diff --git a/crmd/te_utils.c b/crmd/te_utils.c index dcfbb3b903..9fd4c9f0b4 100644 --- a/crmd/te_utils.c +++ b/crmd/te_utils.c @@ -1,675 +1,677 @@ /* * Copyright (C) 2004 Andrew Beekhof * * This program is free software; you can redistribute it and/or * modify it under the terms of the GNU General Public * License as published by the Free Software Foundation; either * version 2 of the License, or (at your option) any later version. * * This software is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * General Public License for more details. * * You should have received a copy of the GNU General Public * License along with this library; if not, write to the Free Software * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA */ #include #include #include #include #include #include #include #include #include #include #include crm_trigger_t *stonith_reconnect = NULL; /* * stonith cleanup list * * If the DC is shot, proper notifications might not go out. * The stonith cleanup list allows the cluster to (re-)send * notifications once a new DC is elected. */ static GListPtr stonith_cleanup_list = NULL; /*! * \internal * \brief Add a node to the stonith cleanup list * * \param[in] target Name of node to add */ void add_stonith_cleanup(const char *target) { stonith_cleanup_list = g_list_append(stonith_cleanup_list, strdup(target)); } /*! * \internal * \brief Remove a node from the stonith cleanup list * * \param[in] Name of node to remove */ void remove_stonith_cleanup(const char *target) { GListPtr iter = stonith_cleanup_list; while (iter != NULL) { GListPtr tmp = iter; char *iter_name = tmp->data; iter = iter->next; if (safe_str_eq(target, iter_name)) { crm_trace("Removing %s from the cleanup list", iter_name); stonith_cleanup_list = g_list_delete_link(stonith_cleanup_list, tmp); free(iter_name); } } } /*! * \internal * \brief Purge all entries from the stonith cleanup list */ void purge_stonith_cleanup() { if (stonith_cleanup_list) { GListPtr iter = NULL; for (iter = stonith_cleanup_list; iter != NULL; iter = iter->next) { char *target = iter->data; crm_info("Purging %s from stonith cleanup list", target); free(target); } g_list_free(stonith_cleanup_list); stonith_cleanup_list = NULL; } } /*! * \internal * \brief Send stonith updates for all entries in cleanup list, then purge it */ void execute_stonith_cleanup() { GListPtr iter; for (iter = stonith_cleanup_list; iter != NULL; iter = iter->next) { char *target = iter->data; crm_node_t *target_node = crm_get_peer(0, target); const char *uuid = crm_peer_uuid(target_node); crm_notice("Marking %s, target of a previous stonith action, as clean", target); send_stonith_update(NULL, target, uuid); free(target); } g_list_free(stonith_cleanup_list); stonith_cleanup_list = NULL; } /* end stonith cleanup list functions */ static gboolean fail_incompletable_stonith(crm_graph_t * graph) { GListPtr lpc = NULL; const char *task = NULL; xmlNode *last_action = NULL; if (graph == NULL) { return FALSE; } for (lpc = graph->synapses; lpc != NULL; lpc = lpc->next) { GListPtr lpc2 = NULL; synapse_t *synapse = (synapse_t *) lpc->data; if (synapse->confirmed) { continue; } for (lpc2 = synapse->actions; lpc2 != NULL; lpc2 = lpc2->next) { crm_action_t *action = (crm_action_t *) lpc2->data; if (action->type != action_type_crm || action->confirmed) { continue; } task = crm_element_value(action->xml, XML_LRM_ATTR_TASK); if (task && safe_str_eq(task, CRM_OP_FENCE)) { action->failed = TRUE; last_action = action->xml; update_graph(graph, action); crm_notice("Failing action %d (%s): STONITHd terminated", action->id, ID(action->xml)); } } } if (last_action != NULL) { crm_warn("STONITHd failure resulted in un-runnable actions"); abort_for_stonith_failure(tg_restart, NULL, last_action); return TRUE; } return FALSE; } static void tengine_stonith_connection_destroy(stonith_t * st, stonith_event_t * e) { if (is_set(fsa_input_register, R_ST_REQUIRED)) { crm_crit("Fencing daemon connection failed"); mainloop_set_trigger(stonith_reconnect); } else { crm_info("Fencing daemon disconnected"); } /* cbchan will be garbage at this point, arrange for it to be reset */ if(stonith_api) { stonith_api->state = stonith_disconnected; } if (AM_I_DC) { fail_incompletable_stonith(transition_graph); trigger_graph(); } } #if SUPPORT_CMAN # include #endif char *te_client_id = NULL; #ifdef HAVE_SYS_REBOOT_H # include # include #endif static void tengine_stonith_notify(stonith_t * st, stonith_event_t * st_event) { if(te_client_id == NULL) { te_client_id = crm_strdup_printf("%s.%d", crm_system_name, getpid()); } if (st_event == NULL) { crm_err("Notify data not found"); return; } crmd_alert_fencing_op(st_event); if (st_event->result == pcmk_ok && safe_str_eq("on", st_event->action)) { crm_notice("%s was successfully unfenced by %s (at the request of %s)", st_event->target, st_event->executioner ? st_event->executioner : "", st_event->origin); /* TODO: Hook up st_event->device */ return; } else if (safe_str_eq("on", st_event->action)) { crm_err("Unfencing of %s by %s failed: %s (%d)", st_event->target, st_event->executioner ? st_event->executioner : "", pcmk_strerror(st_event->result), st_event->result); return; } else if (st_event->result == pcmk_ok && crm_str_eq(st_event->target, fsa_our_uname, TRUE)) { crm_crit("We were allegedly just fenced by %s for %s!", st_event->executioner ? st_event->executioner : "", st_event->origin); /* Dumps blackbox if enabled */ qb_log_fini(); /* Try to get the above log message to disk - somehow */ /* Get out ASAP and do not come back up. * * Triggering a reboot is also not the worst idea either since * the rest of the cluster thinks we're safely down */ #ifdef RB_HALT_SYSTEM reboot(RB_HALT_SYSTEM); #endif /* * If reboot() fails or is not supported, coming back up will * probably lead to a situation where the other nodes set our * status to 'lost' because of the fencing callback and will * discard subsequent election votes with: * * Election 87 (current: 5171, owner: 103): Processed vote from east-03 (Peer is not part of our cluster) * * So just stay dead, something is seriously messed up anyway. * */ exit(100); /* None of our wrappers since we already called qb_log_fini() */ return; } if (safe_str_eq(st_event->operation, T_STONITH_NOTIFY_FENCE)) { if (st_event->result == pcmk_ok) { st_fail_count_reset(st_event->target); } else { st_fail_count_increment(st_event->target); } } - crm_notice("Peer %s was%s terminated (%s) by %s for %s: %s (ref=%s) by client %s", + crm_notice("Peer %s was%s terminated (%s) by %s on behalf of %s: %s " + CRM_XS " initiator=%s ref=%s", st_event->target, st_event->result == pcmk_ok ? "" : " not", st_event->action, st_event->executioner ? st_event->executioner : "", - st_event->origin, pcmk_strerror(st_event->result), st_event->id, - st_event->client_origin ? st_event->client_origin : ""); + (st_event->client_origin? st_event->client_origin : ""), + pcmk_strerror(st_event->result), + st_event->origin, st_event->id); #if SUPPORT_CMAN if (st_event->result == pcmk_ok && is_cman_cluster()) { int local_rc = 0; int confirm = 0; char *target_copy = strdup(st_event->target); /* In case fenced hasn't noticed yet * * Any fencing that has been inititated will be completed by way of the fence_pcmk redirect */ local_rc = fenced_external(target_copy); if (local_rc != 0) { crm_err("Could not notify CMAN that '%s' is now fenced: %d", st_event->target, local_rc); } else { crm_notice("Notified CMAN that '%s' is now fenced", st_event->target); } /* In case fenced is already trying to shoot it */ confirm = open("/var/run/cluster/fenced_override", O_NONBLOCK|O_WRONLY); if (confirm >= 0) { int ignore = 0; int len = strlen(target_copy); errno = 0; local_rc = write(confirm, target_copy, len); ignore = write(confirm, "\n", 1); if(ignore < 0 && errno == EBADF) { crm_trace("CMAN not expecting %s to be fenced (yet)", st_event->target); } else if (local_rc < len) { crm_perror(LOG_ERR, "Confirmation of CMAN fencing event for '%s' failed: %d", st_event->target, local_rc); } else { fsync(confirm); crm_notice("Confirmed CMAN fencing event for '%s'", st_event->target); } close(confirm); } free(target_copy); } #endif if (st_event->result == pcmk_ok) { crm_node_t *peer = crm_find_peer_full(0, st_event->target, CRM_GET_PEER_ANY); const char *uuid = NULL; gboolean we_are_executioner = safe_str_eq(st_event->executioner, fsa_our_uname); if (peer == NULL) { return; } uuid = crm_peer_uuid(peer); crm_trace("target=%s dc=%s", st_event->target, fsa_our_dc); if(AM_I_DC) { /* The DC always sends updates */ send_stonith_update(NULL, st_event->target, uuid); /* @TODO Ideally, at this point, we'd check whether the fenced node * hosted any guest nodes, and call remote_node_down() for them. * Unfortunately, the crmd doesn't have a simple, reliable way to * map hosts to guests. It might be possible to track this in the * peer cache via crm_remote_peer_cache_refresh(). For now, we rely * on the PE creating fence pseudo-events for the guests. */ if (st_event->client_origin && safe_str_neq(st_event->client_origin, te_client_id)) { /* Abort the current transition graph if it wasn't us * that invoked stonith to fence someone */ crm_info("External fencing operation from %s fenced %s", st_event->client_origin, st_event->target); abort_transition(INFINITY, tg_restart, "External Fencing Operation", NULL); } /* Assume it was our leader if we don't currently have one */ } else if (((fsa_our_dc == NULL) || safe_str_eq(fsa_our_dc, st_event->target)) && !is_set(peer->flags, crm_remote_node)) { crm_notice("Target %s our leader %s (recorded: %s)", fsa_our_dc ? "was" : "may have been", st_event->target, fsa_our_dc ? fsa_our_dc : ""); /* Given the CIB resyncing that occurs around elections, * have one node update the CIB now and, if the new DC is different, * have them do so too after the election */ if (we_are_executioner) { send_stonith_update(NULL, st_event->target, uuid); } add_stonith_cleanup(st_event->target); } /* If the target is a remote node, and we host its connection, * immediately fail all monitors so it can be recovered quickly. * The connection won't necessarily drop when a remote node is fenced, * so the failure might not otherwise be detected until the next poke. */ if (is_set(peer->flags, crm_remote_node)) { remote_ra_fail(st_event->target); } crmd_peer_down(peer, TRUE); } } gboolean te_connect_stonith(gpointer user_data) { int lpc = 0; int rc = pcmk_ok; if (stonith_api == NULL) { stonith_api = stonith_api_new(); } if (stonith_api->state != stonith_disconnected) { crm_trace("Still connected"); return TRUE; } for (lpc = 0; lpc < 30; lpc++) { crm_debug("Attempting connection to fencing daemon..."); sleep(1); rc = stonith_api->cmds->connect(stonith_api, crm_system_name, NULL); if (rc == pcmk_ok) { break; } if (user_data != NULL) { if (is_set(fsa_input_register, R_ST_REQUIRED)) { crm_err("Sign-in failed: triggered a retry"); mainloop_set_trigger(stonith_reconnect); } else { crm_info("Sign-in failed, but no longer required"); } return TRUE; } crm_err("Sign-in failed: pausing and trying again in 2s..."); sleep(1); } CRM_CHECK(rc == pcmk_ok, return TRUE); /* If not, we failed 30 times... just get out */ stonith_api->cmds->register_notification(stonith_api, T_STONITH_NOTIFY_DISCONNECT, tengine_stonith_connection_destroy); stonith_api->cmds->register_notification(stonith_api, T_STONITH_NOTIFY_FENCE, tengine_stonith_notify); crm_trace("Connected"); return TRUE; } gboolean stop_te_timer(crm_action_timer_t * timer) { const char *timer_desc = "action timer"; if (timer == NULL) { return FALSE; } if (timer->reason == timeout_abort) { timer_desc = "global timer"; crm_trace("Stopping %s", timer_desc); } if (timer->source_id != 0) { crm_trace("Stopping %s", timer_desc); g_source_remove(timer->source_id); timer->source_id = 0; } else { crm_trace("%s was already stopped", timer_desc); return FALSE; } return TRUE; } gboolean te_graph_trigger(gpointer user_data) { enum transition_status graph_rc = -1; if (transition_graph == NULL) { crm_debug("Nothing to do"); return TRUE; } crm_trace("Invoking graph %d in state %s", transition_graph->id, fsa_state2string(fsa_state)); switch (fsa_state) { case S_STARTING: case S_PENDING: case S_NOT_DC: case S_HALT: case S_ILLEGAL: case S_STOPPING: case S_TERMINATE: return TRUE; break; default: break; } if (transition_graph->complete == FALSE) { int limit = transition_graph->batch_limit; transition_graph->batch_limit = throttle_get_total_job_limit(limit); graph_rc = run_graph(transition_graph); transition_graph->batch_limit = limit; /* Restore the configured value */ /* significant overhead... */ /* print_graph(LOG_DEBUG_3, transition_graph); */ if (graph_rc == transition_active) { crm_trace("Transition not yet complete"); return TRUE; } else if (graph_rc == transition_pending) { crm_trace("Transition not yet complete - no actions fired"); return TRUE; } if (graph_rc != transition_complete) { crm_warn("Transition failed: %s", transition_status(graph_rc)); print_graph(LOG_NOTICE, transition_graph); } } crm_debug("Transition %d is now complete", transition_graph->id); transition_graph->complete = TRUE; notify_crmd(transition_graph); return TRUE; } void trigger_graph_processing(const char *fn, int line) { crm_trace("%s:%d - Triggered graph processing", fn, line); mainloop_set_trigger(transition_trigger); } void abort_transition_graph(int abort_priority, enum transition_action abort_action, const char *abort_text, xmlNode * reason, const char *fn, int line) { int add[] = { 0, 0, 0 }; int del[] = { 0, 0, 0 }; int level = LOG_INFO; xmlNode *diff = NULL; xmlNode *change = NULL; CRM_CHECK(transition_graph != NULL, return); switch (fsa_state) { case S_STARTING: case S_PENDING: case S_NOT_DC: case S_HALT: case S_ILLEGAL: case S_STOPPING: case S_TERMINATE: crm_info("Abort %s suppressed: state=%s (complete=%d)", abort_text, fsa_state2string(fsa_state), transition_graph->complete); return; default: break; } /* Make sure any queued calculations are discarded ASAP */ free(fsa_pe_ref); fsa_pe_ref = NULL; if (transition_graph->complete == FALSE) { if(update_abort_priority(transition_graph, abort_priority, abort_action, abort_text)) { level = LOG_NOTICE; } } if(reason) { xmlNode *search = NULL; for(search = reason; search; search = search->parent) { if (safe_str_eq(XML_TAG_DIFF, TYPE(search))) { diff = search; break; } } if(diff) { xml_patch_versions(diff, add, del); for(search = reason; search; search = search->parent) { if (safe_str_eq(XML_DIFF_CHANGE, TYPE(search))) { change = search; break; } } } } if(reason == NULL) { do_crm_log(level, "Transition aborted: %s "CRM_XS" source=%s:%d complete=%s", abort_text, fn, line, (transition_graph->complete? "true" : "false")); } else if(change == NULL) { char *local_path = xml_get_path(reason); do_crm_log(level, "Transition aborted by %s.%s: %s " CRM_XS " cib=%d.%d.%d source=%s:%d path=%s complete=%s", TYPE(reason), ID(reason), abort_text, add[0], add[1], add[2], fn, line, local_path, (transition_graph->complete? "true" : "false")); free(local_path); } else { const char *kind = NULL; const char *op = crm_element_value(change, XML_DIFF_OP); const char *path = crm_element_value(change, XML_DIFF_PATH); if(change == reason) { if(strcmp(op, "create") == 0) { reason = reason->children; } else if(strcmp(op, "modify") == 0) { reason = first_named_child(reason, XML_DIFF_RESULT); if(reason) { reason = reason->children; } } } kind = TYPE(reason); if(strcmp(op, "delete") == 0) { const char *shortpath = strrchr(path, '/'); do_crm_log(level, "Transition aborted by deletion of %s: %s " CRM_XS " cib=%d.%d.%d source=%s:%d path=%s complete=%s", (shortpath? (shortpath + 1) : path), abort_text, add[0], add[1], add[2], fn, line, path, (transition_graph->complete? "true" : "false")); } else if (safe_str_eq(XML_CIB_TAG_NVPAIR, kind)) { do_crm_log(level, "Transition aborted by %s doing %s %s=%s: %s " CRM_XS " cib=%d.%d.%d source=%s:%d path=%s complete=%s", crm_element_value(reason, XML_ATTR_ID), op, crm_element_value(reason, XML_NVPAIR_ATTR_NAME), crm_element_value(reason, XML_NVPAIR_ATTR_VALUE), abort_text, add[0], add[1], add[2], fn, line, path, (transition_graph->complete? "true" : "false")); } else if (safe_str_eq(XML_LRM_TAG_RSC_OP, kind)) { const char *magic = crm_element_value(reason, XML_ATTR_TRANSITION_MAGIC); do_crm_log(level, "Transition aborted by operation %s '%s' on %s: %s " CRM_XS " magic=%s cib=%d.%d.%d source=%s:%d complete=%s", crm_element_value(reason, XML_LRM_ATTR_TASK_KEY), op, crm_element_value(reason, XML_LRM_ATTR_TARGET), abort_text, magic, add[0], add[1], add[2], fn, line, (transition_graph->complete? "true" : "false")); } else if (safe_str_eq(XML_CIB_TAG_STATE, kind) || safe_str_eq(XML_CIB_TAG_NODE, kind)) { const char *uname = crm_peer_uname(ID(reason)); do_crm_log(level, "Transition aborted by %s '%s' on %s: %s " CRM_XS " cib=%d.%d.%d source=%s:%d complete=%s", kind, op, (uname? uname : ID(reason)), abort_text, add[0], add[1], add[2], fn, line, (transition_graph->complete? "true" : "false")); } else { do_crm_log(level, "Transition aborted by %s.%s '%s': %s " CRM_XS " cib=%d.%d.%d source=%s:%d path=%s complete=%s", TYPE(reason), ID(reason), (op? op : "change"), abort_text, add[0], add[1], add[2], fn, line, path, (transition_graph->complete? "true" : "false")); } } if (transition_graph->complete) { if (transition_timer->period_ms > 0) { crm_timer_stop(transition_timer); crm_timer_start(transition_timer); } else { register_fsa_input(C_FSA_INTERNAL, I_PE_CALC, NULL); } return; } mainloop_set_trigger(transition_trigger); } diff --git a/cts/patterns.py b/cts/patterns.py index 305ea73967..0230f12f61 100644 --- a/cts/patterns.py +++ b/cts/patterns.py @@ -1,546 +1,546 @@ import sys, os from cts.CTSvars import * patternvariants = {} class BasePatterns: def __init__(self, name): self.name = name patternvariants[name] = self self.ignore = [ "avoid confusing Valgrind", ] self.BadNews = [] self.components = {} self.commands = { "StatusCmd" : "crmadmin -t 60000 -S %s 2>/dev/null", "CibQuery" : "cibadmin -Ql", "CibAddXml" : "cibadmin --modify -c --xml-text %s", "CibDelXpath" : "cibadmin --delete --xpath %s", # 300,000 == 5 minutes "RscRunning" : CTSvars.CRM_DAEMON_DIR + "/lrmd_test -R -r %s", "CIBfile" : "%s:"+CTSvars.CRM_CONFIG_DIR+"/cib.xml", "TmpDir" : "/tmp", "BreakCommCmd" : "iptables -A INPUT -s %s -j DROP >/dev/null 2>&1", "FixCommCmd" : "iptables -D INPUT -s %s -j DROP >/dev/null 2>&1", # tc qdisc add dev lo root handle 1: cbq avpkt 1000 bandwidth 1000mbit # tc class add dev lo parent 1: classid 1:1 cbq rate "$RATE"kbps allot 17000 prio 5 bounded isolated # tc filter add dev lo parent 1: protocol ip prio 16 u32 match ip dst 127.0.0.1 match ip sport $PORT 0xFFFF flowid 1:1 # tc qdisc add dev lo parent 1: netem delay "$LATENCY"msec "$(($LATENCY/4))"msec 10% 2> /dev/null > /dev/null "ReduceCommCmd" : "", "RestoreCommCmd" : "tc qdisc del dev lo root", "UUIDQueryCmd" : "crmadmin -N", "SetCheckInterval" : "cibadmin --modify -c --xml-text ''", "ClearCheckInterval" : "cibadmin --delete --xpath \"//nvpair[@name='cluster-recheck-interval']\"", "MaintenanceModeOn" : "cibadmin --modify -c --xml-text ''", "MaintenanceModeOff" : "cibadmin --delete --xpath \"//nvpair[@name='maintenance-mode']\"", "StandbyCmd" : "crm_attribute -VQ -U %s -n standby -l forever -v %s 2>/dev/null", "StandbyQueryCmd" : "crm_attribute -QG -U %s -n standby -l forever -d off 2>/dev/null", } self.search = { "Pat:DC_IDLE" : "crmd.*State transition.*-> S_IDLE", # This wont work if we have multiple partitions "Pat:Local_started" : "%s\W.*The local CRM is operational", "Pat:Slave_started" : "%s\W.*State transition.*-> S_NOT_DC", "Pat:Master_started": "%s\W.*State transition.*-> S_IDLE", "Pat:We_stopped" : "heartbeat.*%s.*Heartbeat shutdown complete", "Pat:Logd_stopped" : "%s\W.*logd:.*Exiting write process", "Pat:They_stopped" : "%s\W.*LOST:.* %s ", "Pat:They_dead" : "node %s.*: is dead", "Pat:TransitionComplete" : "Transition status: Complete: complete", "Pat:Fencing_start" : "(Initiating remote operation|Requesting peer fencing ).* (for|of) %s", "Pat:Fencing_ok" : r"stonith.*:\s*Operation .* of %s by .* for .*@.*: OK", "Pat:Fencing_recover" : r"pengine.*: Recover %s", "Pat:RscOpOK" : r"crmd.*:\s+Result of %s operation for %s.*: (0 \()?ok", "Pat:RscRemoteOpOK" : r"crmd.*:\s+Result of %s operation for %s on %s: (0 \()?ok", - "Pat:NodeFenced" : r"crmd.*:\s*Peer\s+%s\s+was\s+terminated\s+\(.*\)\s+by\s+.*\s+for\s+.*:\s+OK", + "Pat:NodeFenced" : r"crmd.*:\s* Peer %s was terminated \(.*\) by .* on behalf of .*: OK", "Pat:FenceOpOK" : "Operation .* for host '%s' with device .* returned: 0", } def get_component(self, key): if key in self.components: return self.components[key] print("Unknown component '%s' for %s" % (key, self.name)) return [] def get_patterns(self, key): if key == "BadNews": return self.BadNews elif key == "BadNewsIgnore": return self.ignore elif key == "Commands": return self.commands elif key == "Search": return self.search elif key == "Components": return self.components def __getitem__(self, key): if key == "Name": return self.name elif key in self.commands: return self.commands[key] elif key in self.search: return self.search[key] else: print("Unknown template '%s' for %s" % (key, self.name)) return None class crm_lha(BasePatterns): def __init__(self, name): BasePatterns.__init__(self, name) self.commands.update({ "StartCmd" : "service heartbeat start > /dev/null 2>&1", "StopCmd" : "service heartbeat stop > /dev/null 2>&1", "EpochCmd" : "crm_node -H -e", "QuorumCmd" : "crm_node -H -q", "PartitionCmd" : "crm_node -H -p", }) self.search.update({ # Patterns to look for in the log files for various occasions... "Pat:ChildKilled" : "%s\W.*heartbeat.*%s.*killed by signal 9", "Pat:ChildRespawn" : "%s\W.*heartbeat.*Respawning client.*%s", "Pat:ChildExit" : "(ERROR|error): Client .* exited with return code", }) self.BadNews = [ r"error:", r"crit:", r"ERROR:", r"CRIT:", r"Shutting down...NOW", r"Timer I_TERMINATE just popped", r"input=I_ERROR", r"input=I_FAIL", r"input=I_INTEGRATED cause=C_TIMER_POPPED", r"input=I_FINALIZED cause=C_TIMER_POPPED", r"input=I_ERROR", r", exiting\.", r"WARN.*Ignoring HA message.*vote.*not in our membership list", r"pengine.*Attempting recovery of resource", r"is taking more than 2x its timeout", r"Confirm not received from", r"Welcome reply not received from", r"Attempting to schedule .* after a stop", r"Resource .* was active at shutdown", r"duplicate entries for call_id", r"Search terminated:", r"No need to invoke the TE", r"global_timer_callback:", r"Faking parameter digest creation", r"Parameters to .* action changed:", r"Parameters to .* changed", ] self.ignore = self.ignore + [ r"(ERROR|error):.*\s+assert\s+at\s+crm_glib_handler:" "(ERROR|error): Message hist queue is filling up", "stonithd.*CRIT: external_hostlist:.*'vmware gethosts' returned an empty hostlist", "stonithd.*(ERROR|error): Could not list nodes for stonith RA external/vmware.", "pengine.*Preventing .* from re-starting", ] class crm_cs_v0(BasePatterns): def __init__(self, name): BasePatterns.__init__(self, name) self.commands.update({ "EpochCmd" : "crm_node -e --openais", "QuorumCmd" : "crm_node -q --openais", "PartitionCmd" : "crm_node -p --openais", "StartCmd" : "service corosync start", "StopCmd" : "service corosync stop", }) self.search.update({ # The next pattern is too early # "Pat:We_stopped" : "%s.*Service engine unloaded: Pacemaker Cluster Manager", # The next pattern would be preferred, but it doesn't always come out # "Pat:We_stopped" : "%s.*Corosync Cluster Engine exiting with status", "Pat:We_stopped" : "%s\W.*Service engine unloaded: corosync cluster quorum service", "Pat:They_stopped" : "%s\W.*crmd.*Node %s(\[|\s).*state is now lost", "Pat:They_dead" : "corosync:.*Node %s is now: lost", "Pat:ChildExit" : "Child process .* exited", "Pat:ChildKilled" : "%s\W.*corosync.*Child process %s terminated with signal 9", "Pat:ChildRespawn" : "%s\W.*corosync.*Respawning failed child process: %s", "Pat:InfraUp" : "%s\W.*corosync.*Initializing transport", "Pat:PacemakerUp" : "%s\W.*corosync.*CRM: Initialized", }) self.ignore = self.ignore + [ r"crm_mon:", r"crmadmin:", r"update_trace_data", r"async_notify:.*strange, client not found", r"Parse error: Ignoring unknown option .*nodename", r"error.*: Operation 'reboot' .* with device 'FencingFail' returned:", r"Child process .* terminated with signal 9", r"getinfo response error: 1$", "sbd.* error: inquisitor_child: DEBUG MODE IS ACTIVE", r"sbd.* pcmk:\s*error:.*Connection to cib_ro failed", r"sbd.* pcmk:\s*error:.*Connection to cib_ro.* closed .I/O condition=17", ] self.BadNews = [ r"error:", r"crit:", r"ERROR:", r"CRIT:", r"Shutting down...NOW", r"Timer I_TERMINATE just popped", r"input=I_ERROR", r"input=I_FAIL", r"input=I_INTEGRATED cause=C_TIMER_POPPED", r"input=I_FINALIZED cause=C_TIMER_POPPED", r"input=I_ERROR", r", exiting\.", r"(WARN|warn).*Ignoring HA message.*vote.*not in our membership list", r"pengine.*Attempting recovery of resource", r"is taking more than 2x its timeout", r"Confirm not received from", r"Welcome reply not received from", r"Attempting to schedule .* after a stop", r"Resource .* was active at shutdown", r"duplicate entries for call_id", r"Search terminated:", r":global_timer_callback", r"Faking parameter digest creation", r"Parameters to .* action changed:", r"Parameters to .* changed", r"The .* process .* terminated with signal", r"Child process .* terminated with signal", r"pengine:.*Recover .*\(.* -\> .*\)", r"rsyslogd.* imuxsock lost .* messages from pid .* due to rate-limiting", r"Peer is not part of our cluster", r"We appear to be in an election loop", r"Unknown node -> we will not deliver message", r"(Blackbox dump requested|Problem detected)", r"pacemakerd.*Could not connect to Cluster Configuration Database API", r"Receiving messages from a node we think is dead", r"share the same cluster nodeid", r"share the same name", #r"crm_ipc_send:.*Request .* failed", #r"crm_ipc_send:.*Sending to .* is disabled until pending reply is received", # Not inherently bad, but worth tracking #r"No need to invoke the TE", #r"ping.*: DEBUG: Updated connected = 0", #r"Digest mis-match:", r"crmd:.*Transition failed: terminated", r"Local CIB .* differs from .*:", r"warn.*:\s*Continuing but .* will NOT be used", r"warn.*:\s*Cluster configuration file .* is corrupt", #r"Executing .* fencing operation", #r"fence_pcmk.* Call to fence", #r"fence_pcmk", r"cman killed by node", r"Election storm", r"stalled the FSA with pending inputs", ] self.components["common-ignore"] = [ "Pending action:", "error: crm_log_message_adv:", r"resource( was|s were) active at shutdown", "pending LRM operations at shutdown", "Lost connection to the CIB service", "Connection to the CIB terminated...", "Sending message to CIB service FAILED", "apply_xml_diff:.*Diff application failed!", r"crmd.*:\s*Action A_RECOVER .* not supported", "unconfirmed_actions:.*Waiting on .* unconfirmed actions", "cib_native_msgready:.*Message pending on command channel", r"crmd.*:\s*Performing A_EXIT_1 - forcefully exiting the CRMd", "verify_stopped:.*Resource .* was active at shutdown. You may ignore this error if it is unmanaged.", "error: attrd_connection_destroy:.*Lost connection to attrd", r".*:\s*Executing .* fencing operation \(.*\) on ", r".*:\s*Requesting fencing \([^)]+\) of node ", r"(Blackbox dump requested|Problem detected)", # "error: native_create_actions: Resource .*stonith::.* is active on 2 nodes attempting recovery", # "error: process_pe_message: Transition .* ERRORs found during PE processing", ] self.components["corosync-ignore"] = [ r"error:.*Connection to the CPG API failed: Library error", r"The .* process .* exited", r"pacemakerd.*error:.*Child process .* exited", r"cib.*error:.*Corosync connection lost", r"stonith-ng.*error:.*Corosync connection terminated", r"The cib process .* exited: Invalid argument", r"The attrd process .* exited: Transport endpoint is not connected", r"The crmd process .* exited: Link has been severed", r"error:.*Child process cib .* exited: Invalid argument", r"error:.*Child process attrd .* exited: Transport endpoint is not connected", r"error:.*Child process crmd .* exited: Link has been severed", r"lrmd.*error:.*Connection to stonith-ng failed", r"lrmd.*error:.*Connection to stonith-ng.* closed", r"lrmd.*error:.*LRMD lost STONITH connection", r"crmd.*State transition .* S_RECOVERY", r"crmd.*error:.*Input I_ERROR .*received in state", r"crmd.*error:.*Input I_TERMINATE .*received in state", r"crmd.*error:.*Connection to cman failed", r"crmd.*error:.*Could not recover from internal error", r"error:.*Connection to cib_shm failed", r"error:.*Connection to cib_shm.* closed", r"error:.*STONITH connection failed", r"error: Connection to stonith-ng failed", r"crit: Fencing daemon connection failed", r"error: Connection to stonith-ng.* closed", ] self.components["corosync"] = [ r"pacemakerd.*error:.*Connection destroyed", r"attrd.*:\s*crit:.*Lost connection to Corosync service", r"stonith.*:\s*(Corosync connection terminated|Shutting down)", r"cib.*:\s*Corosync connection lost!\s+Exiting.", r"crmd.*:\s*(connection terminated|Disconnected from Corosync)", r"pengine.*Scheduling Node .* for STONITH", r"crmd.*:\s*Peer .* was terminated \(.*\) by .* for .*:\s*OK", ] self.components["cib-ignore"] = [ "lrmd.*Connection to stonith-ng failed", "lrmd.*Connection to stonith-ng.* closed", "lrmd.*LRMD lost STONITH connection", "lrmd.*STONITH connection failed, finalizing .* pending operations", ] self.components["cib"] = [ "State transition .* S_RECOVERY", "Respawning .* crmd", "Respawning .* attrd", "Connection to cib_.* failed", "Connection to cib_.* closed", r"crmd.*:.*Connection to the CIB terminated...", r"attrd.*:.*(Lost connection to CIB service|Connection to the CIB terminated)", "(Child process|The) crmd .* exited: Generic Pacemaker error", "(Child process|The) attrd .* exited: (Connection reset by peer|Transport endpoint is not connected)", r"crmd.*: Input I_TERMINATE .*from do_recover", "crmd.*I_ERROR.*crmd_cib_connection_destroy", "crmd.*Could not recover from internal error", ] self.components["lrmd"] = [ "State transition .* S_RECOVERY", "LRM Connection failed", "Respawning .* crmd", "Connection to lrmd failed", "Connection to lrmd.* closed", "crmd.*I_ERROR.*lrm_connection_destroy", "(Child process|The) crmd .* exited: Generic Pacemaker error", r"crmd.*: Input I_TERMINATE .*from do_recover", "crmd.*Could not recover from internal error", ] self.components["lrmd-ignore"] = [] self.components["crmd"] = [ # "WARN: determine_online_status: Node .* is unclean", # "Scheduling Node .* for STONITH", # "Executing .* fencing operation", # Only if the node wasn't the DC: "State transition S_IDLE", "State transition .* -> S_IDLE", ] self.components["crmd-ignore"] = [] self.components["attrd"] = [] self.components["attrd-ignore"] = [] self.components["pengine"] = [ "State transition .* S_RECOVERY", "Respawning .* crmd", "(The|Child process) crmd .* exited: Generic Pacemaker error", "Connection to pengine failed", "Connection to pengine.* closed", "Connection to the Policy Engine failed", "crmd.*I_ERROR.*save_cib_contents", r"crmd.*: Input I_TERMINATE .*from do_recover", "crmd.*Could not recover from internal error", ] self.components["pengine-ignore"] = [] self.components["stonith"] = [ "Connection to stonith-ng failed", "LRMD lost STONITH connection", "Connection to stonith-ng.* closed", "Fencing daemon connection failed", r"crmd.*:\s*warn.*:\s*Callback already present", ] self.components["stonith-ignore"] = [ r"pengine.*: Recover Fencing", r"Updating failcount for Fencing", r"error:.*Connection to stonith-ng failed", r"error:.*Connection to stonith-ng.*closed \(I/O condition=17\)", r"crit:.*Fencing daemon connection failed", r"error:.*Sign-in failed: triggered a retry", "STONITH connection failed, finalizing .* pending operations.", r"crmd.*:\s+Result of .* operation for Fencing.*Error", ] self.components["stonith-ignore"].extend(self.components["common-ignore"]) class crm_mcp(crm_cs_v0): ''' The crm version 4 cluster manager class. It implements the things we need to talk to and manipulate crm clusters running on top of native corosync (no plugins) ''' def __init__(self, name): crm_cs_v0.__init__(self, name) self.commands.update({ "StartCmd" : "service corosync start && service pacemaker start", "StopCmd" : "service pacemaker stop; [ ! -e /usr/sbin/pacemaker_remoted ] || service pacemaker_remote stop; service corosync stop", "EpochCmd" : "crm_node -e", "QuorumCmd" : "crm_node -q", "PartitionCmd" : "crm_node -p", }) self.search.update({ # Close enough... "Corosync Cluster Engine exiting normally" isn't printed # reliably and there's little interest in doing anything about it "Pat:We_stopped" : "%s\W.*Unloading all Corosync service engines", "Pat:They_stopped" : "%s\W.*crmd.*Node %s(\[|\s).*state is now lost", "Pat:They_dead" : "crmd.*Node %s(\[|\s).*state is now lost", "Pat:ChildExit" : "The .* process exited", "Pat:ChildKilled" : "%s\W.*pacemakerd.*The %s process .* terminated with signal 9", "Pat:ChildRespawn" : "%s\W.*pacemakerd.*Respawning failed child process: %s", "Pat:PacemakerUp" : "%s\W.*pacemakerd.*Starting Pacemaker", }) # if self.Env["have_systemd"]: # self.update({ # # When systemd is in use, we can look for this instead # "Pat:We_stopped" : "%s.*Stopped Corosync Cluster Engine", # }) class crm_mcp_docker(crm_mcp): ''' The crm version 4 cluster manager class. It implements the things we need to talk to and manipulate crm clusters running on top of native corosync (no plugins) ''' def __init__(self, name): crm_mcp.__init__(self, name) self.commands.update({ "StartCmd" : "pcmk_start", "StopCmd" : "pcmk_stop", }) class crm_cman(crm_cs_v0): ''' The crm version 3 cluster manager class. It implements the things we need to talk to and manipulate crm clusters running on top of openais ''' def __init__(self, name): crm_cs_v0.__init__(self, name) self.commands.update({ "StartCmd" : "service pacemaker start", "StopCmd" : "service pacemaker stop; [ ! -e /usr/sbin/pacemaker_remoted ] || service pacemaker_remote stop", "EpochCmd" : "crm_node -e --cman", "QuorumCmd" : "crm_node -q --cman", "PartitionCmd" : "crm_node -p --cman", }) self.search.update({ "Pat:We_stopped" : "%s.*Unloading all Corosync service engines", "Pat:They_stopped" : "%s\W.*crmd.*Node %s(\[|\s).*state is now lost", "Pat:They_dead" : "crmd.*Node %s(\[|\s).*state is now lost", "Pat:ChildKilled" : "%s\W.*pacemakerd.*The %s process .* terminated with signal 9", "Pat:ChildRespawn" : "%s\W.*pacemakerd.*Respawning failed child process: %s", "Pat:PacemakerUp" : "%s\W.*pacemakerd.*Starting Pacemaker", }) class PatternSelector: def __init__(self, name=None): self.name = name self.base = BasePatterns("crm-base") if not name: crm_cs_v0("crm-plugin-v0") crm_cman("crm-cman") crm_mcp("crm-mcp") crm_lha("crm-lha") elif name == "crm-lha": crm_lha(name) elif name == "crm-plugin-v0": crm_cs_v0(name) elif name == "crm-cman": crm_cman(name) elif name == "crm-mcp": crm_mcp(name) elif name == "crm-mcp-docker": crm_mcp_docker(name) def get_variant(self, variant): if variant in patternvariants: return patternvariants[variant] print("defaulting to crm-base for %s" % variant) return self.base def get_patterns(self, variant, kind): return self.get_variant(variant).get_patterns(kind) def get_template(self, variant, key): v = self.get_variant(variant) return v[key] def get_component(self, variant, kind): return self.get_variant(variant).get_component(kind) def __getitem__(self, key): return self.get_template(self.name, key) # python cts/CTSpatt.py -k crm-mcp -t StartCmd if __name__ == '__main__': pdir=os.path.dirname(sys.path[0]) sys.path.insert(0, pdir) # So that things work from the source directory kind=None template=None skipthis=None args=sys.argv[1:] for i in range(0, len(args)): if skipthis: skipthis=None continue elif args[i] == "-k" or args[i] == "--kind": skipthis=1 kind = args[i+1] elif args[i] == "-t" or args[i] == "--template": skipthis=1 template = args[i+1] else: print("Illegal argument " + args[i]) print(PatternSelector(kind)[template])