify_start_resource+ and +OCF_RESKEY_CRM_meta_notify_start_uname+, and should be treated as an array of whitespace-separated elements. +OCF_RESKEY_CRM_meta_notify_inactive_resource+ is an exception, as the matching +uname+ variable does not exist since inactive resources are not running on any node. Thus, in order to indicate that +clone:0+ will be started on +sles-1+, +clone:2+ will be started on +sles-3+, and +clone:3+ will be started on +sles-2+, the cluster would set: .Notification variables ====== [source,Bash] ------- OCF_RESKEY_CRM_meta_notify_start_resource="clone:0 clone:2 clone:3" OCF_RESKEY_CRM_meta_notify_start_uname="sles-1 sles-3 sles-2" ------- ====== [NOTE] ==== Pacemaker will log but otherwise ignore failures of notify actions. ==== ==== Interpretation of Notification Variables ==== .Pre-notification (stop): * Active resources: +$OCF_RESKEY_CRM_meta_notify_active_resource+ * Inactive resources: +$OCF_RESKEY_CRM_meta_notify_inactive_resource+ * Resources to be started: +$OCF_RESKEY_CRM_meta_notify_start_resource+ * Resources to be stopped: +$OCF_RESKEY_CRM_meta_notify_stop_resource+ .Post-notification (stop) / Pre-notification (start): * Active resources ** +$OCF_RESKEY_CRM_meta_notify_active_resource+ ** minus +$OCF_RESKEY_CRM_meta_notify_stop_resource+ * Inactive resources ** +$OCF_RESKEY_CRM_meta_notify_inactive_resource+ ** plus +$OCF_RESKEY_CRM_meta_notify_stop_resource+ * Resources that were started: +$OCF_RESKEY_CRM_meta_notify_start_resource+ * Resources that were stopped: +$OCF_RESKEY_CRM_meta_notify_stop_resource+ .Post-notification (start): * Active resources: ** +$OCF_RESKEY_CRM_meta_notify_active_resource+ ** minus +$OCF_RESKEY_CRM_meta_notify_stop_resource+ ** plus +$OCF_RESKEY_CRM_meta_notify_start_resource+ * Inactive resources: ** +$OCF_RESKEY_CRM_meta_notify_inactive_resource+ ** plus +$OCF_RESKEY_CRM_meta_notify_stop_resource+ ** minus +$OCF_RESKEY_CRM_meta_notify_start_resource+ * Resources that were started: +$OCF_RESKEY_CRM_meta_notify_start_resource+ * Resources that were stopped: +$OCF_RESKEY_CRM_meta_notify_stop_resource+ ==== Extra Notifications for Promotable Clones ==== .Extra environment variables supplied for promotable clones [width="95%",cols="5,<3",options="header",align="center"] |========================================================= |Variable |Description |OCF_RESKEY_CRM_meta_notify_master_resource |Resources that are running in +Master+ mode indexterm:[Environment Variable,OCF_RESKEY_CRM_meta_notify_,master_resource] indexterm:[master_resource,Notification Environment Variable] |OCF_RESKEY_CRM_meta_notify_slave_resource |Resources that are running in +Slave+ mode indexterm:[Environment Variable,OCF_RESKEY_CRM_meta_notify_,slave_resource] indexterm:[slave_resource,Notification Environment Variable] |OCF_RESKEY_CRM_meta_notify_promote_resource |Resources to be promoted indexterm:[Environment Variable,OCF_RESKEY_CRM_meta_notify_,promote_resource] indexterm:[promote_resource,Notification Environment Variable] |OCF_RESKEY_CRM_meta_notify_demote_resource |Resources to be demoted indexterm:[Environment Variable,OCF_RESKEY_CRM_meta_notify_,demote_resource] indexterm:[demote_resource,Notification Environment Variable] |OCF_RESKEY_CRM_meta_notify_promote_uname |Nodes on which resources will be promoted indexterm:[Environment Variable,OCF_RESKEY_CRM_meta_notify_,promote_uname] indexterm:[promote_uname,Notification Environment Variable] |OCF_RESKEY_CRM_meta_notify_demote_uname |Nodes on which resources will be demoted indexterm:[Environment Variable,OCF_RESKEY_CRM_meta_notify_,demote_uname] indexterm:[demote_uname,Notification Environment Variable] |OCF_RESKEY_CRM_meta_notify_master_uname |Nodes on which resources are running in +Master+ mode indexterm:[Environment Variable,OCF_RESKEY_CRM_meta_notify_,master_uname] indexterm:[master_uname,Notification Environment Variable] |OCF_RESKEY_CRM_meta_notify_slave_uname |Nodes on which resources are running in +Slave+ mode indexterm:[Environment Variable,OCF_RESKEY_CRM_meta_notify_,slave_uname] indexterm:[slave_uname,Notification Environment Variable] |========================================================= ==== Interpretation of Promotable Notification Variables ==== .Pre-notification (demote): * +Active+ resources: +$OCF_RESKEY_CRM_meta_notify_active_resource+ * +Master+ resources: +$OCF_RESKEY_CRM_meta_notify_master_resource+ * +Slave+ resources: +$OCF_RESKEY_CRM_meta_notify_slave_resource+ * Inactive resources: +$OCF_RESKEY_CRM_meta_notify_inactive_resource+ * Resources to be started: +$OCF_RESKEY_CRM_meta_notify_start_resource+ * Resources to be promoted: +$OCF_RESKEY_CRM_meta_notify_promote_resource+ * Resources to be demoted: +$OCF_RESKEY_CRM_meta_notify_demote_resource+ * Resources to be stopped: +$OCF_RESKEY_CRM_meta_notify_stop_resource+ .Post-notification (demote) / Pre-notification (stop): * +Active+ resources: +$OCF_RESKEY_CRM_meta_notify_active_resource+ * +Master+ resources: ** +$OCF_RESKEY_CRM_meta_notify_master_resource+ ** minus +$OCF_RESKEY_CRM_meta_notify_demote_resource+ * +Slave+ resources: +$OCF_RESKEY_CRM_meta_notify_slave_resource+ * Inactive resources: +$OCF_RESKEY_CRM_meta_notify_inactive_resource+ * Resources to be started: +$OCF_RESKEY_CRM_meta_notify_start_resource+ * Resources to be promoted: +$OCF_RESKEY_CRM_meta_notify_promote_resource+ * Resources to be demoted: +$OCF_RESKEY_CRM_meta_notify_demote_resource+ * Resources to be stopped: +$OCF_RESKEY_CRM_meta_notify_stop_resource+ * Resources that were demoted: +$OCF_RESKEY_CRM_meta_notify_demote_resource+ .Post-notification (stop) / Pre-notification (start) * +Active+ resources: ** +$OCF_RESKEY_CRM_meta_notify_active_resource+ ** minus +$OCF_RESKEY_CRM_meta_notify_stop_resource+ * +Master+ resources: ** +$OCF_RESKEY_CRM_meta_notify_master_resource+ ** minus +$OCF_RESKEY_CRM_meta_notify_demote_resource+ * +Slave+ resources: ** +$OCF_RESKEY_CRM_meta_notify_slave_resource+ ** minus +$OCF_RESKEY_CRM_meta_notify_stop_resource+ * Inactive resources: ** +$OCF_RESKEY_CRM_meta_notify_inactive_resource+ ** plus +$OCF_RESKEY_CRM_meta_notify_stop_resource+ * Resources to be started: +$OCF_RESKEY_CRM_meta_notify_start_resource+ * Resources to be promoted: +$OCF_RESKEY_CRM_meta_notify_promote_resource+ * Resources to be demoted: +$OCF_RESKEY_CRM_meta_notify_demote_resource+ * Resources to be stopped: +$OCF_RESKEY_CRM_meta_notify_stop_resource+ * Resources that were demoted: +$OCF_RESKEY_CRM_meta_notify_demote_resource+ * Resources that were stopped: +$OCF_RESKEY_CRM_meta_notify_stop_resource+ .Post-notification (start) / Pre-notification (promote) * +Active+ resources: ** +$OCF_RESKEY_CRM_meta_notify_active_resource+ ** minus +$OCF_RESKEY_CRM_meta_notify_stop_resource+ ** plus +$OCF_RESKEY_CRM_meta_notify_start_resource+ * +Master+ resources: ** +$OCF_RESKEY_CRM_meta_notify_master_resource+ ** minus +$OCF_RESKEY_CRM_meta_notify_demote_resource+ * +Slave+ resources: ** +$OCF_RESKEY_CRM_meta_notify_slave_resource+ ** minus +$OCF_RESKEY_CRM_meta_notify_stop_resource+ ** plus +$OCF_RESKEY_CRM_meta_notify_start_resource+ * Inactive resources: ** +$OCF_RESKEY_CRM_meta_notify_inactive_resource+ ** plus +$OCF_RESKEY_CRM_meta_notify_stop_resource+ ** minus +$OCF_RESKEY_CRM_meta_notify_start_resource+ * Resources to be started: +$OCF_RESKEY_CRM_meta_notify_start_resource+ * Resources to be promoted: +$OCF_RESKEY_CRM_meta_notify_promote_resource+ * Resources to be demoted: +$OCF_RESKEY_CRM_meta_notify_demote_resource+ * Resources to be stopped: +$OCF_RESKEY_CRM_meta_notify_stop_resource+ * Resources that were started: +$OCF_RESKEY_CRM_meta_notify_start_resource+ * Resources that were demoted: +$OCF_RESKEY_CRM_meta_notify_demote_resource+ * Resources that were stopped: +$OCF_RESKEY_CRM_meta_notify_stop_resource+ .Post-notification (promote) * +Active+ resources: ** +$OCF_RESKEY_CRM_meta_notify_active_resource+ ** minus +$OCF_RESKEY_CRM_meta_notify_stop_resource+ ** plus +$OCF_RESKEY_CRM_meta_notify_start_resource+ * +Master+ resources: ** +$OCF_RESKEY_CRM_meta_notify_master_resource+ ** minus +$OCF_RESKEY_CRM_meta_notify_demote_resource+ ** plus +$OCF_RESKEY_CRM_meta_notify_promote_resource+ * +Slave+ resources: ** +$OCF_RESKEY_CRM_meta_notify_slave_resource+ ** minus +$OCF_RESKEY_CRM_meta_notify_stop_resource+ ** plus +$OCF_RESKEY_CRM_meta_notify_start_resource+ ** minus +$OCF_RESKEY_CRM_meta_notify_promote_resource+ * Inactive resources: ** +$OCF_RESKEY_CRM_meta_notify_inactive_resource+ ** plus +$OCF_RESKEY_CRM_meta_notify_stop_resource+ ** minus +$OCF_RESKEY_CRM_meta_notify_start_resource+ * Resources to be started: +$OCF_RESKEY_CRM_meta_notify_start_resource+ * Resources to be promoted: +$OCF_RESKEY_CRM_meta_notify_promote_resource+ * Resources to be demoted: +$OCF_RESKEY_CRM_meta_notify_demote_resource+ * Resources to be stopped: +$OCF_RESKEY_CRM_meta_notify_stop_resource+ * Resources that were started: +$OCF_RESKEY_CRM_meta_notify_start_resource+ * Resources that were promoted: +$OCF_RESKEY_CRM_meta_notify_promote_resource+ * Resources that were demoted: +$OCF_RESKEY_CRM_meta_notify_demote_resource+ * Resources that were stopped: +$OCF_RESKEY_CRM_meta_notify_stop_resource+ === Monitoring Promotable Clone Resources === The usual monitor actions are insufficient to monitor a promotable clone resource, because Pacemaker needs to verify not only that the resource is active, but also that its actual role matches its intended one. Define two monitoring actions: the usual one will cover the slave role, and an additional one with +role="master"+ will cover the master role. .Monitoring both states of a promotable clone resource ====== [source,XML] ------- ------- ====== [IMPORTANT] =========== It is crucial that _every_ monitor operation has a different interval! Pacemaker currently differentiates between operations only by resource and interval; so if (for example) a promotable clone resource had the same monitor interval for both roles, Pacemaker would ignore the role when checking the status -- which would cause unexpected return codes, and therefore unnecessary complications. =========== [[s-promotion-scores]] === Determining Which Instance is Promoted === Pacemaker can choose a promotable clone instance to be promoted in one of two ways: * Promotion scores: These are node attributes set via the `crm_master` utility, which generally would be called by the resource agent's start action if it supports promotable clones. This tool automatically detects both the resource and host, and should be used to set a preference for being promoted. Based on this, +promoted-max+, and +promoted-node-max+, the instance(s) with the highest preference will be promoted. * Constraints: Location constraints can indicate which nodes are most preferred as masters. .Explicitly preferring node1 to be promoted to master ====== [source,XML] ------- ------- ====== [[s-resource-bundle]] == Bundles - Isolated Environments == indexterm:[bundle] indexterm:[Resource,bundle] indexterm:[container,bundle] indexterm:[Docker,bundle] indexterm:[podman,bundle] indexterm:[rkt,bundle] Pacemaker supports a special syntax for launching a https://en.wikipedia.org/wiki/Operating-system-level_virtualization[container] with any infrastructure it requires: the 'bundle'. Pacemaker bundles support https://www.docker.com/[Docker], https://podman.io/[podman], and https://coreos.com/rkt/[rkt] container technologies. footnote:[Docker is a trademark of Docker, Inc. No endorsement by or association with Docker, Inc. is implied.] .A bundle for a containerized web server ==== [source,XML] ---- + options="rw,Z"/> + options="rw,Z"/> ---- ==== === Bundle Prerequisites === indexterm:[bundle,prerequisites] Before configuring a bundle in Pacemaker, the user must install the appropriate container launch technology (Docker, podman, or rkt), and supply a fully configured container image, on every node allowed to run the bundle. Pacemaker will create an implicit resource of type +ocf:heartbeat:docker+, +ocf:heartbeat:podman+, or +ocf:heartbeat:rkt+ to manage a bundle's container. The user must ensure that the appropriate resource agent is installed on every node allowed to run the bundle. === Bundle Properties === .Properties of a Bundle [width="95%",cols="3m,<5",options="header",align="center"] |========================================================= |Field |Description |id |A unique name for the bundle (required) indexterm:[id,bundle] indexterm:[bundle,Property,id] |description |Arbitrary text (not used by Pacemaker) indexterm:[description,bundle] indexterm:[bundle,Property,description] |========================================================= A bundle must contain exactly one ++, ++, or ++ element. === Bundle Container Properties === indexterm:[container,Properties] indexterm:[Docker,Properties] indexterm:[podman,Properties] indexterm:[rkt,Properties] .Properties of a Bundle's docker, podman, or rkt Element [width="95%",cols="3m,4,<5",options="header",align="center"] |==== |Field |Default |Description |image | |Container image tag (required) indexterm:[image,container] indexterm:[container,Property,image] |replicas |Value of +promoted-max+ if that is positive, else 1 |A positive integer specifying the number of container instances to launch indexterm:[replicas,container] indexterm:[container,Property,replicas] |replicas-per-host |1 |A positive integer specifying the number of container instances allowed to run on a single node indexterm:[replicas-per-host,container] indexterm:[container,Property,replicas-per-host] |promoted-max |0 |A non-negative integer that, if positive, indicates that the containerized service should be treated as a promotable service, with this many replicas allowed to run the service in the master role indexterm:[promoted-max,container] indexterm:[container,Property,promoted-max] |network | |If specified, this will be passed to the `docker run`, `podman run`, or `rkt run` command as the network setting for the container. indexterm:[network,container] indexterm:[container,Property,network] |run-command |`/usr/sbin/pacemaker-remoted` if bundle contains a +primitive+, otherwise none |This command will be run inside the container when launching it ("PID 1"). If the bundle contains a +primitive+, this command 'must' start pacemaker-remoted (but could, for example, be a script that does other stuff, too). indexterm:[run-command,container] indexterm:[container,Property,run-command] |options | |Extra command-line options to pass to the `docker run`, `podman run`, or `rkt run` command indexterm:[options,container] indexterm:[container,Property,options] |==== [NOTE] ==== Considerations when using cluster configurations or container images from Pacemaker 1.1: - If the container image has a pre-2.0.0 version of Pacemaker, set +run-command+ to +/usr/sbin/pacemaker_remoted+ (note the underbar instead of dash). - +masters+ is accepted as an alias for +promoted-max+, but is deprecated since 2.0.0, and support for it will be removed in a future version. ==== === Bundle Network Properties === A bundle may optionally contain one ++ element. indexterm:[bundle,network] .Properties of a Bundle's Network Element [width="95%",cols="2m,1,<4",options="header",align="center"] |========================================================= |Field |Default |Description |add-host |TRUE |If TRUE, and +ip-range-start+ is used, Pacemaker will automatically ensure that +/etc/hosts+ inside the containers has entries for each <> and its assigned IP. indexterm:[add-host,network] indexterm:[network,Property,add-host] |ip-range-start | |If specified, Pacemaker will create an implicit +ocf:heartbeat:IPaddr2+ resource for each container instance, starting with this IP address, using up to +replicas+ sequential addresses. These addresses can be used from the host's network to reach the service inside the container, though it is not visible within the container itself. Only IPv4 addresses are currently supported. indexterm:[ip-range-start,network] indexterm:[network,Property,ip-range-start] |host-netmask |32 |If +ip-range-start+ is specified, the IP addresses are created with this CIDR netmask (as a number of bits). indexterm:[host-netmask,network] indexterm:[network,Property,host-netmask] |host-interface | |If +ip-range-start+ is specified, the IP addresses are created on this host interface (by default, it will be determined from the IP address). indexterm:[host-interface,network] indexterm:[network,Property,host-interface] |control-port |3121 |If the bundle contains a +primitive+, the cluster will use this integer TCP port for communication with Pacemaker Remote inside the container. Changing this is useful when the container is unable to listen on the default port, for example, when the container uses the host's network rather than +ip-range-start+ (in which case +replicas-per-host+ must be 1), or when the bundle may run on a Pacemaker Remote node that is already listening on the default port. Any PCMK_remote_port environment variable set on the host or in the container is ignored for bundle connections. indexterm:[control-port,network] indexterm:[network,Property,control-port] |========================================================= [[s-resource-bundle-note-replica-names]] [NOTE] ==== Replicas are named by the bundle id plus a dash and an integer counter starting with zero. For example, if a bundle named +httpd-bundle+ has +replicas=2+, its containers will be named +httpd-bundle-0+ and +httpd-bundle-1+. ==== Additionally, a ++ element may optionally contain one or more ++ elements. indexterm:[bundle,network,port-mapping] .Properties of a Bundle's Port-Mapping Element [width="95%",cols="2m,1,<4",options="header",align="center"] |========================================================= |Field |Default |Description |id | |A unique name for the port mapping (required) indexterm:[id,port-mapping] indexterm:[port-mapping,Property,id] |port | |If this is specified, connections to this TCP port number on the host network (on the container's assigned IP address, if +ip-range-start+ is specified) will be forwarded to the container network. Exactly one of +port+ or +range+ must be specified in a +port-mapping+. indexterm:[port,port-mapping] indexterm:[port-mapping,Property,port] |internal-port |value of +port+ |If +port+ and this are specified, connections to +port+ on the host's network will be forwarded to this port on the container network. indexterm:[internal-port,port-mapping] indexterm:[port-mapping,Property,internal-port] |range | |If this is specified, connections to these TCP port numbers (expressed as 'first_port'-'last_port') on the host network (on the container's assigned IP address, if +ip-range-start+ is specified) will be forwarded to the same ports in the container network. Exactly one of +port+ or +range+ must be specified in a +port-mapping+. indexterm:[range,port-mapping] indexterm:[port-mapping,Property,range] |========================================================= [NOTE] ==== If the bundle contains a +primitive+, Pacemaker will automatically map the +control-port+, so it is not necessary to specify that port in a +port-mapping+. ==== === Bundle Storage Properties === A bundle may optionally contain one ++ element. A ++ element has no properties of its own, but may contain one or more ++ elements. indexterm:[bundle,storage,storage-mapping] .Properties of a Bundle's Storage-Mapping Element [width="95%",cols="2m,1,<4",options="header",align="center"] |========================================================= |Field |Default |Description |id | |A unique name for the storage mapping (required) indexterm:[id,storage-mapping] indexterm:[storage-mapping,Property,id] |source-dir | |The absolute path on the host's filesystem that will be mapped into the container. Exactly one of +source-dir+ and +source-dir-root+ must be specified in a +storage-mapping+. indexterm:[source-dir,storage-mapping] indexterm:[storage-mapping,Property,source-dir] |source-dir-root | |The start of a path on the host's filesystem that will be mapped into the container, using a different subdirectory on the host for each container instance. The subdirectory will be named the same as the <>. Exactly one of +source-dir+ and +source-dir-root+ must be specified in a +storage-mapping+. indexterm:[source-dir-root,storage-mapping] indexterm:[storage-mapping,Property,source-dir-root] |target-dir | |The path name within the container where the host storage will be mapped (required) indexterm:[target-dir,storage-mapping] indexterm:[storage-mapping,Property,target-dir] |options | -|File system mount options to use when mapping the storage +|A comma-separated list of file system mount options to use when mapping the + storage indexterm:[options,storage-mapping] indexterm:[storage-mapping,Property,options] |========================================================= [NOTE] ==== Pacemaker does not define the behavior if the source directory does not already exist on the host. However, it is expected that the container technology and/or its resource agent will create the source directory in that case. ==== [NOTE] ==== If the bundle contains a +primitive+, Pacemaker will automatically map the equivalent of +source-dir=/etc/pacemaker/authkey target-dir=/etc/pacemaker/authkey+ and +source-dir-root=/var/log/pacemaker/bundles target-dir=/var/log+ into the container, so it is not necessary to specify those paths in a +storage-mapping+. ==== [IMPORTANT] ==== The +PCMK_authkey_location+ environment variable must not be set to anything other than the default of `/etc/pacemaker/authkey` on any node in the cluster. ==== +[IMPORTANT] +==== +If SELinux is used in enforcing mode on the host, you must ensure the container +is allowed to use any storage you mount into it. For Docker and podman bundles, +adding "Z" to the mount options will create a container-specific label for the +mount that allows the container access. +==== + === Bundle Primitive === A bundle may optionally contain one ++ resource (see <>). The primitive may have operations, instance attributes and meta-attributes defined, as usual. If a bundle contains a primitive resource, the container image must include the Pacemaker Remote daemon, and at least one of +ip-range-start+ or +control-port+ must be configured in the bundle. Pacemaker will create an implicit +ocf:pacemaker:remote+ resource for the connection, launch Pacemaker Remote within the container, and monitor and manage the primitive resource via Pacemaker Remote. If the bundle has more than one container instance (replica), the primitive resource will function as an implicit clone (see <>) -- a promotable clone if the bundle has +masters+ greater than zero (see <>). [IMPORTANT] ==== Containers in bundles with a +primitive+ must have an accessible networking environment, so that Pacemaker on the cluster nodes can contact Pacemaker Remote inside the container. For example, the Docker option `--net=none` should not be used with a +primitive+. The default (using a distinct network space inside the container) works in combination with +ip-range-start+. If the Docker option `--net=host` is used (making the container share the host's network space), a unique +control-port+ should be specified for each bundle. Any firewall must allow access to the +control-port+. ==== [[s-bundle-attributes]] === Bundle Node Attributes === If the bundle has a +primitive+, the primitive's resource agent may want to set node attributes such as <>. However, with containers, it is not apparent which node should get the attribute. If the container uses shared storage that is the same no matter which node the container is hosted on, then it is appropriate to use the promotion score on the bundle node itself. On the other hand, if the container uses storage exported from the underlying host, then it may be more appropriate to use the promotion score on the underlying host. Since this depends on the particular situation, the +container-attribute-target+ resource meta-attribute allows the user to specify which approach to use. If it is set to +host+, then user-defined node attributes will be checked on the underlying host. If it is anything else, the local node (in this case the bundle node) is used as usual. This only applies to user-defined attributes; the cluster will always check the local node for cluster-defined attributes such as +#uname+. If +container-attribute-target+ is +host+, the cluster will pass additional environment variables to the primitive's resource agent that allow it to set node attributes appropriately: +CRM_meta_container_attribute_target+ (identical to the meta-attribute value) and +CRM_meta_physical_host+ (the name of the underlying host). [NOTE] ==== When called by a resource agent, the attrd_updater and crm_attribute commands will automatically check those environment variables and set attributes appropriately. ==== === Bundle Meta-Attributes === Any meta-attribute set on a bundle will be inherited by the bundle's primitive and any resources implicitly created by Pacemaker for the bundle. This includes options such as +priority+, +target-role+, and +is-managed+. See <> for more information. === Limitations of Bundles === Restarting pacemaker while a bundle is unmanaged or the cluster is in maintenance mode may cause the bundle to fail. Bundles may not be explicitly cloned or included in groups. This includes the bundle's primitive and any resources implicitly created by Pacemaker for the bundle. (If +replicas+ is greater than 1, the bundle will behave like a clone implicitly.) Bundles do not have instance attributes, utilization attributes, or operations, though a bundle's primitive may have them. A bundle with a primitive can run on a Pacemaker Remote node only if the bundle uses a distinct +control-port+. diff --git a/lib/pengine/bundle.c b/lib/pengine/bundle.c index 3b32f0454b..b223f03c04 100644 --- a/lib/pengine/bundle.c +++ b/lib/pengine/bundle.c @@ -1,1681 +1,1682 @@ /* * Copyright 2004-2019 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 #define PE__VARIANT_BUNDLE 1 #include "./variant.h" static char * next_ip(const char *last_ip) { unsigned int oct1 = 0; unsigned int oct2 = 0; unsigned int oct3 = 0; unsigned int oct4 = 0; int rc = sscanf(last_ip, "%u.%u.%u.%u", &oct1, &oct2, &oct3, &oct4); if (rc != 4) { /*@ TODO check for IPv6 */ return NULL; } else if (oct3 > 253) { return NULL; } else if (oct4 > 253) { ++oct3; oct4 = 1; } else { ++oct4; } return crm_strdup_printf("%u.%u.%u.%u", oct1, oct2, oct3, oct4); } static int allocate_ip(pe__bundle_variant_data_t *data, pe__bundle_replica_t *replica, char *buffer, int max) { if(data->ip_range_start == NULL) { return 0; } else if(data->ip_last) { replica->ipaddr = next_ip(data->ip_last); } else { replica->ipaddr = strdup(data->ip_range_start); } data->ip_last = replica->ipaddr; switch (data->agent_type) { case PE__CONTAINER_AGENT_DOCKER: case PE__CONTAINER_AGENT_PODMAN: if (data->add_host) { return snprintf(buffer, max, " --add-host=%s-%d:%s", data->prefix, replica->offset, replica->ipaddr); } case PE__CONTAINER_AGENT_RKT: return snprintf(buffer, max, " --hosts-entry=%s=%s-%d", replica->ipaddr, data->prefix, replica->offset); default: // PE__CONTAINER_AGENT_UNKNOWN break; } return 0; } static xmlNode * create_resource(const char *name, const char *provider, const char *kind) { xmlNode *rsc = create_xml_node(NULL, XML_CIB_TAG_RESOURCE); crm_xml_add(rsc, XML_ATTR_ID, name); crm_xml_add(rsc, XML_AGENT_ATTR_CLASS, PCMK_RESOURCE_CLASS_OCF); crm_xml_add(rsc, XML_AGENT_ATTR_PROVIDER, provider); crm_xml_add(rsc, XML_ATTR_TYPE, kind); return rsc; } /*! * \internal * \brief Check whether cluster can manage resource inside container * * \param[in] data Container variant data * * \return TRUE if networking configuration is acceptable, FALSE otherwise * * \note The resource is manageable if an IP range or control port has been * specified. If a control port is used without an IP range, replicas per * host must be 1. */ static bool valid_network(pe__bundle_variant_data_t *data) { if(data->ip_range_start) { return TRUE; } if(data->control_port) { if(data->nreplicas_per_host > 1) { pe_err("Specifying the 'control-port' for %s requires 'replicas-per-host=1'", data->prefix); data->nreplicas_per_host = 1; /* @TODO to be sure: clear_bit(rsc->flags, pe_rsc_unique); */ } return TRUE; } return FALSE; } static bool create_ip_resource(pe_resource_t *parent, pe__bundle_variant_data_t *data, pe__bundle_replica_t *replica, pe_working_set_t *data_set) { if(data->ip_range_start) { char *id = NULL; xmlNode *xml_ip = NULL; xmlNode *xml_obj = NULL; id = crm_strdup_printf("%s-ip-%s", data->prefix, replica->ipaddr); crm_xml_sanitize_id(id); xml_ip = create_resource(id, "heartbeat", "IPaddr2"); free(id); xml_obj = create_xml_node(xml_ip, XML_TAG_ATTR_SETS); crm_xml_set_id(xml_obj, "%s-attributes-%d", data->prefix, replica->offset); crm_create_nvpair_xml(xml_obj, NULL, "ip", replica->ipaddr); if(data->host_network) { crm_create_nvpair_xml(xml_obj, NULL, "nic", data->host_network); } if(data->host_netmask) { crm_create_nvpair_xml(xml_obj, NULL, "cidr_netmask", data->host_netmask); } else { crm_create_nvpair_xml(xml_obj, NULL, "cidr_netmask", "32"); } xml_obj = create_xml_node(xml_ip, "operations"); crm_create_op_xml(xml_obj, ID(xml_ip), "monitor", "60s", NULL); // TODO: Other ops? Timeouts and intervals from underlying resource? if (!common_unpack(xml_ip, &replica->ip, parent, data_set)) { return FALSE; } parent->children = g_list_append(parent->children, replica->ip); } return TRUE; } static bool create_docker_resource(pe_resource_t *parent, pe__bundle_variant_data_t *data, pe__bundle_replica_t *replica, pe_working_set_t *data_set) { int offset = 0, max = 4096; char *buffer = calloc(1, max+1); int doffset = 0, dmax = 1024; char *dbuffer = calloc(1, dmax+1); char *id = NULL; xmlNode *xml_container = NULL; xmlNode *xml_obj = NULL; id = crm_strdup_printf("%s-docker-%d", data->prefix, replica->offset); crm_xml_sanitize_id(id); xml_container = create_resource(id, "heartbeat", PE__CONTAINER_AGENT_DOCKER_S); free(id); xml_obj = create_xml_node(xml_container, XML_TAG_ATTR_SETS); crm_xml_set_id(xml_obj, "%s-attributes-%d", data->prefix, replica->offset); crm_create_nvpair_xml(xml_obj, NULL, "image", data->image); crm_create_nvpair_xml(xml_obj, NULL, "allow_pull", XML_BOOLEAN_TRUE); crm_create_nvpair_xml(xml_obj, NULL, "force_kill", XML_BOOLEAN_FALSE); crm_create_nvpair_xml(xml_obj, NULL, "reuse", XML_BOOLEAN_FALSE); offset += snprintf(buffer+offset, max-offset, " --restart=no"); /* Set a container hostname only if we have an IP to map it to. * The user can set -h or --uts=host themselves if they want a nicer * name for logs, but this makes applications happy who need their * hostname to match the IP they bind to. */ if (data->ip_range_start != NULL) { offset += snprintf(buffer+offset, max-offset, " -h %s-%d", data->prefix, replica->offset); } offset += snprintf(buffer+offset, max-offset, " -e PCMK_stderr=1"); if (data->container_network) { #if 0 offset += snprintf(buffer+offset, max-offset, " --link-local-ip=%s", replica->ipaddr); #endif offset += snprintf(buffer+offset, max-offset, " --net=%s", data->container_network); } if(data->control_port) { offset += snprintf(buffer+offset, max-offset, " -e PCMK_remote_port=%s", data->control_port); } else { offset += snprintf(buffer+offset, max-offset, " -e PCMK_remote_port=%d", DEFAULT_REMOTE_PORT); } for(GListPtr pIter = data->mounts; pIter != NULL; pIter = pIter->next) { pe__bundle_mount_t *mount = pIter->data; - if(mount->flags) { + if (is_set(mount->flags, pe__bundle_mount_subdir)) { char *source = crm_strdup_printf( "%s/%s-%d", mount->source, data->prefix, replica->offset); if(doffset > 0) { doffset += snprintf(dbuffer+doffset, dmax-doffset, ","); } doffset += snprintf(dbuffer+doffset, dmax-doffset, "%s", source); offset += snprintf(buffer+offset, max-offset, " -v %s:%s", source, mount->target); free(source); } else { offset += snprintf(buffer+offset, max-offset, " -v %s:%s", mount->source, mount->target); } if(mount->options) { offset += snprintf(buffer+offset, max-offset, ":%s", mount->options); } } for(GListPtr pIter = data->ports; pIter != NULL; pIter = pIter->next) { pe__bundle_port_t *port = pIter->data; if (replica->ipaddr) { offset += snprintf(buffer+offset, max-offset, " -p %s:%s:%s", replica->ipaddr, port->source, port->target); } else if(safe_str_neq(data->container_network, "host")) { // No need to do port mapping if net=host offset += snprintf(buffer+offset, max-offset, " -p %s:%s", port->source, port->target); } } if (data->launcher_options) { offset += snprintf(buffer+offset, max-offset, " %s", data->launcher_options); } if (data->container_host_options) { offset += snprintf(buffer + offset, max - offset, " %s", data->container_host_options); } crm_create_nvpair_xml(xml_obj, NULL, "run_opts", buffer); free(buffer); crm_create_nvpair_xml(xml_obj, NULL, "mount_points", dbuffer); free(dbuffer); if (replica->child) { if (data->container_command) { crm_create_nvpair_xml(xml_obj, NULL, "run_cmd", data->container_command); } else { crm_create_nvpair_xml(xml_obj, NULL, "run_cmd", SBIN_DIR "/pacemaker-remoted"); } /* TODO: Allow users to specify their own? * * We just want to know if the container is alive, we'll * monitor the child independently */ crm_create_nvpair_xml(xml_obj, NULL, "monitor_cmd", "/bin/true"); /* } else if(child && data->untrusted) { * Support this use-case? * * The ability to have resources started/stopped by us, but * unable to set attributes, etc. * * Arguably better to control API access this with ACLs like * "normal" remote nodes * * crm_create_nvpair_xml(xml_obj, NULL, * "run_cmd", * "/usr/libexec/pacemaker/pacemaker-execd"); * crm_create_nvpair_xml(xml_obj, NULL, "monitor_cmd", * "/usr/libexec/pacemaker/lrmd_internal_ctl -c poke"); */ } else { if (data->container_command) { crm_create_nvpair_xml(xml_obj, NULL, "run_cmd", data->container_command); } /* TODO: Allow users to specify their own? * * We don't know what's in the container, so we just want * to know if it is alive */ crm_create_nvpair_xml(xml_obj, NULL, "monitor_cmd", "/bin/true"); } xml_obj = create_xml_node(xml_container, "operations"); crm_create_op_xml(xml_obj, ID(xml_container), "monitor", "60s", NULL); // TODO: Other ops? Timeouts and intervals from underlying resource? if (!common_unpack(xml_container, &replica->container, parent, data_set)) { return FALSE; } parent->children = g_list_append(parent->children, replica->container); return TRUE; } static bool create_podman_resource(pe_resource_t *parent, pe__bundle_variant_data_t *data, pe__bundle_replica_t *replica, pe_working_set_t *data_set) { int offset = 0, max = 4096; char *buffer = calloc(1, max+1); int doffset = 0, dmax = 1024; char *dbuffer = calloc(1, dmax+1); char *id = NULL; xmlNode *xml_container = NULL; xmlNode *xml_obj = NULL; id = crm_strdup_printf("%s-podman-%d", data->prefix, replica->offset); crm_xml_sanitize_id(id); xml_container = create_resource(id, "heartbeat", PE__CONTAINER_AGENT_PODMAN_S); free(id); xml_obj = create_xml_node(xml_container, XML_TAG_ATTR_SETS); crm_xml_set_id(xml_obj, "%s-attributes-%d", data->prefix, replica->offset); crm_create_nvpair_xml(xml_obj, NULL, "image", data->image); crm_create_nvpair_xml(xml_obj, NULL, "allow_pull", XML_BOOLEAN_TRUE); crm_create_nvpair_xml(xml_obj, NULL, "force_kill", XML_BOOLEAN_FALSE); crm_create_nvpair_xml(xml_obj, NULL, "reuse", XML_BOOLEAN_FALSE); // FIXME: (bandini 2018-08) podman has no restart policies //offset += snprintf(buffer+offset, max-offset, " --restart=no"); /* Set a container hostname only if we have an IP to map it to. * The user can set -h or --uts=host themselves if they want a nicer * name for logs, but this makes applications happy who need their * hostname to match the IP they bind to. */ if (data->ip_range_start != NULL) { offset += snprintf(buffer+offset, max-offset, " -h %s-%d", data->prefix, replica->offset); } offset += snprintf(buffer+offset, max-offset, " -e PCMK_stderr=1"); if (data->container_network) { #if 0 // podman has no support for --link-local-ip offset += snprintf(buffer+offset, max-offset, " --link-local-ip=%s", replica->ipaddr); #endif offset += snprintf(buffer+offset, max-offset, " --net=%s", data->container_network); } if(data->control_port) { offset += snprintf(buffer+offset, max-offset, " -e PCMK_remote_port=%s", data->control_port); } else { offset += snprintf(buffer+offset, max-offset, " -e PCMK_remote_port=%d", DEFAULT_REMOTE_PORT); } for(GListPtr pIter = data->mounts; pIter != NULL; pIter = pIter->next) { pe__bundle_mount_t *mount = pIter->data; - if(mount->flags) { + if (is_set(mount->flags, pe__bundle_mount_subdir)) { char *source = crm_strdup_printf( "%s/%s-%d", mount->source, data->prefix, replica->offset); if(doffset > 0) { doffset += snprintf(dbuffer+doffset, dmax-doffset, ","); } doffset += snprintf(dbuffer+doffset, dmax-doffset, "%s", source); offset += snprintf(buffer+offset, max-offset, " -v %s:%s", source, mount->target); free(source); } else { offset += snprintf(buffer+offset, max-offset, " -v %s:%s", mount->source, mount->target); } if(mount->options) { offset += snprintf(buffer+offset, max-offset, ":%s", mount->options); } } for(GListPtr pIter = data->ports; pIter != NULL; pIter = pIter->next) { pe__bundle_port_t *port = pIter->data; if (replica->ipaddr) { offset += snprintf(buffer+offset, max-offset, " -p %s:%s:%s", replica->ipaddr, port->source, port->target); } else if(safe_str_neq(data->container_network, "host")) { // No need to do port mapping if net=host offset += snprintf(buffer+offset, max-offset, " -p %s:%s", port->source, port->target); } } if (data->launcher_options) { offset += snprintf(buffer+offset, max-offset, " %s", data->launcher_options); } if (data->container_host_options) { offset += snprintf(buffer + offset, max - offset, " %s", data->container_host_options); } crm_create_nvpair_xml(xml_obj, NULL, "run_opts", buffer); free(buffer); crm_create_nvpair_xml(xml_obj, NULL, "mount_points", dbuffer); free(dbuffer); if (replica->child) { if (data->container_command) { crm_create_nvpair_xml(xml_obj, NULL, "run_cmd", data->container_command); } else { crm_create_nvpair_xml(xml_obj, NULL, "run_cmd", SBIN_DIR "/pacemaker-remoted"); } /* TODO: Allow users to specify their own? * * We just want to know if the container is alive, we'll * monitor the child independently */ crm_create_nvpair_xml(xml_obj, NULL, "monitor_cmd", "/bin/true"); /* } else if(child && data->untrusted) { * Support this use-case? * * The ability to have resources started/stopped by us, but * unable to set attributes, etc. * * Arguably better to control API access this with ACLs like * "normal" remote nodes * * crm_create_nvpair_xml(xml_obj, NULL, * "run_cmd", * "/usr/libexec/pacemaker/pacemaker-execd"); * crm_create_nvpair_xml(xml_obj, NULL, "monitor_cmd", * "/usr/libexec/pacemaker/lrmd_internal_ctl -c poke"); */ } else { if (data->container_command) { crm_create_nvpair_xml(xml_obj, NULL, "run_cmd", data->container_command); } /* TODO: Allow users to specify their own? * * We don't know what's in the container, so we just want * to know if it is alive */ crm_create_nvpair_xml(xml_obj, NULL, "monitor_cmd", "/bin/true"); } xml_obj = create_xml_node(xml_container, "operations"); crm_create_op_xml(xml_obj, ID(xml_container), "monitor", "60s", NULL); // TODO: Other ops? Timeouts and intervals from underlying resource? if (!common_unpack(xml_container, &replica->container, parent, data_set)) { return FALSE; } parent->children = g_list_append(parent->children, replica->container); return TRUE; } static bool create_rkt_resource(pe_resource_t *parent, pe__bundle_variant_data_t *data, pe__bundle_replica_t *replica, pe_working_set_t *data_set) { int offset = 0, max = 4096; char *buffer = calloc(1, max+1); int doffset = 0, dmax = 1024; char *dbuffer = calloc(1, dmax+1); char *id = NULL; xmlNode *xml_container = NULL; xmlNode *xml_obj = NULL; int volid = 0; id = crm_strdup_printf("%s-rkt-%d", data->prefix, replica->offset); crm_xml_sanitize_id(id); xml_container = create_resource(id, "heartbeat", PE__CONTAINER_AGENT_RKT_S); free(id); xml_obj = create_xml_node(xml_container, XML_TAG_ATTR_SETS); crm_xml_set_id(xml_obj, "%s-attributes-%d", data->prefix, replica->offset); crm_create_nvpair_xml(xml_obj, NULL, "image", data->image); crm_create_nvpair_xml(xml_obj, NULL, "allow_pull", "true"); crm_create_nvpair_xml(xml_obj, NULL, "force_kill", "false"); crm_create_nvpair_xml(xml_obj, NULL, "reuse", "false"); /* Set a container hostname only if we have an IP to map it to. * The user can set -h or --uts=host themselves if they want a nicer * name for logs, but this makes applications happy who need their * hostname to match the IP they bind to. */ if (data->ip_range_start != NULL) { offset += snprintf(buffer+offset, max-offset, " --hostname=%s-%d", data->prefix, replica->offset); } offset += snprintf(buffer+offset, max-offset, " --environment=PCMK_stderr=1"); if (data->container_network) { #if 0 offset += snprintf(buffer+offset, max-offset, " --link-local-ip=%s", replica->ipaddr); #endif offset += snprintf(buffer+offset, max-offset, " --net=%s", data->container_network); } if(data->control_port) { offset += snprintf(buffer+offset, max-offset, " --environment=PCMK_remote_port=%s", data->control_port); } else { offset += snprintf(buffer+offset, max-offset, " --environment=PCMK_remote_port=%d", DEFAULT_REMOTE_PORT); } for(GListPtr pIter = data->mounts; pIter != NULL; pIter = pIter->next) { pe__bundle_mount_t *mount = pIter->data; - if(mount->flags) { + if (is_set(mount->flags, pe__bundle_mount_subdir)) { char *source = crm_strdup_printf( "%s/%s-%d", mount->source, data->prefix, replica->offset); if(doffset > 0) { doffset += snprintf(dbuffer+doffset, dmax-doffset, ","); } doffset += snprintf(dbuffer+doffset, dmax-doffset, "%s", source); offset += snprintf(buffer+offset, max-offset, " --volume vol%d,kind=host,source=%s", volid, source); if(mount->options) { offset += snprintf(buffer+offset, max-offset, ",%s", mount->options); } offset += snprintf(buffer+offset, max-offset, " --mount volume=vol%d,target=%s", volid, mount->target); free(source); } else { offset += snprintf(buffer+offset, max-offset, " --volume vol%d,kind=host,source=%s", volid, mount->source); if(mount->options) { offset += snprintf(buffer+offset, max-offset, ",%s", mount->options); } offset += snprintf(buffer+offset, max-offset, " --mount volume=vol%d,target=%s", volid, mount->target); } volid++; } for(GListPtr pIter = data->ports; pIter != NULL; pIter = pIter->next) { pe__bundle_port_t *port = pIter->data; if (replica->ipaddr) { offset += snprintf(buffer+offset, max-offset, " --port=%s:%s:%s", port->target, replica->ipaddr, port->source); } else { offset += snprintf(buffer+offset, max-offset, " --port=%s:%s", port->target, port->source); } } if (data->launcher_options) { offset += snprintf(buffer+offset, max-offset, " %s", data->launcher_options); } if (data->container_host_options) { offset += snprintf(buffer + offset, max - offset, " %s", data->container_host_options); } crm_create_nvpair_xml(xml_obj, NULL, "run_opts", buffer); free(buffer); crm_create_nvpair_xml(xml_obj, NULL, "mount_points", dbuffer); free(dbuffer); if (replica->child) { if (data->container_command) { crm_create_nvpair_xml(xml_obj, NULL, "run_cmd", data->container_command); } else { crm_create_nvpair_xml(xml_obj, NULL, "run_cmd", SBIN_DIR "/pacemaker-remoted"); } /* TODO: Allow users to specify their own? * * We just want to know if the container is alive, we'll * monitor the child independently */ crm_create_nvpair_xml(xml_obj, NULL, "monitor_cmd", "/bin/true"); /* } else if(child && data->untrusted) { * Support this use-case? * * The ability to have resources started/stopped by us, but * unable to set attributes, etc. * * Arguably better to control API access this with ACLs like * "normal" remote nodes * * crm_create_nvpair_xml(xml_obj, NULL, * "run_cmd", * "/usr/libexec/pacemaker/pacemaker-execd"); * crm_create_nvpair_xml(xml_obj, NULL, "monitor_cmd", * "/usr/libexec/pacemaker/lrmd_internal_ctl -c poke"); */ } else { if (data->container_command) { crm_create_nvpair_xml(xml_obj, NULL, "run_cmd", data->container_command); } /* TODO: Allow users to specify their own? * * We don't know what's in the container, so we just want * to know if it is alive */ crm_create_nvpair_xml(xml_obj, NULL, "monitor_cmd", "/bin/true"); } xml_obj = create_xml_node(xml_container, "operations"); crm_create_op_xml(xml_obj, ID(xml_container), "monitor", "60s", NULL); // TODO: Other ops? Timeouts and intervals from underlying resource? if (!common_unpack(xml_container, &replica->container, parent, data_set)) { return FALSE; } parent->children = g_list_append(parent->children, replica->container); return TRUE; } /*! * \brief Ban a node from a resource's (and its children's) allowed nodes list * * \param[in,out] rsc Resource to modify * \param[in] uname Name of node to ban */ static void disallow_node(resource_t *rsc, const char *uname) { gpointer match = g_hash_table_lookup(rsc->allowed_nodes, uname); if (match) { ((pe_node_t *) match)->weight = -INFINITY; ((pe_node_t *) match)->rsc_discover_mode = pe_discover_never; } if (rsc->children) { GListPtr child; for (child = rsc->children; child != NULL; child = child->next) { disallow_node((resource_t *) (child->data), uname); } } } static bool create_remote_resource(pe_resource_t *parent, pe__bundle_variant_data_t *data, pe__bundle_replica_t *replica, pe_working_set_t *data_set) { if (replica->child && valid_network(data)) { GHashTableIter gIter; GListPtr rsc_iter = NULL; node_t *node = NULL; xmlNode *xml_remote = NULL; char *id = crm_strdup_printf("%s-%d", data->prefix, replica->offset); char *port_s = NULL; const char *uname = NULL; const char *connect_name = NULL; if (remote_id_conflict(id, data_set)) { free(id); // The biggest hammer we have id = crm_strdup_printf("pcmk-internal-%s-remote-%d", replica->child->id, replica->offset); CRM_ASSERT(remote_id_conflict(id, data_set) == FALSE); } /* REMOTE_CONTAINER_HACK: Using "#uname" as the server name when the * connection does not have its own IP is a magic string that we use to * support nested remotes (i.e. a bundle running on a remote node). */ connect_name = (replica->ipaddr? replica->ipaddr : "#uname"); if (data->control_port == NULL) { port_s = crm_itoa(DEFAULT_REMOTE_PORT); } /* This sets replica->container as replica->remote's container, which is * similar to what happens with guest nodes. This is how the PE knows * that the bundle node is fenced by recovering the container, and that * remote should be ordered relative to the container. */ xml_remote = pe_create_remote_xml(NULL, id, replica->container->id, NULL, NULL, NULL, connect_name, (data->control_port? data->control_port : port_s)); free(port_s); /* Abandon our created ID, and pull the copy from the XML, because we * need something that will get freed during data set cleanup to use as * the node ID and uname. */ free(id); id = NULL; uname = ID(xml_remote); /* Ensure a node has been created for the guest (it may have already * been, if it has a permanent node attribute), and ensure its weight is * -INFINITY so no other resources can run on it. */ node = pe_find_node(data_set->nodes, uname); if (node == NULL) { node = pe_create_node(uname, uname, "remote", "-INFINITY", data_set); } else { node->weight = -INFINITY; } node->rsc_discover_mode = pe_discover_never; /* unpack_remote_nodes() ensures that each remote node and guest node * has a pe_node_t entry. Ideally, it would do the same for bundle nodes. * Unfortunately, a bundle has to be mostly unpacked before it's obvious * what nodes will be needed, so we do it just above. * * Worse, that means that the node may have been utilized while * unpacking other resources, without our weight correction. The most * likely place for this to happen is when common_unpack() calls * resource_location() to set a default score in symmetric clusters. * This adds a node *copy* to each resource's allowed nodes, and these * copies will have the wrong weight. * * As a hacky workaround, fix those copies here. * * @TODO Possible alternative: ensure bundles are unpacked before other * resources, so the weight is correct before any copies are made. */ for (rsc_iter = data_set->resources; rsc_iter; rsc_iter = rsc_iter->next) { disallow_node((resource_t *) (rsc_iter->data), uname); } replica->node = node_copy(node); replica->node->weight = 500; replica->node->rsc_discover_mode = pe_discover_exclusive; /* Ensure the node shows up as allowed and with the correct discovery set */ if (replica->child->allowed_nodes != NULL) { g_hash_table_destroy(replica->child->allowed_nodes); } replica->child->allowed_nodes = g_hash_table_new_full(crm_str_hash, g_str_equal, NULL, free); g_hash_table_insert(replica->child->allowed_nodes, (gpointer) replica->node->details->id, node_copy(replica->node)); { node_t *copy = node_copy(replica->node); copy->weight = -INFINITY; g_hash_table_insert(replica->child->parent->allowed_nodes, (gpointer) replica->node->details->id, copy); } if (!common_unpack(xml_remote, &replica->remote, parent, data_set)) { return FALSE; } g_hash_table_iter_init(&gIter, replica->remote->allowed_nodes); while (g_hash_table_iter_next(&gIter, NULL, (void **)&node)) { if (pe__is_guest_or_remote_node(node)) { /* Remote resources can only run on 'normal' cluster node */ node->weight = -INFINITY; } } replica->node->details->remote_rsc = replica->remote; // Ensure pe__is_guest_node() functions correctly immediately replica->remote->container = replica->container; /* A bundle's #kind is closer to "container" (guest node) than the * "remote" set by pe_create_node(). */ g_hash_table_insert(replica->node->details->attrs, strdup(CRM_ATTR_KIND), strdup("container")); /* One effect of this is that setup_container() will add * replica->remote to replica->container's fillers, which will make * pe__resource_contains_guest_node() true for replica->container. * * replica->child does NOT get added to replica->container's fillers. * The only noticeable effect if it did would be for its fail count to * be taken into account when checking replica->container's migration * threshold. */ parent->children = g_list_append(parent->children, replica->remote); } return TRUE; } static bool create_container(pe_resource_t *parent, pe__bundle_variant_data_t *data, pe__bundle_replica_t *replica, pe_working_set_t *data_set) { switch (data->agent_type) { case PE__CONTAINER_AGENT_DOCKER: if (!create_docker_resource(parent, data, replica, data_set)) { return FALSE; } break; case PE__CONTAINER_AGENT_PODMAN: if (!create_podman_resource(parent, data, replica, data_set)) { return FALSE; } break; case PE__CONTAINER_AGENT_RKT: if (!create_rkt_resource(parent, data, replica, data_set)) { return FALSE; } break; default: // PE__CONTAINER_AGENT_UNKNOWN return FALSE; } if (create_ip_resource(parent, data, replica, data_set) == FALSE) { return FALSE; } if(create_remote_resource(parent, data, replica, data_set) == FALSE) { return FALSE; } if (replica->child && replica->ipaddr) { add_hash_param(replica->child->meta, "external-ip", replica->ipaddr); } if (replica->remote) { /* * Allow the remote connection resource to be allocated to a * different node than the one on which the container is active. * * This makes it possible to have Pacemaker Remote nodes running * containers with pacemaker-remoted inside in order to start * services inside those containers. */ set_bit(replica->remote->flags, pe_rsc_allow_remote_remotes); } return TRUE; } static void mount_add(pe__bundle_variant_data_t *bundle_data, const char *source, - const char *target, const char *options, int flags) + const char *target, const char *options, uint32_t flags) { pe__bundle_mount_t *mount = calloc(1, sizeof(pe__bundle_mount_t)); mount->source = strdup(source); mount->target = strdup(target); if (options) { mount->options = strdup(options); } mount->flags = flags; bundle_data->mounts = g_list_append(bundle_data->mounts, mount); } static void mount_free(pe__bundle_mount_t *mount) { free(mount->source); free(mount->target); free(mount->options); free(mount); } static void port_free(pe__bundle_port_t *port) { free(port->source); free(port->target); free(port); } static pe__bundle_replica_t * replica_for_remote(pe_resource_t *remote) { resource_t *top = remote; pe__bundle_variant_data_t *bundle_data = NULL; if (top == NULL) { return NULL; } while (top->parent != NULL) { top = top->parent; } get_bundle_variant_data(bundle_data, top); for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pe__bundle_replica_t *replica = gIter->data; if (replica->remote == remote) { return replica; } } CRM_LOG_ASSERT(FALSE); return NULL; } bool pe__bundle_needs_remote_name(pe_resource_t *rsc) { const char *value; if (rsc == NULL) { return FALSE; } value = g_hash_table_lookup(rsc->parameters, XML_RSC_ATTR_REMOTE_RA_ADDR); if (safe_str_eq(value, "#uname") == FALSE) { return FALSE; } else { const char *match[3][2] = { { XML_ATTR_TYPE, "remote" }, { XML_AGENT_ATTR_CLASS, PCMK_RESOURCE_CLASS_OCF }, { XML_AGENT_ATTR_PROVIDER, "pacemaker" }, }; for (int m = 0; m < 3; m++) { value = crm_element_value(rsc->xml, match[m][0]); if (safe_str_neq(value, match[m][1])) { return FALSE; } } } return TRUE; } const char * pe__add_bundle_remote_name(pe_resource_t *rsc, xmlNode *xml, const char *field) { // REMOTE_CONTAINER_HACK: Allow remote nodes that start containers with pacemaker remote inside pe_node_t *node = NULL; pe__bundle_replica_t *replica = NULL; if (!pe__bundle_needs_remote_name(rsc)) { return NULL; } replica = replica_for_remote(rsc); if (replica == NULL) { return NULL; } node = replica->container->allocated_to; if (node == NULL) { /* If it won't be running anywhere after the * transition, go with where it's running now. */ node = pe__current_node(replica->container); } if(node == NULL) { crm_trace("Cannot determine address for bundle connection %s", rsc->id); return NULL; } crm_trace("Setting address for bundle connection %s to bundle host %s", rsc->id, node->details->uname); if(xml != NULL && field != NULL) { crm_xml_add(xml, field, node->details->uname); } return node->details->uname; } gboolean pe__unpack_bundle(pe_resource_t *rsc, pe_working_set_t *data_set) { const char *value = NULL; xmlNode *xml_obj = NULL; xmlNode *xml_resource = NULL; pe__bundle_variant_data_t *bundle_data = NULL; CRM_ASSERT(rsc != NULL); pe_rsc_trace(rsc, "Processing resource %s...", rsc->id); bundle_data = calloc(1, sizeof(pe__bundle_variant_data_t)); rsc->variant_opaque = bundle_data; bundle_data->prefix = strdup(rsc->id); xml_obj = first_named_child(rsc->xml, PE__CONTAINER_AGENT_DOCKER_S); if (xml_obj != NULL) { bundle_data->agent_type = PE__CONTAINER_AGENT_DOCKER; } else { xml_obj = first_named_child(rsc->xml, PE__CONTAINER_AGENT_RKT_S); if (xml_obj != NULL) { bundle_data->agent_type = PE__CONTAINER_AGENT_RKT; } else { xml_obj = first_named_child(rsc->xml, PE__CONTAINER_AGENT_PODMAN_S); if (xml_obj != NULL) { bundle_data->agent_type = PE__CONTAINER_AGENT_PODMAN; } else { return FALSE; } } } value = crm_element_value(xml_obj, XML_RSC_ATTR_PROMOTED_MAX); if (value == NULL) { // @COMPAT deprecated since 2.0.0 value = crm_element_value(xml_obj, "masters"); } bundle_data->promoted_max = crm_parse_int(value, "0"); if (bundle_data->promoted_max < 0) { pe_err("%s for %s must be nonnegative integer, using 0", XML_RSC_ATTR_PROMOTED_MAX, rsc->id); bundle_data->promoted_max = 0; } value = crm_element_value(xml_obj, "replicas"); if ((value == NULL) && bundle_data->promoted_max) { bundle_data->nreplicas = bundle_data->promoted_max; } else { bundle_data->nreplicas = crm_parse_int(value, "1"); } if (bundle_data->nreplicas < 1) { pe_err("'replicas' for %s must be positive integer, using 1", rsc->id); bundle_data->nreplicas = 1; } /* * Communication between containers on the same host via the * floating IPs only works if the container is started with: * --userland-proxy=false --ip-masq=false */ value = crm_element_value(xml_obj, "replicas-per-host"); bundle_data->nreplicas_per_host = crm_parse_int(value, "1"); if (bundle_data->nreplicas_per_host < 1) { pe_err("'replicas-per-host' for %s must be positive integer, using 1", rsc->id); bundle_data->nreplicas_per_host = 1; } if (bundle_data->nreplicas_per_host == 1) { clear_bit(rsc->flags, pe_rsc_unique); } bundle_data->container_command = crm_element_value_copy(xml_obj, "run-command"); bundle_data->launcher_options = crm_element_value_copy(xml_obj, "options"); bundle_data->image = crm_element_value_copy(xml_obj, "image"); bundle_data->container_network = crm_element_value_copy(xml_obj, "network"); xml_obj = first_named_child(rsc->xml, "network"); if(xml_obj) { bundle_data->ip_range_start = crm_element_value_copy(xml_obj, "ip-range-start"); bundle_data->host_netmask = crm_element_value_copy(xml_obj, "host-netmask"); bundle_data->host_network = crm_element_value_copy(xml_obj, "host-interface"); bundle_data->control_port = crm_element_value_copy(xml_obj, "control-port"); value = crm_element_value(xml_obj, "add-host"); if (check_boolean(value) == FALSE) { bundle_data->add_host = TRUE; } else { crm_str_to_boolean(value, &bundle_data->add_host); } for (xmlNode *xml_child = __xml_first_child_element(xml_obj); xml_child != NULL; xml_child = __xml_next_element(xml_child)) { pe__bundle_port_t *port = calloc(1, sizeof(pe__bundle_port_t)); port->source = crm_element_value_copy(xml_child, "port"); if(port->source == NULL) { port->source = crm_element_value_copy(xml_child, "range"); } else { port->target = crm_element_value_copy(xml_child, "internal-port"); } if(port->source != NULL && strlen(port->source) > 0) { if(port->target == NULL) { port->target = strdup(port->source); } bundle_data->ports = g_list_append(bundle_data->ports, port); } else { pe_err("Invalid port directive %s", ID(xml_child)); port_free(port); } } } xml_obj = first_named_child(rsc->xml, "storage"); for (xmlNode *xml_child = __xml_first_child_element(xml_obj); xml_child != NULL; xml_child = __xml_next_element(xml_child)) { const char *source = crm_element_value(xml_child, "source-dir"); const char *target = crm_element_value(xml_child, "target-dir"); const char *options = crm_element_value(xml_child, "options"); - int flags = 0; + int flags = pe__bundle_mount_none; if (source == NULL) { source = crm_element_value(xml_child, "source-dir-root"); - flags = 1; + set_bit(flags, pe__bundle_mount_subdir); } if (source && target) { mount_add(bundle_data, source, target, options, flags); } else { pe_err("Invalid mount directive %s", ID(xml_child)); } } xml_obj = first_named_child(rsc->xml, "primitive"); if (xml_obj && valid_network(bundle_data)) { char *value = NULL; xmlNode *xml_set = NULL; xml_resource = create_xml_node(NULL, XML_CIB_TAG_INCARNATION); /* @COMPAT We no longer use the tag, but we need to keep it as * part of the resource name, so that bundles don't restart in a rolling * upgrade. (It also avoids needing to change regression tests.) */ crm_xml_set_id(xml_resource, "%s-%s", bundle_data->prefix, (bundle_data->promoted_max? "master" : (const char *)xml_resource->name)); xml_set = create_xml_node(xml_resource, XML_TAG_META_SETS); crm_xml_set_id(xml_set, "%s-%s-meta", bundle_data->prefix, xml_resource->name); crm_create_nvpair_xml(xml_set, NULL, XML_RSC_ATTR_ORDERED, XML_BOOLEAN_TRUE); value = crm_itoa(bundle_data->nreplicas); crm_create_nvpair_xml(xml_set, NULL, XML_RSC_ATTR_INCARNATION_MAX, value); free(value); value = crm_itoa(bundle_data->nreplicas_per_host); crm_create_nvpair_xml(xml_set, NULL, XML_RSC_ATTR_INCARNATION_NODEMAX, value); free(value); crm_create_nvpair_xml(xml_set, NULL, XML_RSC_ATTR_UNIQUE, (bundle_data->nreplicas_per_host > 1)? XML_BOOLEAN_TRUE : XML_BOOLEAN_FALSE); if (bundle_data->promoted_max) { crm_create_nvpair_xml(xml_set, NULL, XML_RSC_ATTR_PROMOTABLE, XML_BOOLEAN_TRUE); value = crm_itoa(bundle_data->promoted_max); crm_create_nvpair_xml(xml_set, NULL, XML_RSC_ATTR_PROMOTED_MAX, value); free(value); } //crm_xml_add(xml_obj, XML_ATTR_ID, bundle_data->prefix); add_node_copy(xml_resource, xml_obj); } else if(xml_obj) { pe_err("Cannot control %s inside %s without either ip-range-start or control-port", rsc->id, ID(xml_obj)); return FALSE; } if(xml_resource) { int lpc = 0; GListPtr childIter = NULL; resource_t *new_rsc = NULL; pe__bundle_port_t *port = NULL; int offset = 0, max = 1024; char *buffer = NULL; if (common_unpack(xml_resource, &new_rsc, rsc, data_set) == FALSE) { pe_err("Failed unpacking resource %s", ID(rsc->xml)); if (new_rsc != NULL && new_rsc->fns != NULL) { new_rsc->fns->free(new_rsc); } return FALSE; } bundle_data->child = new_rsc; /* Currently, we always map the default authentication key location * into the same location inside the container. * * Ideally, we would respect the host's PCMK_authkey_location, but: * - it may be different on different nodes; * - the actual connection will do extra checking to make sure the key * file exists and is readable, that we can't do here on the DC * - tools such as crm_resource and crm_simulate may not have the same * environment variables as the cluster, causing operation digests to * differ * * Always using the default location inside the container is fine, * because we control the pacemaker_remote environment, and it avoids * having to pass another environment variable to the container. * * @TODO A better solution may be to have only pacemaker_remote use the * environment variable, and have the cluster nodes use a new * cluster option for key location. This would introduce the limitation * of the location being the same on all cluster nodes, but that's * reasonable. */ mount_add(bundle_data, DEFAULT_REMOTE_KEY_LOCATION, - DEFAULT_REMOTE_KEY_LOCATION, NULL, 0); + DEFAULT_REMOTE_KEY_LOCATION, NULL, pe__bundle_mount_none); - mount_add(bundle_data, CRM_BUNDLE_DIR, "/var/log", NULL, 1); + mount_add(bundle_data, CRM_BUNDLE_DIR, "/var/log", NULL, + pe__bundle_mount_subdir); port = calloc(1, sizeof(pe__bundle_port_t)); if(bundle_data->control_port) { port->source = strdup(bundle_data->control_port); } else { /* If we wanted to respect PCMK_remote_port, we could use * crm_default_remote_port() here and elsewhere in this file instead * of DEFAULT_REMOTE_PORT. * * However, it gains nothing, since we control both the container * environment and the connection resource parameters, and the user * can use a different port if desired by setting control-port. */ port->source = crm_itoa(DEFAULT_REMOTE_PORT); } port->target = strdup(port->source); bundle_data->ports = g_list_append(bundle_data->ports, port); buffer = calloc(1, max+1); for (childIter = bundle_data->child->children; childIter != NULL; childIter = childIter->next) { pe__bundle_replica_t *replica = calloc(1, sizeof(pe__bundle_replica_t)); replica->child = childIter->data; replica->child->exclusive_discover = TRUE; replica->offset = lpc++; // Ensure the child's notify gets set based on the underlying primitive's value if (is_set(replica->child->flags, pe_rsc_notify)) { set_bit(bundle_data->child->flags, pe_rsc_notify); } offset += allocate_ip(bundle_data, replica, buffer+offset, max-offset); bundle_data->replicas = g_list_append(bundle_data->replicas, replica); bundle_data->attribute_target = g_hash_table_lookup(replica->child->meta, XML_RSC_ATTR_TARGET); } bundle_data->container_host_options = buffer; if (bundle_data->attribute_target) { g_hash_table_replace(rsc->meta, strdup(XML_RSC_ATTR_TARGET), strdup(bundle_data->attribute_target)); g_hash_table_replace(bundle_data->child->meta, strdup(XML_RSC_ATTR_TARGET), strdup(bundle_data->attribute_target)); } } else { // Just a naked container, no pacemaker-remote int offset = 0, max = 1024; char *buffer = calloc(1, max+1); for (int lpc = 0; lpc < bundle_data->nreplicas; lpc++) { pe__bundle_replica_t *replica = calloc(1, sizeof(pe__bundle_replica_t)); replica->offset = lpc; offset += allocate_ip(bundle_data, replica, buffer+offset, max-offset); bundle_data->replicas = g_list_append(bundle_data->replicas, replica); } bundle_data->container_host_options = buffer; } for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pe__bundle_replica_t *replica = gIter->data; if (!create_container(rsc, bundle_data, replica, data_set)) { pe_err("Failed unpacking resource %s", rsc->id); rsc->fns->free(rsc); return FALSE; } } if (bundle_data->child) { rsc->children = g_list_append(rsc->children, bundle_data->child); } return TRUE; } static int replica_resource_active(pe_resource_t *rsc, gboolean all) { if (rsc) { gboolean child_active = rsc->fns->active(rsc, all); if (child_active && !all) { return TRUE; } else if (!child_active && all) { return FALSE; } } return -1; } gboolean pe__bundle_active(pe_resource_t *rsc, gboolean all) { pe__bundle_variant_data_t *bundle_data = NULL; GListPtr iter = NULL; get_bundle_variant_data(bundle_data, rsc); for (iter = bundle_data->replicas; iter != NULL; iter = iter->next) { pe__bundle_replica_t *replica = iter->data; int rsc_active; rsc_active = replica_resource_active(replica->ip, all); if (rsc_active >= 0) { return (gboolean) rsc_active; } rsc_active = replica_resource_active(replica->child, all); if (rsc_active >= 0) { return (gboolean) rsc_active; } rsc_active = replica_resource_active(replica->container, all); if (rsc_active >= 0) { return (gboolean) rsc_active; } rsc_active = replica_resource_active(replica->remote, all); if (rsc_active >= 0) { return (gboolean) rsc_active; } } /* If "all" is TRUE, we've already checked that no resources were inactive, * so return TRUE; if "all" is FALSE, we didn't find any active resources, * so return FALSE. */ return all; } /*! * \internal * \brief Find the bundle replica corresponding to a given node * * \param[in] bundle Top-level bundle resource * \param[in] node Node to search for * * \return Bundle replica if found, NULL otherwise */ pe_resource_t * pe__find_bundle_replica(const pe_resource_t *bundle, const pe_node_t *node) { pe__bundle_variant_data_t *bundle_data = NULL; CRM_ASSERT(bundle && node); get_bundle_variant_data(bundle_data, bundle); for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pe__bundle_replica_t *replica = gIter->data; CRM_ASSERT(replica && replica->node); if (replica->node->details == node->details) { return replica->child; } } return NULL; } static void print_rsc_in_list(resource_t *rsc, const char *pre_text, long options, void *print_data) { if (rsc != NULL) { if (options & pe_print_html) { status_print("
  • "); } rsc->fns->print(rsc, pre_text, options, print_data); if (options & pe_print_html) { status_print("
    • \n"); } } } static const char* container_agent_str(enum pe__container_agent t) { switch (t) { case PE__CONTAINER_AGENT_DOCKER: return PE__CONTAINER_AGENT_DOCKER_S; case PE__CONTAINER_AGENT_RKT: return PE__CONTAINER_AGENT_RKT_S; case PE__CONTAINER_AGENT_PODMAN: return PE__CONTAINER_AGENT_PODMAN_S; default: // PE__CONTAINER_AGENT_UNKNOWN break; } return PE__CONTAINER_AGENT_UNKNOWN_S; } static void bundle_print_xml(pe_resource_t *rsc, const char *pre_text, long options, void *print_data) { pe__bundle_variant_data_t *bundle_data = NULL; char *child_text = NULL; CRM_CHECK(rsc != NULL, return); if (pre_text == NULL) { pre_text = ""; } child_text = crm_concat(pre_text, " ", ' '); get_bundle_variant_data(bundle_data, rsc); status_print("%sid); status_print("type=\"%s\" ", container_agent_str(bundle_data->agent_type)); status_print("image=\"%s\" ", bundle_data->image); status_print("unique=\"%s\" ", is_set(rsc->flags, pe_rsc_unique)? "true" : "false"); status_print("managed=\"%s\" ", is_set(rsc->flags, pe_rsc_managed) ? "true" : "false"); status_print("failed=\"%s\" ", is_set(rsc->flags, pe_rsc_failed) ? "true" : "false"); status_print(">\n"); for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pe__bundle_replica_t *replica = gIter->data; CRM_ASSERT(replica); status_print("%s \n", pre_text, replica->offset); print_rsc_in_list(replica->ip, child_text, options, print_data); print_rsc_in_list(replica->child, child_text, options, print_data); print_rsc_in_list(replica->container, child_text, options, print_data); print_rsc_in_list(replica->remote, child_text, options, print_data); status_print("%s \n", pre_text); } status_print("%s\n", pre_text); free(child_text); } static void print_bundle_replica(pe__bundle_replica_t *replica, const char *pre_text, long options, void *print_data) { node_t *node = NULL; pe_resource_t *rsc = replica->child; int offset = 0; char buffer[LINE_MAX]; if(rsc == NULL) { rsc = replica->container; } if (replica->remote) { offset += snprintf(buffer + offset, LINE_MAX - offset, "%s", rsc_printable_id(replica->remote)); } else { offset += snprintf(buffer + offset, LINE_MAX - offset, "%s", rsc_printable_id(replica->container)); } if (replica->ipaddr) { offset += snprintf(buffer + offset, LINE_MAX - offset, " (%s)", replica->ipaddr); } node = pe__current_node(replica->container); common_print(rsc, pre_text, buffer, node, options, print_data); } void pe__print_bundle(pe_resource_t *rsc, const char *pre_text, long options, void *print_data) { pe__bundle_variant_data_t *bundle_data = NULL; char *child_text = NULL; CRM_CHECK(rsc != NULL, return); if (options & pe_print_xml) { bundle_print_xml(rsc, pre_text, options, print_data); return; } get_bundle_variant_data(bundle_data, rsc); if (pre_text == NULL) { pre_text = " "; } status_print("%sContainer bundle%s: %s [%s]%s%s\n", pre_text, ((bundle_data->nreplicas > 1)? " set" : ""), rsc->id, bundle_data->image, is_set(rsc->flags, pe_rsc_unique) ? " (unique)" : "", is_set(rsc->flags, pe_rsc_managed) ? "" : " (unmanaged)"); if (options & pe_print_html) { status_print("
    \n
      \n"); } for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pe__bundle_replica_t *replica = gIter->data; CRM_ASSERT(replica); if (options & pe_print_html) { status_print("
    • "); } if (is_set(options, pe_print_implicit)) { child_text = crm_strdup_printf(" %s", pre_text); if(g_list_length(bundle_data->replicas) > 1) { status_print(" %sReplica[%d]\n", pre_text, replica->offset); } if (options & pe_print_html) { status_print("
      \n
        \n"); } print_rsc_in_list(replica->ip, child_text, options, print_data); print_rsc_in_list(replica->container, child_text, options, print_data); print_rsc_in_list(replica->remote, child_text, options, print_data); print_rsc_in_list(replica->child, child_text, options, print_data); if (options & pe_print_html) { status_print("
      \n"); } } else { child_text = crm_strdup_printf("%s ", pre_text); print_bundle_replica(replica, child_text, options, print_data); } free(child_text); if (options & pe_print_html) { status_print("
      • \n"); } } if (options & pe_print_html) { status_print("
    \n"); } } static void free_bundle_replica(pe__bundle_replica_t *replica) { if (replica == NULL) { return; } if (replica->node) { free(replica->node); replica->node = NULL; } if (replica->ip) { free_xml(replica->ip->xml); replica->ip->xml = NULL; replica->ip->fns->free(replica->ip); replica->ip = NULL; } if (replica->container) { free_xml(replica->container->xml); replica->container->xml = NULL; replica->container->fns->free(replica->container); replica->container = NULL; } if (replica->remote) { free_xml(replica->remote->xml); replica->remote->xml = NULL; replica->remote->fns->free(replica->remote); replica->remote = NULL; } free(replica->ipaddr); free(replica); } void pe__free_bundle(pe_resource_t *rsc) { pe__bundle_variant_data_t *bundle_data = NULL; CRM_CHECK(rsc != NULL, return); get_bundle_variant_data(bundle_data, rsc); pe_rsc_trace(rsc, "Freeing %s", rsc->id); free(bundle_data->prefix); free(bundle_data->image); free(bundle_data->control_port); free(bundle_data->host_network); free(bundle_data->host_netmask); free(bundle_data->ip_range_start); free(bundle_data->container_network); free(bundle_data->launcher_options); free(bundle_data->container_command); free(bundle_data->container_host_options); g_list_free_full(bundle_data->replicas, (GDestroyNotify) free_bundle_replica); g_list_free_full(bundle_data->mounts, (GDestroyNotify)mount_free); g_list_free_full(bundle_data->ports, (GDestroyNotify)port_free); g_list_free(rsc->children); if(bundle_data->child) { free_xml(bundle_data->child->xml); bundle_data->child->xml = NULL; bundle_data->child->fns->free(bundle_data->child); } common_free(rsc); } enum rsc_role_e pe__bundle_resource_state(const pe_resource_t *rsc, gboolean current) { enum rsc_role_e container_role = RSC_ROLE_UNKNOWN; return container_role; } /*! * \brief Get the number of configured replicas in a bundle * * \param[in] rsc Bundle resource * * \return Number of configured replicas, or 0 on error */ int pe_bundle_replicas(const resource_t *rsc) { if ((rsc == NULL) || (rsc->variant != pe_container)) { return 0; } else { pe__bundle_variant_data_t *bundle_data = NULL; get_bundle_variant_data(bundle_data, rsc); return bundle_data->nreplicas; } } diff --git a/lib/pengine/variant.h b/lib/pengine/variant.h index f46aa111c1..7f77eef87a 100644 --- a/lib/pengine/variant.h +++ b/lib/pengine/variant.h @@ -1,146 +1,153 @@ /* * Copyright 2004-2019 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 PE_VARIANT__H # define PE_VARIANT__H # if VARIANT_CLONE typedef struct clone_variant_data_s { int clone_max; int clone_node_max; int promoted_max; int promoted_node_max; int total_clones; // @TODO make these a bitmask gboolean ordered; gboolean applied_master_prefs; gboolean merged_master_weights; notify_data_t *stop_notify; notify_data_t *start_notify; notify_data_t *demote_notify; notify_data_t *promote_notify; xmlNode *xml_obj_child; } clone_variant_data_t; # define get_clone_variant_data(data, rsc) \ CRM_ASSERT(rsc != NULL); \ CRM_ASSERT(rsc->variant == pe_clone); \ data = (clone_variant_data_t *)rsc->variant_opaque; # elif PE__VARIANT_BUNDLE typedef struct { int offset; char *ipaddr; pe_node_t *node; pe_resource_t *ip; pe_resource_t *child; pe_resource_t *container; pe_resource_t *remote; } pe__bundle_replica_t; +enum pe__bundle_mount_flags { + pe__bundle_mount_none = 0x00, + + // mount instance-specific subdirectory rather than source directly + pe__bundle_mount_subdir = 0x01 +}; + typedef struct { char *source; char *target; char *options; - int flags; + uint32_t flags; // bitmask of pe__bundle_mount_flags } pe__bundle_mount_t; typedef struct { char *source; char *target; } pe__bundle_port_t; enum pe__container_agent { PE__CONTAINER_AGENT_UNKNOWN, PE__CONTAINER_AGENT_DOCKER, PE__CONTAINER_AGENT_RKT, PE__CONTAINER_AGENT_PODMAN, }; #define PE__CONTAINER_AGENT_UNKNOWN_S "unknown" #define PE__CONTAINER_AGENT_DOCKER_S "docker" #define PE__CONTAINER_AGENT_RKT_S "rkt" #define PE__CONTAINER_AGENT_PODMAN_S "podman" typedef struct pe__bundle_variant_data_s { int promoted_max; int nreplicas; int nreplicas_per_host; char *prefix; char *image; const char *ip_last; char *host_network; char *host_netmask; char *control_port; char *container_network; char *ip_range_start; gboolean add_host; char *container_host_options; char *container_command; char *launcher_options; const char *attribute_target; resource_t *child; GList *replicas; // pe__bundle_replica_t * GList *ports; // pe__bundle_port_t * GList *mounts; // pe__bundle_mount_t * enum pe__container_agent agent_type; } pe__bundle_variant_data_t; # define get_bundle_variant_data(data, rsc) \ CRM_ASSERT(rsc != NULL); \ CRM_ASSERT(rsc->variant == pe_container); \ CRM_ASSERT(rsc->variant_opaque != NULL); \ data = (pe__bundle_variant_data_t *)rsc->variant_opaque; \ # elif VARIANT_GROUP typedef struct group_variant_data_s { int num_children; resource_t *first_child; resource_t *last_child; gboolean colocated; gboolean ordered; gboolean child_starting; gboolean child_stopping; } group_variant_data_t; # define get_group_variant_data(data, rsc) \ CRM_ASSERT(rsc != NULL); \ CRM_ASSERT(rsc->variant == pe_group); \ CRM_ASSERT(rsc->variant_opaque != NULL); \ data = (group_variant_data_t *)rsc->variant_opaque; \ # elif VARIANT_NATIVE typedef struct native_variant_data_s { int dummy; } native_variant_data_t; # define get_native_variant_data(data, rsc) \ CRM_ASSERT(rsc != NULL); \ CRM_ASSERT(rsc->variant == pe_native); \ CRM_ASSERT(rsc->variant_opaque != NULL); \ data = (native_variant_data_t *)rsc->variant_opaque; # endif #endif