diff --git a/daemons/controld/controld_callbacks.c b/daemons/controld/controld_callbacks.c index 0a03d14254..9965b5c841 100644 --- a/daemons/controld/controld_callbacks.c +++ b/daemons/controld/controld_callbacks.c @@ -1,288 +1,294 @@ /* * Copyright 2004-2018 Andrew Beekhof * * 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 /* 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); } #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; uint32_t changed = 0; bool appeared = FALSE; bool is_remote = is_set(node->flags, crm_remote_node); const char *status = NULL; /* 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); } } crmd_alert_node_event(node); break; case crm_status_processes: if (data) { old = *(const uint32_t *)data; changed = node->processes ^ old; } status = (node->processes & proc_flags) ? ONLINESTATUS : OFFLINESTATUS; crm_info("Client %s/%s now has status [%s] (DC=%s, changed=%6x)", node->uname, peer2text(proc_flags), status, AM_I_DC ? "true" : crm_str(fsa_our_dc), changed); if ((changed & proc_flags) == 0) { /* Peer process did not change */ crm_trace("No change %6x %6x %6x", old, node->processes, proc_flags); return; } else if (is_not_set(fsa_input_register, R_CIB_CONNECTED)) { crm_trace("Not connected"); return; } else if (fsa_state == S_STOPPING) { crm_trace("Stopping"); return; } appeared = (node->processes & proc_flags) != 0; if (safe_str_eq(node->uname, fsa_our_uname) && (node->processes & proc_flags) == 0) { /* 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 && appeared == FALSE) { - crm_info("Peer %s left us", node->uname); - erase_status_tag(node->uname, XML_TAG_TRANSIENT_NODEATTRS, cib_scope_local); + } else if(AM_I_DC) { + if (appeared == FALSE) { + crm_info("Peer %s left us", node->uname); + erase_status_tag(node->uname, XML_TAG_TRANSIENT_NODEATTRS, cib_scope_local); + } else { + crm_info("New peer %s we want to sync fence history with", + node->uname); + te_trigger_stonith_history_sync(); + } } break; } if (AM_I_DC) { xmlNode *update = NULL; int flags = node_update_peer; gboolean alive = is_remote? appeared : crm_is_peer_active(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 (alive && type == crm_status_processes) { 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 ((alive == FALSE) && safe_str_eq(task, CRM_OP_SHUTDOWN)) { crm_notice("%s of peer %s is complete "CRM_XS" op=%d", task, node->uname, down->id); /* down->confirmed = TRUE; */ stop_te_timer(down->timer); if (!is_remote) { flags |= node_update_join | node_update_expected; crmd_peer_down(node, FALSE); check_join_state(fsa_state, __FUNCTION__); } update_graph(transition_graph, down); trigger_graph(); } else { crm_trace("Node %s is %salive, was expected to %s (op %d)", node->uname, (alive? "" : "not "), 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__); fsa_cib_anon_update(XML_CIB_TAG_STATUS, update, cib_scope_local | cib_quorum_override | cib_can_create); 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/controld/controld_te_utils.c b/daemons/controld/controld_te_utils.c index 3fb4e85728..21e6398190 100644 --- a/daemons/controld/controld_te_utils.c +++ b/daemons/controld/controld_te_utils.c @@ -1,663 +1,717 @@ /* * Copyright 2004-2018 Andrew Beekhof * * 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 crm_trigger_t *stonith_reconnect = NULL; +static crm_trigger_t *stonith_history_sync_trigger = NULL; +static mainloop_timer_t *stonith_history_sync_timer = 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): fencer terminated", action->id, ID(action->xml)); } } } if (last_action != NULL) { crm_warn("Fencer failure resulted in unrunnable 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(); } } 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.%lu", crm_system_name, (unsigned long) 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(CRM_EX_FATAL); // None of our wrappers since we already called qb_log_fini() return; } /* Update the count of stonith failures for this target, in case we become * DC later. The current DC has already updated its fail count in * tengine_stonith_callback(). */ if (!AM_I_DC && 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 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->client_origin? st_event->client_origin : ""), pcmk_strerror(st_event->result), st_event->origin, st_event->id); 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 controller 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); } } +static gboolean +do_stonith_history_sync(gpointer user_data) +{ + if (stonith_api && (stonith_api->state != stonith_disconnected)) { + stonith_history_t *history = NULL; + + stonith_api->cmds->history(stonith_api, + st_opt_sync_call | st_opt_broadcast, + NULL, &history, 5); + stonith_history_free(history); + return TRUE; + } else { + crm_info("Skip triggering stonith history-sync as stonith is disconnected"); + return FALSE; + } +} + +static gboolean +stonith_history_sync_set_trigger(gpointer user_data) +{ + mainloop_set_trigger(stonith_history_sync_trigger); + return FALSE; +} + +void +te_trigger_stonith_history_sync(void) +{ + /* trigger a sync in 5s to give more nodes the + * chance to show up so that we don't create + * unnecessary stonith-history-sync traffic + */ + + /* as we are finally checking the stonith-connection + * in do_stonith_history_sync we should be fine + * leaving stonith_history_sync_time & stonith_history_sync_trigger + * arround + */ + if (stonith_history_sync_trigger == NULL) { + stonith_history_sync_trigger = + mainloop_add_trigger(G_PRIORITY_LOW, + do_stonith_history_sync, NULL); + } + + if(stonith_history_sync_timer == NULL) { + stonith_history_sync_timer = + mainloop_timer_add("history_sync", 5000, + FALSE, stonith_history_sync_set_trigger, + NULL); + } + mainloop_timer_start(stonith_history_sync_timer); +} + 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) { if (timer == NULL) { return FALSE; } if (timer->source_id != 0) { crm_trace("Stopping action timer"); g_source_remove(timer->source_id); timer->source_id = 0; } else { crm_trace("Action timer was already stopped"); 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_TRACE, 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); } static struct abort_timer_s { bool aborted; guint id; int priority; enum transition_action action; const char *text; } abort_timer = { 0, }; static gboolean abort_timer_popped(gpointer data) { if (AM_I_DC && (abort_timer.aborted == FALSE)) { abort_transition(abort_timer.priority, abort_timer.action, abort_timer.text, NULL); } abort_timer.id = 0; return FALSE; // do not immediately reschedule timer } /*! * \internal * \brief Abort transition after delay, if not already aborted in that time * * \param[in] abort_text Must be literal string */ void abort_after_delay(int abort_priority, enum transition_action abort_action, const char *abort_text, guint delay_ms) { if (abort_timer.id) { // Timer already in progress, stop and reschedule g_source_remove(abort_timer.id); } abort_timer.aborted = FALSE; abort_timer.priority = abort_priority; abort_timer.action = abort_action; abort_timer.text = abort_text; abort_timer.id = g_timeout_add(delay_ms, abort_timer_popped, NULL); } 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; } abort_timer.aborted = TRUE; /* 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 %d aborted: %s "CRM_XS" source=%s:%d complete=%s", transition_graph->id, 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 %d aborted by %s.%s: %s " CRM_XS " cib=%d.%d.%d source=%s:%d path=%s complete=%s", transition_graph->id, 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 %d aborted by deletion of %s: %s " CRM_XS " cib=%d.%d.%d source=%s:%d path=%s complete=%s", transition_graph->id, (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 %d aborted by %s doing %s %s=%s: %s " CRM_XS " cib=%d.%d.%d source=%s:%d path=%s complete=%s", transition_graph->id, 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 %d aborted by operation %s '%s' on %s: %s " CRM_XS " magic=%s cib=%d.%d.%d source=%s:%d complete=%s", transition_graph->id, 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 %d aborted by %s '%s' on %s: %s " CRM_XS " cib=%d.%d.%d source=%s:%d complete=%s", transition_graph->id, kind, op, (uname? uname : ID(reason)), abort_text, add[0], add[1], add[2], fn, line, (transition_graph->complete? "true" : "false")); } else { const char *id = ID(reason); do_crm_log(level, "Transition %d aborted by %s.%s '%s': %s " CRM_XS " cib=%d.%d.%d source=%s:%d path=%s complete=%s", transition_graph->id, TYPE(reason), (id? id : ""), (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/daemons/controld/controld_transition.h b/daemons/controld/controld_transition.h index 563030642b..e2361f8c37 100644 --- a/daemons/controld/controld_transition.h +++ b/daemons/controld/controld_transition.h @@ -1,83 +1,85 @@ /* * Copyright 2004-2018 Andrew Beekhof * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #ifndef TENGINE__H # define TENGINE__H # include # include # include # include extern stonith_t *stonith_api; extern void send_stonith_update(crm_action_t * stonith_action, const char *target, const char *uuid); /* stonith cleanup list */ void add_stonith_cleanup(const char *target); void remove_stonith_cleanup(const char *target); void purge_stonith_cleanup(void); void execute_stonith_cleanup(void); /* tengine */ extern crm_action_t *match_down_event(const char *target); extern crm_action_t *get_cancel_action(const char *id, const char *node); void controld_record_action_timeout(crm_action_t *action); extern gboolean fail_incompletable_actions(crm_graph_t * graph, const char *down_node); void process_graph_event(xmlNode *event, const char *event_node); /* utils */ extern crm_action_t *get_action(int id, gboolean confirmed); extern gboolean stop_te_timer(crm_action_timer_t * timer); extern const char *get_rsc_state(const char *task, enum op_status status); /* unpack */ extern gboolean process_te_message(xmlNode * msg, xmlNode * xml_data); extern crm_graph_t *transition_graph; extern crm_trigger_t *transition_trigger; extern char *te_uuid; extern void notify_crmd(crm_graph_t * graph); void cib_fencing_updated(xmlNode *msg, int call_id, int rc, xmlNode *output, void *user_data); void cib_action_updated(xmlNode *msg, int call_id, int rc, xmlNode *output, void *user_data); gboolean action_timer_callback(gpointer data); gboolean te_graph_trigger(gpointer user_data); void te_update_diff(const char *event, xmlNode *msg); void tengine_stonith_callback(stonith_t *stonith, stonith_callback_data_t *data); void update_stonith_max_attempts(const char* value); extern void trigger_graph_processing(const char *fn, int line); void abort_after_delay(int abort_priority, enum transition_action abort_action, const char *abort_text, guint delay_ms); extern void abort_transition_graph(int abort_priority, enum transition_action abort_action, const char *abort_text, xmlNode * reason, const char *fn, int line); # define trigger_graph() trigger_graph_processing(__FUNCTION__, __LINE__) # define abort_transition(pri, action, text, reason) \ abort_transition_graph(pri, action, text, reason,__FUNCTION__,__LINE__); extern gboolean te_connect_stonith(gpointer user_data); +extern void te_trigger_stonith_history_sync(void); + extern crm_trigger_t *transition_trigger; extern crm_trigger_t *stonith_reconnect; extern char *failed_stop_offset; extern char *failed_start_offset; extern int active_timeout; extern int stonith_op_active; void te_action_confirmed(crm_action_t * action); void te_reset_job_counts(void); #endif diff --git a/daemons/fenced/Makefile.am b/daemons/fenced/Makefile.am index e9dbf1a9de..59d2674d20 100644 --- a/daemons/fenced/Makefile.am +++ b/daemons/fenced/Makefile.am @@ -1,49 +1,50 @@ # # Copyright 2004-2018 International Business Machines # Author: Sun Jiang Dong # # This source code is licensed under the GNU General Public License version 2 # or later (GPLv2+) WITHOUT ANY WARRANTY. # include $(top_srcdir)/Makefile.common halibdir = $(CRM_DAEMON_DIR) halib_PROGRAMS = pacemaker-fenced cts-fence-helper sbin_SCRIPTS = fence_legacy noinst_HEADERS = pacemaker-fenced.h if BUILD_XML_HELP man7_MANS = pacemaker-fenced.7 endif cts_fence_helper_SOURCES = cts-fence-helper.c cts_fence_helper_LDADD = $(top_builddir)/lib/common/libcrmcommon.la \ $(top_builddir)/lib/fencing/libstonithd.la pacemaker_fenced_CPPFLAGS = -I$(top_srcdir)/daemons/schedulerd $(AM_CPPFLAGS) pacemaker_fenced_YFLAGS = -d pacemaker_fenced_CFLAGS = $(CFLAGS_HARDENED_EXE) pacemaker_fenced_LDFLAGS = $(LDFLAGS_HARDENED_EXE) pacemaker_fenced_LDADD = $(top_builddir)/lib/common/libcrmcommon.la \ $(top_builddir)/lib/cluster/libcrmcluster.la \ $(top_builddir)/lib/fencing/libstonithd.la \ $(top_builddir)/lib/pengine/libpe_status.la \ $(top_builddir)/daemons/schedulerd/libpengine.la \ $(CLUSTERLIBS) pacemaker_fenced_SOURCES = pacemaker-fenced.c \ fenced_commands.c \ - fenced_remote.c + fenced_remote.c \ + fenced_history.c CLEANFILES = $(man7_MANS) $(man8_MANS) if BUILD_LEGACY_LINKS install-exec-hook: cd $(DESTDIR)$(CRM_DAEMON_DIR) && rm -f stonithd && $(LN_S) pacemaker-fenced stonithd uninstall-hook: cd $(DESTDIR)$(CRM_DAEMON_DIR) && rm -f stonithd endif diff --git a/daemons/fenced/fenced_commands.c b/daemons/fenced/fenced_commands.c index 4eb8e02f2b..1fdcee7f4f 100644 --- a/daemons/fenced/fenced_commands.c +++ b/daemons/fenced/fenced_commands.c @@ -1,2734 +1,2742 @@ /* * Copyright 2009-2018 Andrew Beekhof * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #if SUPPORT_CIBSECRETS # include #endif #include GHashTable *device_list = NULL; GHashTable *topology = NULL; GList *cmd_list = NULL; struct device_search_s { /* target of fence action */ char *host; /* requested fence action */ char *action; /* timeout to use if a device is queried dynamically for possible targets */ int per_device_timeout; /* number of registered fencing devices at time of request */ int replies_needed; /* number of device replies received so far */ int replies_received; /* whether the target is eligible to perform requested action (or off) */ bool allow_suicide; /* private data to pass to search callback function */ void *user_data; /* function to call when all replies have been received */ void (*callback) (GList * devices, void *user_data); /* devices capable of performing requested action (or off if remapping) */ GListPtr capable; }; static gboolean stonith_device_dispatch(gpointer user_data); static void st_child_done(GPid pid, int rc, const char *output, gpointer user_data); static void stonith_send_reply(xmlNode * reply, int call_options, const char *remote_peer, const char *client_id); static void search_devices_record_result(struct device_search_s *search, const char *device, gboolean can_fence); typedef struct async_command_s { int id; int pid; int fd_stdout; int options; int default_timeout; /* seconds */ int timeout; /* seconds */ int start_delay; /* milliseconds */ int delay_id; char *op; char *origin; char *client; char *client_name; char *remote_op_id; char *victim; uint32_t victim_nodeid; char *action; char *device; char *mode; GListPtr device_list; GListPtr device_next; void *internal_user_data; void (*done_cb) (GPid pid, int rc, const char *output, gpointer user_data); guint timer_sigterm; guint timer_sigkill; /*! If the operation timed out, this is the last signal * we sent to the process to get it to terminate */ int last_timeout_signo; stonith_device_t *active_on; } async_command_t; static xmlNode *stonith_construct_async_reply(async_command_t * cmd, const char *output, xmlNode * data, int rc); static gboolean is_action_required(const char *action, stonith_device_t *device) { return device && device->automatic_unfencing && safe_str_eq(action, "on"); } static int get_action_delay_max(stonith_device_t * device, const char * action) { const char *value = NULL; int delay_max_ms = 0; if (safe_str_neq(action, "off") && safe_str_neq(action, "reboot")) { return 0; } value = g_hash_table_lookup(device->params, STONITH_ATTR_DELAY_MAX); if (value) { delay_max_ms = crm_get_msec(value); } return delay_max_ms; } static int get_action_delay_base(stonith_device_t * device, const char * action) { const char *value = NULL; int delay_base_ms = 0; if (safe_str_neq(action, "off") && safe_str_neq(action, "reboot")) { return 0; } value = g_hash_table_lookup(device->params, STONITH_ATTR_DELAY_BASE); if (value) { delay_base_ms = crm_get_msec(value); } return delay_base_ms; } /*! * \internal * \brief Override STONITH timeout with pcmk_*_timeout if available * * \param[in] device STONITH device to use * \param[in] action STONITH action name * \param[in] default_timeout Timeout to use if device does not have * a pcmk_*_timeout parameter for action * * \return Value of pcmk_(action)_timeout if available, otherwise default_timeout * \note For consistency, it would be nice if reboot/off/on timeouts could be * set the same way as start/stop/monitor timeouts, i.e. with an * entry in the fencing resource configuration. However that * is insufficient because fencing devices may be registered directly via * the fencer's register_device() API instead of going through the CIB * (e.g. stonith_admin uses it for its -R option, and the executor uses it * to ensure a device is registered when a command is issued). As device * properties, pcmk_*_timeout parameters can be grabbed by the fencer when * the device is registered, whether by CIB change or API call. */ static int get_action_timeout(stonith_device_t * device, const char *action, int default_timeout) { if (action && device && device->params) { char buffer[64] = { 0, }; const char *value = NULL; /* If "reboot" was requested but the device does not support it, * we will remap to "off", so check timeout for "off" instead */ if (safe_str_eq(action, "reboot") && is_not_set(device->flags, st_device_supports_reboot)) { crm_trace("%s doesn't support reboot, using timeout for off instead", device->id); action = "off"; } /* If the device config specified an action-specific timeout, use it */ snprintf(buffer, sizeof(buffer), "pcmk_%s_timeout", action); value = g_hash_table_lookup(device->params, buffer); if (value) { return atoi(value); } } return default_timeout; } static void free_async_command(async_command_t * cmd) { if (!cmd) { return; } if (cmd->delay_id) { g_source_remove(cmd->delay_id); } cmd_list = g_list_remove(cmd_list, cmd); g_list_free_full(cmd->device_list, free); free(cmd->device); free(cmd->action); free(cmd->victim); free(cmd->remote_op_id); free(cmd->client); free(cmd->client_name); free(cmd->origin); free(cmd->mode); free(cmd->op); free(cmd); } static async_command_t * create_async_command(xmlNode * msg) { async_command_t *cmd = NULL; xmlNode *op = get_xpath_object("//@" F_STONITH_ACTION, msg, LOG_ERR); const char *action = crm_element_value(op, F_STONITH_ACTION); CRM_CHECK(action != NULL, crm_log_xml_warn(msg, "NoAction"); return NULL); crm_log_xml_trace(msg, "Command"); cmd = calloc(1, sizeof(async_command_t)); crm_element_value_int(msg, F_STONITH_CALLID, &(cmd->id)); crm_element_value_int(msg, F_STONITH_CALLOPTS, &(cmd->options)); crm_element_value_int(msg, F_STONITH_TIMEOUT, &(cmd->default_timeout)); cmd->timeout = cmd->default_timeout; cmd->origin = crm_element_value_copy(msg, F_ORIG); cmd->remote_op_id = crm_element_value_copy(msg, F_STONITH_REMOTE_OP_ID); cmd->client = crm_element_value_copy(msg, F_STONITH_CLIENTID); cmd->client_name = crm_element_value_copy(msg, F_STONITH_CLIENTNAME); cmd->op = crm_element_value_copy(msg, F_STONITH_OPERATION); cmd->action = strdup(action); cmd->victim = crm_element_value_copy(op, F_STONITH_TARGET); cmd->mode = crm_element_value_copy(op, F_STONITH_MODE); cmd->device = crm_element_value_copy(op, F_STONITH_DEVICE); CRM_CHECK(cmd->op != NULL, crm_log_xml_warn(msg, "NoOp"); free_async_command(cmd); return NULL); CRM_CHECK(cmd->client != NULL, crm_log_xml_warn(msg, "NoClient")); cmd->done_cb = st_child_done; cmd_list = g_list_append(cmd_list, cmd); return cmd; } static int get_action_limit(stonith_device_t * device) { const char *value = NULL; int action_limit = 1; value = g_hash_table_lookup(device->params, STONITH_ATTR_ACTION_LIMIT); if (value) { action_limit = crm_parse_int(value, "1"); if (action_limit == 0) { /* pcmk_action_limit should not be 0. Enforce it to be 1. */ action_limit = 1; } } return action_limit; } static int get_active_cmds(stonith_device_t * device) { int counter = 0; GListPtr gIter = NULL; GListPtr gIterNext = NULL; CRM_CHECK(device != NULL, return 0); for (gIter = cmd_list; gIter != NULL; gIter = gIterNext) { async_command_t *cmd = gIter->data; gIterNext = gIter->next; if (cmd->active_on == device) { counter++; } } return counter; } static gboolean stonith_device_execute(stonith_device_t * device) { int exec_rc = 0; const char *action_str = NULL; async_command_t *cmd = NULL; stonith_action_t *action = NULL; int active_cmds = 0; int action_limit = 0; CRM_CHECK(device != NULL, return FALSE); active_cmds = get_active_cmds(device); action_limit = get_action_limit(device); if (action_limit > -1 && active_cmds >= action_limit) { crm_trace("%s is over its action limit of %d (%u active action%s)", device->id, action_limit, active_cmds, active_cmds > 1 ? "s" : ""); return TRUE; } if (device->pending_ops) { GList *first = device->pending_ops; cmd = first->data; if (cmd && cmd->delay_id) { crm_trace ("Operation %s%s%s on %s was asked to run too early, waiting for start_delay timeout of %dms", cmd->action, cmd->victim ? " for node " : "", cmd->victim ? cmd->victim : "", device->id, cmd->start_delay); return TRUE; } device->pending_ops = g_list_remove_link(device->pending_ops, first); g_list_free_1(first); } if (cmd == NULL) { crm_trace("Nothing further to do for %s", device->id); return TRUE; } if(safe_str_eq(device->agent, STONITH_WATCHDOG_AGENT)) { if(safe_str_eq(cmd->action, "reboot")) { pcmk_panic(__FUNCTION__); return TRUE; } else if(safe_str_eq(cmd->action, "off")) { pcmk_panic(__FUNCTION__); return TRUE; } else { crm_info("Faking success for %s watchdog operation", cmd->action); cmd->done_cb(0, 0, NULL, cmd); return TRUE; } } #if SUPPORT_CIBSECRETS if (replace_secret_params(device->id, device->params) < 0) { /* replacing secrets failed! */ if (safe_str_eq(cmd->action,"stop")) { /* don't fail on stop! */ crm_info("proceeding with the stop operation for %s", device->id); } else { crm_err("failed to get secrets for %s, " "considering resource not configured", device->id); exec_rc = PCMK_OCF_NOT_CONFIGURED; cmd->done_cb(0, exec_rc, NULL, cmd); return TRUE; } } #endif action_str = cmd->action; if (safe_str_eq(cmd->action, "reboot") && is_not_set(device->flags, st_device_supports_reboot)) { crm_warn("Agent '%s' does not advertise support for 'reboot', performing 'off' action instead", device->agent); action_str = "off"; } action = stonith_action_create(device->agent, action_str, cmd->victim, cmd->victim_nodeid, cmd->timeout, device->params, device->aliases); /* for async exec, exec_rc is pid if positive and error code if negative/zero */ exec_rc = stonith_action_execute_async(action, (void *)cmd, cmd->done_cb); if (exec_rc > 0) { crm_debug("Operation %s%s%s on %s now running with pid=%d, timeout=%ds", cmd->action, cmd->victim ? " for node " : "", cmd->victim ? cmd->victim : "", device->id, exec_rc, cmd->timeout); cmd->active_on = device; } else { crm_warn("Operation %s%s%s on %s failed: %s (%d)", cmd->action, cmd->victim ? " for node " : "", cmd->victim ? cmd->victim : "", device->id, pcmk_strerror(exec_rc), exec_rc); cmd->done_cb(0, exec_rc, NULL, cmd); } return TRUE; } static gboolean stonith_device_dispatch(gpointer user_data) { return stonith_device_execute(user_data); } static gboolean start_delay_helper(gpointer data) { async_command_t *cmd = data; stonith_device_t *device = NULL; cmd->delay_id = 0; device = cmd->device ? g_hash_table_lookup(device_list, cmd->device) : NULL; if (device) { mainloop_set_trigger(device->work); } return FALSE; } static void schedule_stonith_command(async_command_t * cmd, stonith_device_t * device) { int delay_max = 0; int delay_base = 0; CRM_CHECK(cmd != NULL, return); CRM_CHECK(device != NULL, return); if (cmd->device) { free(cmd->device); } if (device->include_nodeid && cmd->victim) { crm_node_t *node = crm_get_peer(0, cmd->victim); cmd->victim_nodeid = node->id; } cmd->device = strdup(device->id); cmd->timeout = get_action_timeout(device, cmd->action, cmd->default_timeout); if (cmd->remote_op_id) { crm_debug("Scheduling %s on %s for remote peer %s with op id (%s) (timeout=%ds)", cmd->action, device->id, cmd->origin, cmd->remote_op_id, cmd->timeout); } else { crm_debug("Scheduling %s on %s for %s (timeout=%ds)", cmd->action, device->id, cmd->client, cmd->timeout); } device->pending_ops = g_list_append(device->pending_ops, cmd); mainloop_set_trigger(device->work); delay_max = get_action_delay_max(device, cmd->action); delay_base = get_action_delay_base(device, cmd->action); if (delay_max == 0) { delay_max = delay_base; } if (delay_max < delay_base) { crm_warn("Base-delay (%dms) is larger than max-delay (%dms) " "for %s on %s - limiting to max-delay", delay_base, delay_max, cmd->action, device->id); delay_base = delay_max; } if (delay_max > 0) { // coverity[dont_call] We're not using rand() for security cmd->start_delay = ((delay_max != delay_base)?(rand() % (delay_max - delay_base)):0) + delay_base; crm_notice("Delaying %s on %s for %dms (timeout=%ds, base=%dms, " "max=%dms)", cmd->action, device->id, cmd->start_delay, cmd->timeout, delay_base, delay_max); cmd->delay_id = g_timeout_add(cmd->start_delay, start_delay_helper, cmd); } } static void free_device(gpointer data) { GListPtr gIter = NULL; stonith_device_t *device = data; g_hash_table_destroy(device->params); g_hash_table_destroy(device->aliases); for (gIter = device->pending_ops; gIter != NULL; gIter = gIter->next) { async_command_t *cmd = gIter->data; crm_warn("Removal of device '%s' purged operation %s", device->id, cmd->action); cmd->done_cb(0, -ENODEV, NULL, cmd); } g_list_free(device->pending_ops); g_list_free_full(device->targets, free); mainloop_destroy_trigger(device->work); free_xml(device->agent_metadata); free(device->namespace); free(device->on_target_actions); free(device->agent); free(device->id); free(device); } void free_device_list() { if (device_list != NULL) { g_hash_table_destroy(device_list); device_list = NULL; } } void init_device_list() { if (device_list == NULL) { device_list = g_hash_table_new_full(crm_str_hash, g_str_equal, NULL, free_device); } } static GHashTable * build_port_aliases(const char *hostmap, GListPtr * targets) { char *name = NULL; int last = 0, lpc = 0, max = 0, added = 0; GHashTable *aliases = crm_strcase_table_new(); if (hostmap == NULL) { return aliases; } max = strlen(hostmap); for (; lpc <= max; lpc++) { switch (hostmap[lpc]) { /* Assignment chars */ case '=': case ':': if (lpc > last) { free(name); name = calloc(1, 1 + lpc - last); memcpy(name, hostmap + last, lpc - last); } last = lpc + 1; break; /* Delimeter chars */ /* case ',': Potentially used to specify multiple ports */ case 0: case ';': case ' ': case '\t': if (name) { char *value = NULL; value = calloc(1, 1 + lpc - last); memcpy(value, hostmap + last, lpc - last); crm_debug("Adding alias '%s'='%s'", name, value); g_hash_table_replace(aliases, name, value); if (targets) { *targets = g_list_append(*targets, strdup(value)); } value = NULL; name = NULL; added++; } else if (lpc > last) { crm_debug("Parse error at offset %d near '%s'", lpc - last, hostmap + last); } last = lpc + 1; break; } if (hostmap[lpc] == 0) { break; } } if (added == 0) { crm_info("No host mappings detected in '%s'", hostmap); } free(name); return aliases; } static void parse_host_line(const char *line, int max, GListPtr * output) { int lpc = 0; int last = 0; if (max <= 0) { return; } /* Check for any complaints about additional parameters that the device doesn't understand */ if (strstr(line, "invalid") || strstr(line, "variable")) { crm_debug("Skipping: %s", line); return; } crm_trace("Processing %d bytes: [%s]", max, line); /* Skip initial whitespace */ for (lpc = 0; lpc <= max && isspace(line[lpc]); lpc++) { last = lpc + 1; } /* Now the actual content */ for (lpc = 0; lpc <= max; lpc++) { gboolean a_space = isspace(line[lpc]); if (a_space && lpc < max && isspace(line[lpc + 1])) { /* fast-forward to the end of the spaces */ } else if (a_space || line[lpc] == ',' || line[lpc] == ';' || line[lpc] == 0) { int rc = 1; char *entry = NULL; if (lpc != last) { entry = calloc(1, 1 + lpc - last); rc = sscanf(line + last, "%[a-zA-Z0-9_-.]", entry); } if (entry == NULL) { /* Skip */ } else if (rc != 1) { crm_warn("Could not parse (%d %d): %s", last, lpc, line + last); } else if (safe_str_neq(entry, "on") && safe_str_neq(entry, "off")) { crm_trace("Adding '%s'", entry); *output = g_list_append(*output, entry); entry = NULL; } free(entry); last = lpc + 1; } } } static GListPtr parse_host_list(const char *hosts) { int lpc = 0; int max = 0; int last = 0; GListPtr output = NULL; if (hosts == NULL) { return output; } max = strlen(hosts); for (lpc = 0; lpc <= max; lpc++) { if (hosts[lpc] == '\n' || hosts[lpc] == 0) { int len = lpc - last; if(len > 1) { char *line = strndup(hosts + last, len); line[len] = 0; /* Because it might be '\n' */ parse_host_line(line, len, &output); free(line); } last = lpc + 1; } } crm_trace("Parsed %d entries from '%s'", g_list_length(output), hosts); return output; } GHashTable *metadata_cache = NULL; void free_metadata_cache() { if (metadata_cache != NULL) { g_hash_table_destroy(metadata_cache); metadata_cache = NULL; } } static void init_metadata_cache() { if (metadata_cache == NULL) { metadata_cache = crm_str_table_new(); } } static xmlNode * get_agent_metadata(const char *agent) { xmlNode *xml = NULL; char *buffer = NULL; init_metadata_cache(); buffer = g_hash_table_lookup(metadata_cache, agent); if(safe_str_eq(agent, STONITH_WATCHDOG_AGENT)) { return NULL; } else if(buffer == NULL) { stonith_t *st = stonith_api_new(); int rc = st->cmds->metadata(st, st_opt_sync_call, agent, NULL, &buffer, 10); stonith_api_delete(st); if (rc || !buffer) { crm_err("Could not retrieve metadata for fencing agent %s", agent); return NULL; } g_hash_table_replace(metadata_cache, strdup(agent), buffer); } xml = string2xml(buffer); return xml; } static gboolean is_nodeid_required(xmlNode * xml) { xmlXPathObjectPtr xpath = NULL; if (stand_alone) { return FALSE; } if (!xml) { return FALSE; } xpath = xpath_search(xml, "//parameter[@name='nodeid']"); if (numXpathResults(xpath) <= 0) { freeXpathObject(xpath); return FALSE; } freeXpathObject(xpath); return TRUE; } #define MAX_ACTION_LEN 256 static char * add_action(char *actions, const char *action) { int offset = 0; if (actions == NULL) { actions = calloc(1, MAX_ACTION_LEN); } else { offset = strlen(actions); } if (offset > 0) { offset += snprintf(actions+offset, MAX_ACTION_LEN - offset, " "); } offset += snprintf(actions+offset, MAX_ACTION_LEN - offset, "%s", action); return actions; } static void read_action_metadata(stonith_device_t *device) { xmlXPathObjectPtr xpath = NULL; int max = 0; int lpc = 0; if (device->agent_metadata == NULL) { return; } xpath = xpath_search(device->agent_metadata, "//action"); max = numXpathResults(xpath); if (max <= 0) { freeXpathObject(xpath); return; } for (lpc = 0; lpc < max; lpc++) { const char *on_target = NULL; const char *action = NULL; xmlNode *match = getXpathResult(xpath, lpc); CRM_LOG_ASSERT(match != NULL); if(match == NULL) { continue; }; on_target = crm_element_value(match, "on_target"); action = crm_element_value(match, "name"); if(safe_str_eq(action, "list")) { set_bit(device->flags, st_device_supports_list); } else if(safe_str_eq(action, "status")) { set_bit(device->flags, st_device_supports_status); } else if(safe_str_eq(action, "reboot")) { set_bit(device->flags, st_device_supports_reboot); } else if (safe_str_eq(action, "on")) { /* "automatic" means the cluster will unfence node when it joins */ const char *automatic = crm_element_value(match, "automatic"); /* "required" is a deprecated synonym for "automatic" */ const char *required = crm_element_value(match, "required"); if (crm_is_true(automatic) || crm_is_true(required)) { device->automatic_unfencing = TRUE; } } if (action && crm_is_true(on_target)) { device->on_target_actions = add_action(device->on_target_actions, action); } } freeXpathObject(xpath); } /*! * \internal * \brief Set a pcmk_*_action parameter if not already set * * \param[in,out] params Device parameters * \param[in] action Name of action * \param[in] value Value to use if action is not already set */ static void map_action(GHashTable *params, const char *action, const char *value) { char *key = crm_strdup_printf("pcmk_%s_action", action); if (g_hash_table_lookup(params, key)) { crm_warn("Ignoring %s='%s', see %s instead", STONITH_ATTR_ACTION_OP, value, key); free(key); } else { crm_warn("Mapping %s='%s' to %s='%s'", STONITH_ATTR_ACTION_OP, value, key, value); g_hash_table_insert(params, key, strdup(value)); } } /*! * \internal * \brief Create device parameter table from XML * * \param[in] name Device name (used for logging only) * \param[in,out] params Device parameters */ static GHashTable * xml2device_params(const char *name, xmlNode *dev) { GHashTable *params = xml2list(dev); const char *value; /* Action should never be specified in the device configuration, * but we support it for users who are familiar with other software * that worked that way. */ value = g_hash_table_lookup(params, STONITH_ATTR_ACTION_OP); if (value != NULL) { crm_warn("%s has '%s' parameter, which should never be specified in configuration", name, STONITH_ATTR_ACTION_OP); if (*value == '\0') { crm_warn("Ignoring empty '%s' parameter", STONITH_ATTR_ACTION_OP); } else if (strcmp(value, "reboot") == 0) { crm_warn("Ignoring %s='reboot' (see stonith-action cluster property instead)", STONITH_ATTR_ACTION_OP); } else if (strcmp(value, "off") == 0) { map_action(params, "reboot", value); } else { map_action(params, "off", value); map_action(params, "reboot", value); } g_hash_table_remove(params, STONITH_ATTR_ACTION_OP); } return params; } static stonith_device_t * build_device_from_xml(xmlNode * msg) { const char *value = NULL; xmlNode *dev = get_xpath_object("//" F_STONITH_DEVICE, msg, LOG_ERR); stonith_device_t *device = NULL; device = calloc(1, sizeof(stonith_device_t)); device->id = crm_element_value_copy(dev, XML_ATTR_ID); device->agent = crm_element_value_copy(dev, "agent"); device->namespace = crm_element_value_copy(dev, "namespace"); device->params = xml2device_params(device->id, dev); value = g_hash_table_lookup(device->params, STONITH_ATTR_HOSTLIST); if (value) { device->targets = parse_host_list(value); } value = g_hash_table_lookup(device->params, STONITH_ATTR_HOSTMAP); device->aliases = build_port_aliases(value, &(device->targets)); device->agent_metadata = get_agent_metadata(device->agent); read_action_metadata(device); value = g_hash_table_lookup(device->params, "nodeid"); if (!value) { device->include_nodeid = is_nodeid_required(device->agent_metadata); } value = crm_element_value(dev, "rsc_provides"); if (safe_str_eq(value, "unfencing")) { device->automatic_unfencing = TRUE; } if (is_action_required("on", device)) { crm_info("The fencing device '%s' requires unfencing", device->id); } if (device->on_target_actions) { crm_info("The fencing device '%s' requires actions (%s) to be executed on the target node", device->id, device->on_target_actions); } device->work = mainloop_add_trigger(G_PRIORITY_HIGH, stonith_device_dispatch, device); /* TODO: Hook up priority */ return device; } static const char * target_list_type(stonith_device_t * dev) { const char *check_type = NULL; check_type = g_hash_table_lookup(dev->params, STONITH_ATTR_HOSTCHECK); if (check_type == NULL) { if (g_hash_table_lookup(dev->params, STONITH_ATTR_HOSTLIST)) { check_type = "static-list"; } else if (g_hash_table_lookup(dev->params, STONITH_ATTR_HOSTMAP)) { check_type = "static-list"; } else if(is_set(dev->flags, st_device_supports_list)){ check_type = "dynamic-list"; } else if(is_set(dev->flags, st_device_supports_status)){ check_type = "status"; } else { check_type = "none"; } } return check_type; } void schedule_internal_command(const char *origin, stonith_device_t * device, const char *action, const char *victim, int timeout, void *internal_user_data, void (*done_cb) (GPid pid, int rc, const char *output, gpointer user_data)) { async_command_t *cmd = NULL; cmd = calloc(1, sizeof(async_command_t)); cmd->id = -1; cmd->default_timeout = timeout ? timeout : 60; cmd->timeout = cmd->default_timeout; cmd->action = strdup(action); cmd->victim = victim ? strdup(victim) : NULL; cmd->device = strdup(device->id); cmd->origin = strdup(origin); cmd->client = strdup(crm_system_name); cmd->client_name = strdup(crm_system_name); cmd->internal_user_data = internal_user_data; cmd->done_cb = done_cb; /* cmd, not internal_user_data, is passed to 'done_cb' as the userdata */ schedule_stonith_command(cmd, device); } gboolean string_in_list(GListPtr list, const char *item) { int lpc = 0; int max = g_list_length(list); for (lpc = 0; lpc < max; lpc++) { const char *value = g_list_nth_data(list, lpc); if (safe_str_eq(item, value)) { return TRUE; } else { crm_trace("%d: '%s' != '%s'", lpc, item, value); } } return FALSE; } static void status_search_cb(GPid pid, int rc, const char *output, gpointer user_data) { async_command_t *cmd = user_data; struct device_search_s *search = cmd->internal_user_data; stonith_device_t *dev = cmd->device ? g_hash_table_lookup(device_list, cmd->device) : NULL; gboolean can = FALSE; free_async_command(cmd); if (!dev) { search_devices_record_result(search, NULL, FALSE); return; } mainloop_set_trigger(dev->work); if (rc == 1 /* unknown */ ) { crm_trace("Host %s is not known by %s", search->host, dev->id); } else if (rc == 0 /* active */ || rc == 2 /* inactive */ ) { crm_trace("Host %s is known by %s", search->host, dev->id); can = TRUE; } else { crm_notice("Unknown result when testing if %s can fence %s: rc=%d", dev->id, search->host, rc); } search_devices_record_result(search, dev->id, can); } static void dynamic_list_search_cb(GPid pid, int rc, const char *output, gpointer user_data) { async_command_t *cmd = user_data; struct device_search_s *search = cmd->internal_user_data; stonith_device_t *dev = cmd->device ? g_hash_table_lookup(device_list, cmd->device) : NULL; gboolean can_fence = FALSE; free_async_command(cmd); /* Host/alias must be in the list output to be eligible to be fenced * * Will cause problems if down'd nodes aren't listed or (for virtual nodes) * if the guest is still listed despite being moved to another machine */ if (!dev) { search_devices_record_result(search, NULL, FALSE); return; } mainloop_set_trigger(dev->work); /* If we successfully got the targets earlier, don't disable. */ if (rc != 0 && !dev->targets) { crm_notice("Disabling port list queries for %s (%d): %s", dev->id, rc, output); /* Fall back to status */ g_hash_table_replace(dev->params, strdup(STONITH_ATTR_HOSTCHECK), strdup("status")); g_list_free_full(dev->targets, free); dev->targets = NULL; } else if (!rc) { crm_info("Refreshing port list for %s", dev->id); g_list_free_full(dev->targets, free); dev->targets = parse_host_list(output); dev->targets_age = time(NULL); } if (dev->targets) { const char *alias = g_hash_table_lookup(dev->aliases, search->host); if (!alias) { alias = search->host; } if (string_in_list(dev->targets, alias)) { can_fence = TRUE; } } search_devices_record_result(search, dev->id, can_fence); } /*! * \internal * \brief Returns true if any key in first is not in second or second has a different value for key */ static int device_params_diff(GHashTable *first, GHashTable *second) { char *key = NULL; char *value = NULL; GHashTableIter gIter; g_hash_table_iter_init(&gIter, first); while (g_hash_table_iter_next(&gIter, (void **)&key, (void **)&value)) { if(strstr(key, "CRM_meta") == key) { continue; } else if(strcmp(key, "crm_feature_set") == 0) { continue; } else { char *other_value = g_hash_table_lookup(second, key); if (!other_value || safe_str_neq(other_value, value)) { crm_trace("Different value for %s: %s != %s", key, other_value, value); return 1; } } } return 0; } /*! * \internal * \brief Checks to see if an identical device already exists in the device_list */ static stonith_device_t * device_has_duplicate(stonith_device_t * device) { stonith_device_t *dup = g_hash_table_lookup(device_list, device->id); if (!dup) { crm_trace("No match for %s", device->id); return NULL; } else if (safe_str_neq(dup->agent, device->agent)) { crm_trace("Different agent: %s != %s", dup->agent, device->agent); return NULL; } /* Use calculate_operation_digest() here? */ if (device_params_diff(device->params, dup->params) || device_params_diff(dup->params, device->params)) { return NULL; } crm_trace("Match"); return dup; } int stonith_device_register(xmlNode * msg, const char **desc, gboolean from_cib) { stonith_device_t *dup = NULL; stonith_device_t *device = build_device_from_xml(msg); dup = device_has_duplicate(device); if (dup) { crm_debug("Device '%s' already existed in device list (%d active devices)", device->id, g_hash_table_size(device_list)); free_device(device); device = dup; } else { stonith_device_t *old = g_hash_table_lookup(device_list, device->id); if (from_cib && old && old->api_registered) { /* If the cib is writing over an entry that is shared with a stonith client, * copy any pending ops that currently exist on the old entry to the new one. * Otherwise the pending ops will be reported as failures */ crm_info("Overwriting an existing entry for %s from the cib", device->id); device->pending_ops = old->pending_ops; device->api_registered = TRUE; old->pending_ops = NULL; if (device->pending_ops) { mainloop_set_trigger(device->work); } } g_hash_table_replace(device_list, device->id, device); crm_notice("Added '%s' to the device list (%d active devices)", device->id, g_hash_table_size(device_list)); } if (desc) { *desc = device->id; } if (from_cib) { device->cib_registered = TRUE; } else { device->api_registered = TRUE; } return pcmk_ok; } int stonith_device_remove(const char *id, gboolean from_cib) { stonith_device_t *device = g_hash_table_lookup(device_list, id); if (!device) { crm_info("Device '%s' not found (%d active devices)", id, g_hash_table_size(device_list)); return pcmk_ok; } if (from_cib) { device->cib_registered = FALSE; } else { device->verified = FALSE; device->api_registered = FALSE; } if (!device->cib_registered && !device->api_registered) { g_hash_table_remove(device_list, id); crm_info("Removed '%s' from the device list (%d active devices)", id, g_hash_table_size(device_list)); } else { crm_trace("Not removing '%s' from the device list (%d active devices) " "- still %s%s_registered", id, g_hash_table_size(device_list), device->cib_registered?"cib":"", device->api_registered?"api":""); } return pcmk_ok; } /*! * \internal * \brief Return the number of stonith levels registered for a node * * \param[in] tp Node's topology table entry * * \return Number of non-NULL levels in topology entry * \note This function is used only for log messages. */ static int count_active_levels(stonith_topology_t * tp) { int lpc = 0; int count = 0; for (lpc = 0; lpc < ST_LEVEL_MAX; lpc++) { if (tp->levels[lpc] != NULL) { count++; } } return count; } static void free_topology_entry(gpointer data) { stonith_topology_t *tp = data; int lpc = 0; for (lpc = 0; lpc < ST_LEVEL_MAX; lpc++) { if (tp->levels[lpc] != NULL) { g_list_free_full(tp->levels[lpc], free); } } free(tp->target); free(tp->target_value); free(tp->target_pattern); free(tp->target_attribute); free(tp); } void free_topology_list() { if (topology != NULL) { g_hash_table_destroy(topology); topology = NULL; } } void init_topology_list() { if (topology == NULL) { topology = g_hash_table_new_full(crm_str_hash, g_str_equal, NULL, free_topology_entry); } } char *stonith_level_key(xmlNode *level, int mode) { if(mode == -1) { mode = stonith_level_kind(level); } switch(mode) { case 0: return crm_element_value_copy(level, XML_ATTR_STONITH_TARGET); case 1: return crm_element_value_copy(level, XML_ATTR_STONITH_TARGET_PATTERN); case 2: { const char *name = crm_element_value(level, XML_ATTR_STONITH_TARGET_ATTRIBUTE); const char *value = crm_element_value(level, XML_ATTR_STONITH_TARGET_VALUE); if(name && value) { return crm_strdup_printf("%s=%s", name, value); } } default: return crm_strdup_printf("Unknown-%d-%s", mode, ID(level)); } } int stonith_level_kind(xmlNode * level) { int mode = 0; const char *target = crm_element_value(level, XML_ATTR_STONITH_TARGET); if(target == NULL) { mode++; target = crm_element_value(level, XML_ATTR_STONITH_TARGET_PATTERN); } if(stand_alone == FALSE && target == NULL) { mode++; if(crm_element_value(level, XML_ATTR_STONITH_TARGET_ATTRIBUTE) == NULL) { mode++; } else if(crm_element_value(level, XML_ATTR_STONITH_TARGET_VALUE) == NULL) { mode++; } } return mode; } static stonith_key_value_t * parse_device_list(const char *devices) { int lpc = 0; int max = 0; int last = 0; stonith_key_value_t *output = NULL; if (devices == NULL) { return output; } max = strlen(devices); for (lpc = 0; lpc <= max; lpc++) { if (devices[lpc] == ',' || devices[lpc] == 0) { char *line = strndup(devices + last, lpc - last); output = stonith_key_value_add(output, NULL, line); free(line); last = lpc + 1; } } return output; } /*! * \internal * \brief Register a STONITH level for a target * * Given an XML request specifying the target name, level index, and device IDs * for the level, this will create an entry for the target in the global topology * table if one does not already exist, then append the specified device IDs to * the entry's device list for the specified level. * * \param[in] msg XML request for STONITH level registration * \param[out] desc If not NULL, will be set to string representation ("TARGET[LEVEL]") * * \return pcmk_ok on success, -EINVAL if XML does not specify valid level index */ int stonith_level_register(xmlNode *msg, char **desc) { int id = 0; xmlNode *level; int mode; char *target; stonith_topology_t *tp; stonith_key_value_t *dIter = NULL; stonith_key_value_t *devices = NULL; /* Allow the XML here to point to the level tag directly, or wrapped in * another tag. If directly, don't search by xpath, because it might give * multiple hits (e.g. if the XML is the CIB). */ if (safe_str_eq(TYPE(msg), XML_TAG_FENCING_LEVEL)) { level = msg; } else { level = get_xpath_object("//" XML_TAG_FENCING_LEVEL, msg, LOG_ERR); } CRM_CHECK(level != NULL, return -EINVAL); mode = stonith_level_kind(level); target = stonith_level_key(level, mode); crm_element_value_int(level, XML_ATTR_STONITH_INDEX, &id); if (desc) { *desc = crm_strdup_printf("%s[%d]", target, id); } /* Sanity-check arguments */ if (mode >= 3 || (id <= 0) || (id >= ST_LEVEL_MAX)) { crm_trace("Could not add %s[%d] (%d) to the topology (%d active entries)", target, id, mode, g_hash_table_size(topology)); free(target); crm_log_xml_err(level, "Bad topology"); return -EINVAL; } /* Find or create topology table entry */ tp = g_hash_table_lookup(topology, target); if (tp == NULL) { tp = calloc(1, sizeof(stonith_topology_t)); tp->kind = mode; tp->target = target; tp->target_value = crm_element_value_copy(level, XML_ATTR_STONITH_TARGET_VALUE); tp->target_pattern = crm_element_value_copy(level, XML_ATTR_STONITH_TARGET_PATTERN); tp->target_attribute = crm_element_value_copy(level, XML_ATTR_STONITH_TARGET_ATTRIBUTE); g_hash_table_replace(topology, tp->target, tp); crm_trace("Added %s (%d) to the topology (%d active entries)", target, mode, g_hash_table_size(topology)); } else { free(target); } if (tp->levels[id] != NULL) { crm_info("Adding to the existing %s[%d] topology entry", tp->target, id); } devices = parse_device_list(crm_element_value(level, XML_ATTR_STONITH_DEVICES)); for (dIter = devices; dIter; dIter = dIter->next) { const char *device = dIter->value; crm_trace("Adding device '%s' for %s[%d]", device, tp->target, id); tp->levels[id] = g_list_append(tp->levels[id], strdup(device)); } stonith_key_value_freeall(devices, 1, 1); crm_info("Target %s has %d active fencing levels", tp->target, count_active_levels(tp)); return pcmk_ok; } int stonith_level_remove(xmlNode *msg, char **desc) { int id = 0; stonith_topology_t *tp; char *target; /* Unlike additions, removal requests should always have one level tag */ xmlNode *level = get_xpath_object("//" XML_TAG_FENCING_LEVEL, msg, LOG_ERR); CRM_CHECK(level != NULL, return -EINVAL); target = stonith_level_key(level, -1); crm_element_value_int(level, XML_ATTR_STONITH_INDEX, &id); if (desc) { *desc = crm_strdup_printf("%s[%d]", target, id); } /* Sanity-check arguments */ if (id >= ST_LEVEL_MAX) { free(target); return -EINVAL; } tp = g_hash_table_lookup(topology, target); if (tp == NULL) { crm_info("Topology for %s not found (%d active entries)", target, g_hash_table_size(topology)); } else if (id == 0 && g_hash_table_remove(topology, target)) { crm_info("Removed all %s related entries from the topology (%d active entries)", target, g_hash_table_size(topology)); } else if (id > 0 && tp->levels[id] != NULL) { g_list_free_full(tp->levels[id], free); tp->levels[id] = NULL; crm_info("Removed level '%d' from topology for %s (%d active levels remaining)", id, target, count_active_levels(tp)); } free(target); return pcmk_ok; } static int stonith_device_action(xmlNode * msg, char **output) { int rc = pcmk_ok; xmlNode *dev = get_xpath_object("//" F_STONITH_DEVICE, msg, LOG_ERR); const char *id = crm_element_value(dev, F_STONITH_DEVICE); async_command_t *cmd = NULL; stonith_device_t *device = NULL; if (id) { crm_trace("Looking for '%s'", id); device = g_hash_table_lookup(device_list, id); } if (device && device->api_registered == FALSE) { rc = -ENODEV; } else if (device) { cmd = create_async_command(msg); if (cmd == NULL) { return -EPROTO; } schedule_stonith_command(cmd, device); rc = -EINPROGRESS; } else { crm_info("Device %s not found", id ? id : ""); rc = -ENODEV; } return rc; } static void search_devices_record_result(struct device_search_s *search, const char *device, gboolean can_fence) { search->replies_received++; if (can_fence && device) { search->capable = g_list_append(search->capable, strdup(device)); } if (search->replies_needed == search->replies_received) { crm_debug("Finished Search. %d devices can perform action (%s) on node %s", g_list_length(search->capable), search->action ? search->action : "", search->host ? search->host : ""); search->callback(search->capable, search->user_data); free(search->host); free(search->action); free(search); } } /*! * \internal * \brief Check whether the local host is allowed to execute a fencing action * * \param[in] device Fence device to check * \param[in] action Fence action to check * \param[in] target Hostname of fence target * \param[in] allow_suicide Whether self-fencing is allowed for this operation * * \return TRUE if local host is allowed to execute action, FALSE otherwise */ static gboolean localhost_is_eligible(const stonith_device_t *device, const char *action, const char *target, gboolean allow_suicide) { gboolean localhost_is_target = safe_str_eq(target, stonith_our_uname); if (device && action && device->on_target_actions && strstr(device->on_target_actions, action)) { if (!localhost_is_target) { crm_trace("%s operation with %s can only be executed for localhost not %s", action, device->id, target); return FALSE; } } else if (localhost_is_target && !allow_suicide) { crm_trace("%s operation does not support self-fencing", action); return FALSE; } return TRUE; } static void can_fence_host_with_device(stonith_device_t * dev, struct device_search_s *search) { gboolean can = FALSE; const char *check_type = NULL; const char *host = search->host; const char *alias = NULL; CRM_LOG_ASSERT(dev != NULL); if (dev == NULL) { goto search_report_results; } else if (host == NULL) { can = TRUE; goto search_report_results; } /* Short-circuit query if this host is not allowed to perform the action */ if (safe_str_eq(search->action, "reboot")) { /* A "reboot" *might* get remapped to "off" then "on", so short-circuit * only if all three are disallowed. If only one or two are disallowed, * we'll report that with the results. We never allow suicide for * remapped "on" operations because the host is off at that point. */ if (!localhost_is_eligible(dev, "reboot", host, search->allow_suicide) && !localhost_is_eligible(dev, "off", host, search->allow_suicide) && !localhost_is_eligible(dev, "on", host, FALSE)) { goto search_report_results; } } else if (!localhost_is_eligible(dev, search->action, host, search->allow_suicide)) { goto search_report_results; } alias = g_hash_table_lookup(dev->aliases, host); if (alias == NULL) { alias = host; } check_type = target_list_type(dev); if (safe_str_eq(check_type, "none")) { can = TRUE; } else if (safe_str_eq(check_type, "static-list")) { /* Presence in the hostmap is sufficient * Only use if all hosts on which the device can be active can always fence all listed hosts */ if (string_in_list(dev->targets, host)) { can = TRUE; } else if (g_hash_table_lookup(dev->params, STONITH_ATTR_HOSTMAP) && g_hash_table_lookup(dev->aliases, host)) { can = TRUE; } } else if (safe_str_eq(check_type, "dynamic-list")) { time_t now = time(NULL); if (dev->targets == NULL || dev->targets_age + 60 < now) { crm_trace("Running %s command to see if %s can fence %s (%s)", check_type, dev->id, search->host, search->action); schedule_internal_command(__FUNCTION__, dev, "list", NULL, search->per_device_timeout, search, dynamic_list_search_cb); /* we'll respond to this search request async in the cb */ return; } if (string_in_list(dev->targets, alias)) { can = TRUE; } } else if (safe_str_eq(check_type, "status")) { crm_trace("Running %s command to see if %s can fence %s (%s)", check_type, dev->id, search->host, search->action); schedule_internal_command(__FUNCTION__, dev, "status", search->host, search->per_device_timeout, search, status_search_cb); /* we'll respond to this search request async in the cb */ return; } else { crm_err("Unknown check type: %s", check_type); } if (safe_str_eq(host, alias)) { crm_notice("%s can%s fence (%s) %s: %s", dev->id, can ? "" : " not", search->action, host, check_type); } else { crm_notice("%s can%s fence (%s) %s (aka. '%s'): %s", dev->id, can ? "" : " not", search->action, host, alias, check_type); } search_report_results: search_devices_record_result(search, dev ? dev->id : NULL, can); } static void search_devices(gpointer key, gpointer value, gpointer user_data) { stonith_device_t *dev = value; struct device_search_s *search = user_data; can_fence_host_with_device(dev, search); } #define DEFAULT_QUERY_TIMEOUT 20 static void get_capable_devices(const char *host, const char *action, int timeout, bool suicide, void *user_data, void (*callback) (GList * devices, void *user_data)) { struct device_search_s *search; int per_device_timeout = DEFAULT_QUERY_TIMEOUT; int devices_needing_async_query = 0; char *key = NULL; const char *check_type = NULL; GHashTableIter gIter; stonith_device_t *device = NULL; if (!g_hash_table_size(device_list)) { callback(NULL, user_data); return; } search = calloc(1, sizeof(struct device_search_s)); if (!search) { callback(NULL, user_data); return; } g_hash_table_iter_init(&gIter, device_list); while (g_hash_table_iter_next(&gIter, (void **)&key, (void **)&device)) { check_type = target_list_type(device); if (safe_str_eq(check_type, "status") || safe_str_eq(check_type, "dynamic-list")) { devices_needing_async_query++; } } /* If we have devices that require an async event in order to know what * nodes they can fence, we have to give the events a timeout. The total * query timeout is divided among those events. */ if (devices_needing_async_query) { per_device_timeout = timeout / devices_needing_async_query; if (!per_device_timeout) { crm_err("STONITH timeout %ds is too low; using %ds, but consider raising to at least %ds", timeout, DEFAULT_QUERY_TIMEOUT, DEFAULT_QUERY_TIMEOUT * devices_needing_async_query); per_device_timeout = DEFAULT_QUERY_TIMEOUT; } else if (per_device_timeout < DEFAULT_QUERY_TIMEOUT) { crm_notice("STONITH timeout %ds is low for the current configuration;" " consider raising to at least %ds", timeout, DEFAULT_QUERY_TIMEOUT * devices_needing_async_query); } } search->host = host ? strdup(host) : NULL; search->action = action ? strdup(action) : NULL; search->per_device_timeout = per_device_timeout; /* We are guaranteed this many replies. Even if a device gets * unregistered some how during the async search, we will get * the correct number of replies. */ search->replies_needed = g_hash_table_size(device_list); search->allow_suicide = suicide; search->callback = callback; search->user_data = user_data; /* kick off the search */ crm_debug("Searching through %d devices to see what is capable of action (%s) for target %s", search->replies_needed, search->action ? search->action : "", search->host ? search->host : ""); g_hash_table_foreach(device_list, search_devices, search); } struct st_query_data { xmlNode *reply; char *remote_peer; char *client_id; char *target; char *action; int call_options; }; /*! * \internal * \brief Add action-specific attributes to query reply XML * * \param[in,out] xml XML to add attributes to * \param[in] action Fence action * \param[in] device Fence device */ static void add_action_specific_attributes(xmlNode *xml, const char *action, stonith_device_t *device) { int action_specific_timeout; int delay_max; int delay_base; CRM_CHECK(xml && action && device, return); if (is_action_required(action, device)) { crm_trace("Action %s is required on %s", action, device->id); crm_xml_add_int(xml, F_STONITH_DEVICE_REQUIRED, 1); } action_specific_timeout = get_action_timeout(device, action, 0); if (action_specific_timeout) { crm_trace("Action %s has timeout %dms on %s", action, action_specific_timeout, device->id); crm_xml_add_int(xml, F_STONITH_ACTION_TIMEOUT, action_specific_timeout); } delay_max = get_action_delay_max(device, action); if (delay_max > 0) { crm_trace("Action %s has maximum random delay %dms on %s", action, delay_max, device->id); crm_xml_add_int(xml, F_STONITH_DELAY_MAX, delay_max / 1000); } delay_base = get_action_delay_base(device, action); if (delay_base > 0) { crm_xml_add_int(xml, F_STONITH_DELAY_BASE, delay_base / 1000); } if ((delay_max > 0) && (delay_base == 0)) { crm_trace("Action %s has maximum random delay %dms on %s", action, delay_max, device->id); } else if ((delay_max == 0) && (delay_base > 0)) { crm_trace("Action %s has a static delay of %dms on %s", action, delay_base, device->id); } else if ((delay_max > 0) && (delay_base > 0)) { crm_trace("Action %s has a minimum delay of %dms and a randomly chosen " "maximum delay of %dms on %s", action, delay_base, delay_max, device->id); } } /*! * \internal * \brief Add "disallowed" attribute to query reply XML if appropriate * * \param[in,out] xml XML to add attribute to * \param[in] action Fence action * \param[in] device Fence device * \param[in] target Fence target * \param[in] allow_suicide Whether self-fencing is allowed */ static void add_disallowed(xmlNode *xml, const char *action, stonith_device_t *device, const char *target, gboolean allow_suicide) { if (!localhost_is_eligible(device, action, target, allow_suicide)) { crm_trace("Action %s on %s is disallowed for local host", action, device->id); crm_xml_add(xml, F_STONITH_ACTION_DISALLOWED, XML_BOOLEAN_TRUE); } } /*! * \internal * \brief Add child element with action-specific values to query reply XML * * \param[in,out] xml XML to add attribute to * \param[in] action Fence action * \param[in] device Fence device * \param[in] target Fence target * \param[in] allow_suicide Whether self-fencing is allowed */ static void add_action_reply(xmlNode *xml, const char *action, stonith_device_t *device, const char *target, gboolean allow_suicide) { xmlNode *child = create_xml_node(xml, F_STONITH_ACTION); crm_xml_add(child, XML_ATTR_ID, action); add_action_specific_attributes(child, action, device); add_disallowed(child, action, device, target, allow_suicide); } static void stonith_query_capable_device_cb(GList * devices, void *user_data) { struct st_query_data *query = user_data; int available_devices = 0; xmlNode *dev = NULL; xmlNode *list = NULL; GListPtr lpc = NULL; /* Pack the results into XML */ list = create_xml_node(NULL, __FUNCTION__); crm_xml_add(list, F_STONITH_TARGET, query->target); for (lpc = devices; lpc != NULL; lpc = lpc->next) { stonith_device_t *device = g_hash_table_lookup(device_list, lpc->data); const char *action = query->action; if (!device) { /* It is possible the device got unregistered while * determining who can fence the target */ continue; } available_devices++; dev = create_xml_node(list, F_STONITH_DEVICE); crm_xml_add(dev, XML_ATTR_ID, device->id); crm_xml_add(dev, "namespace", device->namespace); crm_xml_add(dev, "agent", device->agent); crm_xml_add_int(dev, F_STONITH_DEVICE_VERIFIED, device->verified); /* If the originating fencer wants to reboot the node, and we have a * capable device that doesn't support "reboot", remap to "off" instead. */ if (is_not_set(device->flags, st_device_supports_reboot) && safe_str_eq(query->action, "reboot")) { crm_trace("%s doesn't support reboot, using values for off instead", device->id); action = "off"; } /* Add action-specific values if available */ add_action_specific_attributes(dev, action, device); if (safe_str_eq(query->action, "reboot")) { /* A "reboot" *might* get remapped to "off" then "on", so after * sending the "reboot"-specific values in the main element, we add * sub-elements for "off" and "on" values. * * We short-circuited earlier if "reboot", "off" and "on" are all * disallowed for the local host. However if only one or two are * disallowed, we send back the results and mark which ones are * disallowed. If "reboot" is disallowed, this might cause problems * with older fencer versions, which won't check for it. Older * versions will ignore "off" and "on", so they are not a problem. */ add_disallowed(dev, action, device, query->target, is_set(query->call_options, st_opt_allow_suicide)); add_action_reply(dev, "off", device, query->target, is_set(query->call_options, st_opt_allow_suicide)); add_action_reply(dev, "on", device, query->target, FALSE); } /* A query without a target wants device parameters */ if (query->target == NULL) { xmlNode *attrs = create_xml_node(dev, XML_TAG_ATTRS); g_hash_table_foreach(device->params, hash2field, attrs); } } crm_xml_add_int(list, F_STONITH_AVAILABLE_DEVICES, available_devices); if (query->target) { crm_debug("Found %d matching devices for '%s'", available_devices, query->target); } else { crm_debug("%d devices installed", available_devices); } if (list != NULL) { crm_log_xml_trace(list, "Add query results"); add_message_xml(query->reply, F_STONITH_CALLDATA, list); } stonith_send_reply(query->reply, query->call_options, query->remote_peer, query->client_id); free_xml(query->reply); free(query->remote_peer); free(query->client_id); free(query->target); free(query->action); free(query); free_xml(list); g_list_free_full(devices, free); } static void stonith_query(xmlNode * msg, const char *remote_peer, const char *client_id, int call_options) { struct st_query_data *query = NULL; const char *action = NULL; const char *target = NULL; int timeout = 0; xmlNode *dev = get_xpath_object("//@" F_STONITH_ACTION, msg, LOG_TRACE); crm_element_value_int(msg, F_STONITH_TIMEOUT, &timeout); if (dev) { const char *device = crm_element_value(dev, F_STONITH_DEVICE); target = crm_element_value(dev, F_STONITH_TARGET); action = crm_element_value(dev, F_STONITH_ACTION); if (device && safe_str_eq(device, "manual_ack")) { /* No query or reply necessary */ return; } } crm_log_xml_debug(msg, "Query"); query = calloc(1, sizeof(struct st_query_data)); query->reply = stonith_construct_reply(msg, NULL, NULL, pcmk_ok); query->remote_peer = remote_peer ? strdup(remote_peer) : NULL; query->client_id = client_id ? strdup(client_id) : NULL; query->target = target ? strdup(target) : NULL; query->action = action ? strdup(action) : NULL; query->call_options = call_options; get_capable_devices(target, action, timeout, is_set(call_options, st_opt_allow_suicide), query, stonith_query_capable_device_cb); } #define ST_LOG_OUTPUT_MAX 512 static void log_operation(async_command_t * cmd, int rc, int pid, const char *next, const char *output) { if (rc == 0) { next = NULL; } if (cmd->victim != NULL) { do_crm_log(rc == 0 ? LOG_NOTICE : LOG_ERR, "Operation '%s' [%d] (call %d from %s) for host '%s' with device '%s' returned: %d (%s)%s%s", cmd->action, pid, cmd->id, cmd->client_name, cmd->victim, cmd->device, rc, pcmk_strerror(rc), next ? ". Trying: " : "", next ? next : ""); } else { do_crm_log_unlikely(rc == 0 ? LOG_DEBUG : LOG_NOTICE, "Operation '%s' [%d] for device '%s' returned: %d (%s)%s%s", cmd->action, pid, cmd->device, rc, pcmk_strerror(rc), next ? ". Trying: " : "", next ? next : ""); } if (output) { /* Logging the whole string confuses syslog when the string is xml */ char *prefix = crm_strdup_printf("%s:%d", cmd->device, pid); crm_log_output(rc == 0 ? LOG_DEBUG : LOG_WARNING, prefix, output); free(prefix); } } static void stonith_send_async_reply(async_command_t * cmd, const char *output, int rc, GPid pid) { xmlNode *reply = NULL; gboolean bcast = FALSE; reply = stonith_construct_async_reply(cmd, output, NULL, rc); if (safe_str_eq(cmd->action, "metadata")) { /* Too verbose to log */ crm_trace("Metadata query for %s", cmd->device); output = NULL; } else if (crm_str_eq(cmd->action, "monitor", TRUE) || crm_str_eq(cmd->action, "list", TRUE) || crm_str_eq(cmd->action, "status", TRUE)) { crm_trace("Never broadcast %s replies", cmd->action); } else if (!stand_alone && safe_str_eq(cmd->origin, cmd->victim) && safe_str_neq(cmd->action, "on")) { crm_trace("Broadcast %s reply for %s", cmd->action, cmd->victim); crm_xml_add(reply, F_SUBTYPE, "broadcast"); bcast = TRUE; } log_operation(cmd, rc, pid, NULL, output); crm_log_xml_trace(reply, "Reply"); if (bcast) { crm_xml_add(reply, F_STONITH_OPERATION, T_STONITH_NOTIFY); send_cluster_message(NULL, crm_msg_stonith_ng, reply, FALSE); } else if (cmd->origin) { crm_trace("Directed reply to %s", cmd->origin); send_cluster_message(crm_get_peer(0, cmd->origin), crm_msg_stonith_ng, reply, FALSE); } else { crm_trace("Directed local %ssync reply to %s", (cmd->options & st_opt_sync_call) ? "" : "a-", cmd->client_name); do_local_reply(reply, cmd->client, cmd->options & st_opt_sync_call, FALSE); } if (stand_alone) { /* Do notification with a clean data object */ xmlNode *notify_data = create_xml_node(NULL, T_STONITH_NOTIFY_FENCE); crm_xml_add_int(notify_data, F_STONITH_RC, rc); crm_xml_add(notify_data, F_STONITH_TARGET, cmd->victim); crm_xml_add(notify_data, F_STONITH_OPERATION, cmd->op); crm_xml_add(notify_data, F_STONITH_DELEGATE, "localhost"); crm_xml_add(notify_data, F_STONITH_DEVICE, cmd->device); crm_xml_add(notify_data, F_STONITH_REMOTE_OP_ID, cmd->remote_op_id); crm_xml_add(notify_data, F_STONITH_ORIGIN, cmd->client); do_stonith_notify(0, T_STONITH_NOTIFY_FENCE, rc, notify_data); + do_stonith_notify(0, T_STONITH_NOTIFY_HISTORY, 0, NULL); } free_xml(reply); } static void cancel_stonith_command(async_command_t * cmd) { stonith_device_t *device; CRM_CHECK(cmd != NULL, return); if (!cmd->device) { return; } device = g_hash_table_lookup(device_list, cmd->device); if (device) { crm_trace("Cancel scheduled %s on %s", cmd->action, device->id); device->pending_ops = g_list_remove(device->pending_ops, cmd); } } static void st_child_done(GPid pid, int rc, const char *output, gpointer user_data) { stonith_device_t *device = NULL; stonith_device_t *next_device = NULL; async_command_t *cmd = user_data; GListPtr gIter = NULL; GListPtr gIterNext = NULL; CRM_CHECK(cmd != NULL, return); cmd->active_on = NULL; /* The device is ready to do something else now */ device = g_hash_table_lookup(device_list, cmd->device); if (device) { if (rc == pcmk_ok && (safe_str_eq(cmd->action, "list") || safe_str_eq(cmd->action, "monitor") || safe_str_eq(cmd->action, "status"))) { device->verified = TRUE; } mainloop_set_trigger(device->work); } crm_debug("Operation '%s' on '%s' completed with rc=%d (%d remaining)", cmd->action, cmd->device, rc, g_list_length(cmd->device_next)); if (rc == 0) { GListPtr iter; /* see if there are any required devices left to execute for this op */ for (iter = cmd->device_next; iter != NULL; iter = iter->next) { next_device = g_hash_table_lookup(device_list, iter->data); if (next_device != NULL && is_action_required(cmd->action, next_device)) { cmd->device_next = iter->next; break; } next_device = NULL; } } else if (rc != 0 && cmd->device_next && (is_action_required(cmd->action, device) == FALSE)) { /* if this device didn't work out, see if there are any others we can try. * if the failed device was 'required', we can't pick another device. */ next_device = g_hash_table_lookup(device_list, cmd->device_next->data); cmd->device_next = cmd->device_next->next; } /* this operation requires more fencing, hooray! */ if (next_device) { log_operation(cmd, rc, pid, cmd->device, output); schedule_stonith_command(cmd, next_device); /* Prevent cmd from being freed */ cmd = NULL; goto done; } stonith_send_async_reply(cmd, output, rc, pid); if (rc != 0) { goto done; } /* Check to see if any operations are scheduled to do the exact * same thing that just completed. If so, rather than * performing the same fencing operation twice, return the result * of this operation for all pending commands it matches. */ for (gIter = cmd_list; gIter != NULL; gIter = gIterNext) { async_command_t *cmd_other = gIter->data; gIterNext = gIter->next; if (cmd == cmd_other) { continue; } /* A pending scheduled command matches the command that just finished if. * 1. The client connections are different. * 2. The node victim is the same. * 3. The fencing action is the same. * 4. The device scheduled to execute the action is the same. */ if (safe_str_eq(cmd->client, cmd_other->client) || safe_str_neq(cmd->victim, cmd_other->victim) || safe_str_neq(cmd->action, cmd_other->action) || safe_str_neq(cmd->device, cmd_other->device)) { continue; } /* Duplicate merging will do the right thing for either type of remapped * reboot. If the executing fencer remapped an unsupported reboot to * off, then cmd->action will be reboot and will be merged with any * other reboot requests. If the originating fencer remapped a * topology reboot to off then on, we will get here once with * cmd->action "off" and once with "on", and they will be merged * separately with similar requests. */ crm_notice ("Merging stonith action %s for node %s originating from client %s with identical stonith request from client %s", cmd_other->action, cmd_other->victim, cmd_other->client_name, cmd->client_name); cmd_list = g_list_remove_link(cmd_list, gIter); stonith_send_async_reply(cmd_other, output, rc, pid); cancel_stonith_command(cmd_other); free_async_command(cmd_other); g_list_free_1(gIter); } done: free_async_command(cmd); } static gint sort_device_priority(gconstpointer a, gconstpointer b) { const stonith_device_t *dev_a = a; const stonith_device_t *dev_b = b; if (dev_a->priority > dev_b->priority) { return -1; } else if (dev_a->priority < dev_b->priority) { return 1; } return 0; } static void stonith_fence_get_devices_cb(GList * devices, void *user_data) { async_command_t *cmd = user_data; stonith_device_t *device = NULL; crm_info("Found %d matching devices for '%s'", g_list_length(devices), cmd->victim); if (g_list_length(devices) > 0) { /* Order based on priority */ devices = g_list_sort(devices, sort_device_priority); device = g_hash_table_lookup(device_list, devices->data); if (device) { cmd->device_list = devices; cmd->device_next = devices->next; devices = NULL; /* list owned by cmd now */ } } /* we have a device, schedule it for fencing. */ if (device) { schedule_stonith_command(cmd, device); /* in progress */ return; } /* no device found! */ stonith_send_async_reply(cmd, NULL, -ENODEV, 0); free_async_command(cmd); g_list_free_full(devices, free); } static int stonith_fence(xmlNode * msg) { const char *device_id = NULL; stonith_device_t *device = NULL; async_command_t *cmd = create_async_command(msg); xmlNode *dev = get_xpath_object("//@" F_STONITH_TARGET, msg, LOG_ERR); if (cmd == NULL) { return -EPROTO; } device_id = crm_element_value(dev, F_STONITH_DEVICE); if (device_id) { device = g_hash_table_lookup(device_list, device_id); if (device == NULL) { crm_err("Requested device '%s' is not available", device_id); return -ENODEV; } schedule_stonith_command(cmd, device); } else { const char *host = crm_element_value(dev, F_STONITH_TARGET); char *nodename = NULL; if (cmd->options & st_opt_cs_nodeid) { int nodeid = crm_atoi(host, NULL); nodename = stonith_get_peer_name(nodeid); if (nodename) { host = nodename; } } /* If we get to here, then self-fencing is implicitly allowed */ get_capable_devices(host, cmd->action, cmd->default_timeout, TRUE, cmd, stonith_fence_get_devices_cb); free(nodename); } return -EINPROGRESS; } xmlNode * stonith_construct_reply(xmlNode * request, const char *output, xmlNode * data, int rc) { int lpc = 0; xmlNode *reply = NULL; const char *name = NULL; const char *value = NULL; const char *names[] = { F_STONITH_OPERATION, F_STONITH_CALLID, F_STONITH_CLIENTID, F_STONITH_CLIENTNAME, F_STONITH_REMOTE_OP_ID, F_STONITH_CALLOPTS }; crm_trace("Creating a basic reply"); reply = create_xml_node(NULL, T_STONITH_REPLY); crm_xml_add(reply, "st_origin", __FUNCTION__); crm_xml_add(reply, F_TYPE, T_STONITH_NG); crm_xml_add(reply, "st_output", output); crm_xml_add_int(reply, F_STONITH_RC, rc); CRM_CHECK(request != NULL, crm_warn("Can't create a sane reply"); return reply); for (lpc = 0; lpc < DIMOF(names); lpc++) { name = names[lpc]; value = crm_element_value(request, name); crm_xml_add(reply, name, value); } if (data != NULL) { crm_trace("Attaching reply output"); add_message_xml(reply, F_STONITH_CALLDATA, data); } return reply; } static xmlNode * stonith_construct_async_reply(async_command_t * cmd, const char *output, xmlNode * data, int rc) { xmlNode *reply = NULL; crm_trace("Creating a basic reply"); reply = create_xml_node(NULL, T_STONITH_REPLY); crm_xml_add(reply, "st_origin", __FUNCTION__); crm_xml_add(reply, F_TYPE, T_STONITH_NG); crm_xml_add(reply, F_STONITH_OPERATION, cmd->op); crm_xml_add(reply, F_STONITH_DEVICE, cmd->device); crm_xml_add(reply, F_STONITH_REMOTE_OP_ID, cmd->remote_op_id); crm_xml_add(reply, F_STONITH_CLIENTID, cmd->client); crm_xml_add(reply, F_STONITH_CLIENTNAME, cmd->client_name); crm_xml_add(reply, F_STONITH_TARGET, cmd->victim); crm_xml_add(reply, F_STONITH_ACTION, cmd->op); crm_xml_add(reply, F_STONITH_ORIGIN, cmd->origin); crm_xml_add_int(reply, F_STONITH_CALLID, cmd->id); crm_xml_add_int(reply, F_STONITH_CALLOPTS, cmd->options); crm_xml_add_int(reply, F_STONITH_RC, rc); crm_xml_add(reply, "st_output", output); if (data != NULL) { crm_info("Attaching reply output"); add_message_xml(reply, F_STONITH_CALLDATA, data); } return reply; } bool fencing_peer_active(crm_node_t *peer) { if (peer == NULL) { return FALSE; } else if (peer->uname == NULL) { return FALSE; } else if (is_set(peer->processes, crm_get_cluster_proc())) { return TRUE; } return FALSE; } /*! * \internal * \brief Determine if we need to use an alternate node to * fence the target. If so return that node's uname * * \retval NULL, no alternate host * \retval uname, uname of alternate host to use */ static const char * check_alternate_host(const char *target) { const char *alternate_host = NULL; crm_trace("Checking if we (%s) can fence %s", stonith_our_uname, target); if (find_topology_for_host(target) && safe_str_eq(target, stonith_our_uname)) { GHashTableIter gIter; crm_node_t *entry = NULL; g_hash_table_iter_init(&gIter, crm_peer_cache); while (g_hash_table_iter_next(&gIter, NULL, (void **)&entry)) { crm_trace("Checking for %s.%d != %s", entry->uname, entry->id, target); if (fencing_peer_active(entry) && safe_str_neq(entry->uname, target)) { alternate_host = entry->uname; break; } } if (alternate_host == NULL) { crm_err("No alternate host available to handle complex self fencing request"); g_hash_table_iter_init(&gIter, crm_peer_cache); while (g_hash_table_iter_next(&gIter, NULL, (void **)&entry)) { crm_notice("Peer[%d] %s", entry->id, entry->uname); } } } return alternate_host; } static void stonith_send_reply(xmlNode * reply, int call_options, const char *remote_peer, const char *client_id) { if (remote_peer) { send_cluster_message(crm_get_peer(0, remote_peer), crm_msg_stonith_ng, reply, FALSE); } else { do_local_reply(reply, client_id, is_set(call_options, st_opt_sync_call), remote_peer != NULL); } } static int handle_request(crm_client_t * client, uint32_t id, uint32_t flags, xmlNode * request, const char *remote_peer) { int call_options = 0; int rc = -EOPNOTSUPP; xmlNode *data = NULL; xmlNode *reply = NULL; char *output = NULL; const char *op = crm_element_value(request, F_STONITH_OPERATION); const char *client_id = crm_element_value(request, F_STONITH_CLIENTID); crm_element_value_int(request, F_STONITH_CALLOPTS, &call_options); if (is_set(call_options, st_opt_sync_call)) { CRM_ASSERT(client == NULL || client->request_id == id); } if (crm_str_eq(op, CRM_OP_REGISTER, TRUE)) { xmlNode *reply = create_xml_node(NULL, "reply"); CRM_ASSERT(client); crm_xml_add(reply, F_STONITH_OPERATION, CRM_OP_REGISTER); crm_xml_add(reply, F_STONITH_CLIENTID, client->id); crm_ipcs_send(client, id, reply, flags); client->request_id = 0; free_xml(reply); return 0; } else if (crm_str_eq(op, STONITH_OP_EXEC, TRUE)) { rc = stonith_device_action(request, &output); } else if (crm_str_eq(op, STONITH_OP_TIMEOUT_UPDATE, TRUE)) { const char *call_id = crm_element_value(request, F_STONITH_CALLID); const char *client_id = crm_element_value(request, F_STONITH_CLIENTID); int op_timeout = 0; crm_element_value_int(request, F_STONITH_TIMEOUT, &op_timeout); do_stonith_async_timeout_update(client_id, call_id, op_timeout); return 0; } else if (crm_str_eq(op, STONITH_OP_QUERY, TRUE)) { if (remote_peer) { create_remote_stonith_op(client_id, request, TRUE); /* Record it for the future notification */ } stonith_query(request, remote_peer, client_id, call_options); return 0; } else if (crm_str_eq(op, T_STONITH_NOTIFY, TRUE)) { const char *flag_name = NULL; CRM_ASSERT(client); flag_name = crm_element_value(request, F_STONITH_NOTIFY_ACTIVATE); if (flag_name) { crm_debug("Setting %s callbacks for %s (%s): ON", flag_name, client->name, client->id); client->options |= get_stonith_flag(flag_name); } flag_name = crm_element_value(request, F_STONITH_NOTIFY_DEACTIVATE); if (flag_name) { crm_debug("Setting %s callbacks for %s (%s): off", flag_name, client->name, client->id); client->options |= get_stonith_flag(flag_name); } if (flags & crm_ipc_client_response) { crm_ipcs_send_ack(client, id, flags, "ack", __FUNCTION__, __LINE__); } return 0; } else if (crm_str_eq(op, STONITH_OP_RELAY, TRUE)) { xmlNode *dev = get_xpath_object("//@" F_STONITH_TARGET, request, LOG_TRACE); crm_notice("Peer %s has received a forwarded fencing request from %s to fence (%s) peer %s", stonith_our_uname, client ? client->name : remote_peer, crm_element_value(dev, F_STONITH_ACTION), crm_element_value(dev, F_STONITH_TARGET)); if (initiate_remote_stonith_op(NULL, request, FALSE) != NULL) { rc = -EINPROGRESS; } } else if (crm_str_eq(op, STONITH_OP_FENCE, TRUE)) { if (remote_peer || stand_alone) { rc = stonith_fence(request); } else if (call_options & st_opt_manual_ack) { remote_fencing_op_t *rop = NULL; xmlNode *dev = get_xpath_object("//@" F_STONITH_TARGET, request, LOG_TRACE); const char *target = crm_element_value(dev, F_STONITH_TARGET); crm_notice("Received manual confirmation that %s is fenced", target); rop = initiate_remote_stonith_op(client, request, TRUE); rc = stonith_manual_ack(request, rop); } else { const char *alternate_host = NULL; xmlNode *dev = get_xpath_object("//@" F_STONITH_TARGET, request, LOG_TRACE); const char *target = crm_element_value(dev, F_STONITH_TARGET); const char *action = crm_element_value(dev, F_STONITH_ACTION); const char *device = crm_element_value(dev, F_STONITH_DEVICE); if (client) { int tolerance = 0; crm_notice("Client %s.%.8s wants to fence (%s) '%s' with device '%s'", client->name, client->id, action, target, device ? device : "(any)"); crm_element_value_int(dev, F_STONITH_TOLERANCE, &tolerance); if (stonith_check_fence_tolerance(tolerance, target, action)) { rc = 0; goto done; } } else { crm_notice("Peer %s wants to fence (%s) '%s' with device '%s'", remote_peer, action, target, device ? device : "(any)"); } alternate_host = check_alternate_host(target); if (alternate_host && client) { const char *client_id = NULL; crm_notice("Forwarding complex self fencing request to peer %s", alternate_host); if (client->id) { client_id = client->id; } else { client_id = crm_element_value(request, F_STONITH_CLIENTID); } /* Create a record of it, otherwise call_id will be 0 if we need to notify of failures */ create_remote_stonith_op(client_id, request, FALSE); crm_xml_add(request, F_STONITH_OPERATION, STONITH_OP_RELAY); crm_xml_add(request, F_STONITH_CLIENTID, client->id); send_cluster_message(crm_get_peer(0, alternate_host), crm_msg_stonith_ng, request, FALSE); rc = -EINPROGRESS; } else if (initiate_remote_stonith_op(client, request, FALSE) != NULL) { rc = -EINPROGRESS; } } } else if (crm_str_eq(op, STONITH_OP_FENCE_HISTORY, TRUE)) { - rc = stonith_fence_history(request, &data); + rc = stonith_fence_history(request, &data, remote_peer, call_options); + if (call_options & st_opt_discard_reply) { + /* we don't expect answers to the broadcast + * we might have sent out + */ + free_xml(data); + return pcmk_ok; + } } else if (crm_str_eq(op, STONITH_OP_DEVICE_ADD, TRUE)) { const char *device_id = NULL; rc = stonith_device_register(request, &device_id, FALSE); do_stonith_notify_device(call_options, op, rc, device_id); } else if (crm_str_eq(op, STONITH_OP_DEVICE_DEL, TRUE)) { xmlNode *dev = get_xpath_object("//" F_STONITH_DEVICE, request, LOG_ERR); const char *device_id = crm_element_value(dev, XML_ATTR_ID); rc = stonith_device_remove(device_id, FALSE); do_stonith_notify_device(call_options, op, rc, device_id); } else if (crm_str_eq(op, STONITH_OP_LEVEL_ADD, TRUE)) { char *device_id = NULL; rc = stonith_level_register(request, &device_id); do_stonith_notify_level(call_options, op, rc, device_id); free(device_id); } else if (crm_str_eq(op, STONITH_OP_LEVEL_DEL, TRUE)) { char *device_id = NULL; rc = stonith_level_remove(request, &device_id); do_stonith_notify_level(call_options, op, rc, device_id); } else if(safe_str_eq(op, CRM_OP_RM_NODE_CACHE)) { int node_id = 0; const char *name = NULL; crm_element_value_int(request, XML_ATTR_ID, &node_id); name = crm_element_value(request, XML_ATTR_UNAME); reap_crm_member(node_id, name); return pcmk_ok; } else { crm_err("Unknown %s from %s", op, client ? client->name : remote_peer); crm_log_xml_warn(request, "UnknownOp"); } done: /* Always reply unless the request is in process still. * If in progress, a reply will happen async after the request * processing is finished */ if (rc != -EINPROGRESS) { crm_trace("Reply handling: %p %u %u %d %d %s", client, client?client->request_id:0, id, is_set(call_options, st_opt_sync_call), call_options, crm_element_value(request, F_STONITH_CALLOPTS)); if (is_set(call_options, st_opt_sync_call)) { CRM_ASSERT(client == NULL || client->request_id == id); } reply = stonith_construct_reply(request, output, data, rc); stonith_send_reply(reply, call_options, remote_peer, client_id); } free(output); free_xml(data); free_xml(reply); return rc; } static void handle_reply(crm_client_t * client, xmlNode * request, const char *remote_peer) { const char *op = crm_element_value(request, F_STONITH_OPERATION); if (crm_str_eq(op, STONITH_OP_QUERY, TRUE)) { process_remote_stonith_query(request); } else if (crm_str_eq(op, T_STONITH_NOTIFY, TRUE)) { process_remote_stonith_exec(request); } else if (crm_str_eq(op, STONITH_OP_FENCE, TRUE)) { /* Reply to a complex fencing op */ process_remote_stonith_exec(request); } else { crm_err("Unknown %s reply from %s", op, client ? client->name : remote_peer); crm_log_xml_warn(request, "UnknownOp"); } } void stonith_command(crm_client_t * client, uint32_t id, uint32_t flags, xmlNode * request, const char *remote_peer) { int call_options = 0; int rc = 0; gboolean is_reply = FALSE; /* Copy op for reporting. The original might get freed by handle_reply() * before we use it in crm_debug(): * handle_reply() * |- process_remote_stonith_exec() * |-- remote_op_done() * |--- handle_local_reply_and_notify() * |---- crm_xml_add(...F_STONITH_OPERATION...) * |--- free_xml(op->request) */ char *op = crm_element_value_copy(request, F_STONITH_OPERATION); if (get_xpath_object("//" T_STONITH_REPLY, request, LOG_TRACE)) { is_reply = TRUE; } crm_element_value_int(request, F_STONITH_CALLOPTS, &call_options); crm_debug("Processing %s%s %u from %s (%16x)", op, is_reply ? " reply" : "", id, client ? client->name : remote_peer, call_options); if (is_set(call_options, st_opt_sync_call)) { CRM_ASSERT(client == NULL || client->request_id == id); } if (is_reply) { handle_reply(client, request, remote_peer); } else { rc = handle_request(client, id, flags, request, remote_peer); } crm_debug("Processed %s%s from %s: %s (%d)", op, is_reply ? " reply" : "", client ? client->name : remote_peer, rc > 0 ? "" : pcmk_strerror(rc), rc); free(op); } diff --git a/daemons/fenced/fenced_history.c b/daemons/fenced/fenced_history.c new file mode 100644 index 0000000000..3ec354ede9 --- /dev/null +++ b/daemons/fenced/fenced_history.c @@ -0,0 +1,354 @@ +/* + * Copyright 2009-2018 Andrew Beekhof + * + * 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 + +/*! + * \internal + * \brief Send a broadcast to all nodes to trigger cleanup or + * history synchronisation + * + * \param[in] history Optional history to be attached + * \param[in] callopts We control cleanup via a flag in the callopts + * \param[in] target Cleanup can be limited to certain fence-targets + */ +static void +stonith_send_broadcast_history(xmlNode *history, + int callopts, + const char *target) +{ + xmlNode *bcast = create_xml_node(NULL, "stonith_command"); + xmlNode *data = create_xml_node(NULL, __FUNCTION__); + + if (target) { + crm_xml_add(data, F_STONITH_TARGET, target); + } + crm_xml_add(bcast, F_TYPE, T_STONITH_NG); + crm_xml_add(bcast, F_SUBTYPE, "broadcast"); + crm_xml_add(bcast, F_STONITH_OPERATION, STONITH_OP_FENCE_HISTORY); + crm_xml_add_int(bcast, F_STONITH_CALLOPTS, callopts); + if (history) { + add_node_copy(data, history); + } + add_message_xml(bcast, F_STONITH_CALLDATA, data); + send_cluster_message(NULL, crm_msg_stonith_ng, bcast, FALSE); + + free_xml(data); + free_xml(bcast); +} + +static gboolean +stonith_remove_history_entry (gpointer key, + gpointer value, + gpointer user_data) +{ + remote_fencing_op_t *op = value; + const char *target = (const char *) user_data; + + if ((op->state == st_failed) || (op->state == st_done)) { + if ((target) && (strcmp(op->target, target) != 0)) { + return FALSE; + } + return TRUE; + } + + return FALSE; /* don't clean pending operations */ +} + +/*! + * \internal + * \brief Send out a cleanup broadcast and do a local history-cleanup + * + * \param[in] request The request message that triggered cleanup + * \param[in] target Cleanup can be limited to certain fence-targets + */ +static void +stonith_fence_history_cleanup(xmlNode *request, const char *target) +{ + if (crm_element_value(request, F_STONITH_CALLID)) { + stonith_send_broadcast_history(NULL, + st_opt_cleanup | st_opt_discard_reply, + target); + /* we'll do the local clean when we receive back our own broadcast */ + } else if (stonith_remote_op_list) { + g_hash_table_foreach_remove(stonith_remote_op_list, + stonith_remove_history_entry, + (gpointer) target); + do_stonith_notify(0, T_STONITH_NOTIFY_HISTORY, 0, NULL); + } +} + +/*! + * \internal + * \brief Convert xml fence-history to a hash-table like stonith_remote_op_list + * + * \param[in] history Fence-history in xml + * + * \return Fence-history as hash-table + */ +static GHashTable * +stonith_xml_history_to_list(xmlNode *history) +{ + xmlNode *xml_op = NULL; + GHashTable *rv = NULL; + + init_stonith_remote_op_hash_table(&rv); + + CRM_LOG_ASSERT(rv != NULL); + + for (xml_op = __xml_first_child(history); xml_op != NULL; + xml_op = __xml_next(xml_op)) { + remote_fencing_op_t *op = NULL; + char *id = crm_element_value_copy(xml_op, F_STONITH_REMOTE_OP_ID); + int completed, state; + + if (!id) { + crm_warn("History to convert to hashtable has no id in entry"); + continue; + } + + crm_trace("Attaching op %s to hashtable", id); + + op = calloc(1, sizeof(remote_fencing_op_t)); + + op->id = id; + op->target = crm_element_value_copy(xml_op, F_STONITH_TARGET); + op->action = crm_element_value_copy(xml_op, F_STONITH_ACTION); + op->originator = crm_element_value_copy(xml_op, F_STONITH_ORIGIN); + op->delegate = crm_element_value_copy(xml_op, F_STONITH_DELEGATE); + op->client_name = crm_element_value_copy(xml_op, F_STONITH_CLIENTNAME); + crm_element_value_int(xml_op, F_STONITH_DATE, &completed); + op->completed = (time_t) completed; + crm_element_value_int(xml_op, F_STONITH_STATE, &state); + op->state = (enum op_state) state; + + g_hash_table_replace(rv, id, op); + CRM_LOG_ASSERT(g_hash_table_lookup(rv, id) != NULL); + } + + return rv; +} + +/*! + * \internal + * \brief Craft xml difference between local fence-history and a history + * coming from remote + * + * \param[in] remote_history Fence-history as hash-table (may be NULL) + * \param[in] add_id If crafting the answer for an API + * history-request there is no need for the id + * \param[in] target Optionally limit to certain fence-target + * + * \return The fence-history as xml + */ +static xmlNode * +stonith_local_history_diff(GHashTable *remote_history, + gboolean add_id, + const char *target) +{ + xmlNode *history = NULL; + int cnt = 0; + + if (stonith_remote_op_list) { + GHashTableIter iter; + remote_fencing_op_t *op = NULL; + + history = create_xml_node(NULL, F_STONITH_HISTORY_LIST); + + g_hash_table_iter_init(&iter, stonith_remote_op_list); + while (g_hash_table_iter_next(&iter, NULL, (void **)&op)) { + xmlNode *entry = NULL; + + if (remote_history && + g_hash_table_lookup(remote_history, op->id)) { + continue; /* skip entries broadcasted already */ + } + + if (target && strcmp(op->target, target) != 0) { + continue; + } + + cnt++; + crm_trace("Attaching op %s", op->id); + entry = create_xml_node(history, STONITH_OP_EXEC); + if (add_id) { + crm_xml_add(entry, F_STONITH_REMOTE_OP_ID, op->id); + } + crm_xml_add(entry, F_STONITH_TARGET, op->target); + crm_xml_add(entry, F_STONITH_ACTION, op->action); + crm_xml_add(entry, F_STONITH_ORIGIN, op->originator); + crm_xml_add(entry, F_STONITH_DELEGATE, op->delegate); + crm_xml_add(entry, F_STONITH_CLIENTNAME, op->client_name); + crm_xml_add_int(entry, F_STONITH_DATE, op->completed); + crm_xml_add_int(entry, F_STONITH_STATE, op->state); + } + } + + if (cnt == 0) { + free_xml(history); + return NULL; + } else { + return history; + } +} + +/*! + * \internal + * \brief Merge fence-history coming from remote into local history + * + * \param[in] history Hash-table holding remote history to be merged in + */ +static void +stonith_merge_in_history_list(GHashTable *history) +{ + GHashTableIter iter; + remote_fencing_op_t *op = NULL; + gboolean updated = FALSE; + + if (!history) { + return; + } + + init_stonith_remote_op_hash_table(&stonith_remote_op_list); + + g_hash_table_iter_init(&iter, history); + while (g_hash_table_iter_next(&iter, NULL, (void **)&op)) { + remote_fencing_op_t *stored_op = + g_hash_table_lookup(stonith_remote_op_list, op->id); + + if (stored_op) { + continue; /* skip over existant - state-merging migh be desirable */ + } + + updated = TRUE; + g_hash_table_iter_steal(&iter); + CRM_LOG_ASSERT(g_hash_table_insert(stonith_remote_op_list, op->id, op)); + } + if (updated) { + do_stonith_notify(0, T_STONITH_NOTIFY_HISTORY, 0, NULL); + } + g_hash_table_destroy(history); /* remove what is left */ +} + +/*! + * \internal + * \brief Handle fence-history messages (either from API or coming in as + * broadcasts + * + * \param[in] msg Request message + * \param[in] output In case of a request from the API used to craft + * a reply from + * \param[in] remote_peer + * \param[in] options call-options from the request + * + * \return always success as there is actully nothing that can go really wrong + */ +int +stonith_fence_history(xmlNode *msg, xmlNode **output, + const char *remote_peer, int options) +{ + int rc = 0; + const char *target = NULL; + xmlNode *dev = get_xpath_object("//@" F_STONITH_TARGET, msg, LOG_TRACE); + char *nodename = NULL; + xmlNode *out_history = NULL; + + if (dev) { + target = crm_element_value(dev, F_STONITH_TARGET); + if (target && (options & st_opt_cs_nodeid)) { + int nodeid = crm_atoi(target, NULL); + + nodename = stonith_get_peer_name(nodeid); + if (nodename) { + target = nodename; + } + } + } + + if (options & st_opt_cleanup) { + crm_trace("Cleaning up operations on %s in %p", target, + stonith_remote_op_list); + + stonith_fence_history_cleanup(msg, target); + } else if (options & st_opt_broadcast) { + if (crm_element_value(msg, F_STONITH_CALLID)) { + /* this is coming from the stonith-API + * + * craft a broadcast with node's history + * so that every node can merge and broadcast + * what it has on top + */ + out_history = stonith_local_history_diff(NULL, TRUE, NULL); + crm_trace("Broadcasting history to peers"); + stonith_send_broadcast_history(out_history, + st_opt_broadcast | st_opt_discard_reply, + NULL); + } else if (remote_peer && + !safe_str_eq(remote_peer, stonith_our_uname)) { + xmlNode *history = + get_xpath_object("//" F_STONITH_HISTORY_LIST, msg, LOG_TRACE); + GHashTable *received_history = + history?stonith_xml_history_to_list(history):NULL; + + /* either a broadcast created directly upon stonith-API request + * or a diff as response to such a thing + * + * in both cases it may have a history or not + * if we have differential data + * merge in what we've received and stop + * otherwise broadcast what we have on top + * marking as differential and merge in afterwards + */ + if (!history || + !crm_is_true(crm_element_value(history, + F_STONITH_DIFFERENTIAL))) { + out_history = + stonith_local_history_diff(received_history, TRUE, NULL); + if (out_history) { + crm_trace("Broadcasting history-diff to peers"); + crm_xml_add(out_history, F_STONITH_DIFFERENTIAL, + XML_BOOLEAN_TRUE); + stonith_send_broadcast_history(out_history, + st_opt_broadcast | st_opt_discard_reply, + NULL); + } else { + crm_trace("History-diff is empty - skip broadcast"); + } + } + stonith_merge_in_history_list(received_history); + } else { + crm_trace("Skipping history-query-broadcast (%s%s)" + " we sent ourselves", + remote_peer?"remote-peer=":"local-ipc", + remote_peer?remote_peer:""); + } + } else { + /* plain history request */ + crm_trace("Looking for operations on %s in %p", target, + stonith_remote_op_list); + *output = stonith_local_history_diff(NULL, FALSE, target); + } + free(nodename); + free_xml(out_history); + return rc; +} diff --git a/daemons/fenced/fenced_remote.c b/daemons/fenced/fenced_remote.c index ade3c92161..753fafaa67 100644 --- a/daemons/fenced/fenced_remote.c +++ b/daemons/fenced/fenced_remote.c @@ -1,2121 +1,2080 @@ /* * Copyright 2009-2018 Andrew Beekhof * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #define TIMEOUT_MULTIPLY_FACTOR 1.2 /* When one fencer queries its peers for devices able to handle a fencing * request, each peer will reply with a list of such devices available to it. * Each reply will be parsed into a st_query_result_t, with each device's * information kept in a device_properties_t. */ typedef struct device_properties_s { /* Whether access to this device has been verified */ gboolean verified; /* The remaining members are indexed by the operation's "phase" */ /* Whether this device has been executed in each phase */ gboolean executed[st_phase_max]; /* Whether this device is disallowed from executing in each phase */ gboolean disallowed[st_phase_max]; /* Action-specific timeout for each phase */ int custom_action_timeout[st_phase_max]; /* Action-specific maximum random delay for each phase */ int delay_max[st_phase_max]; /* Action-specific base delay for each phase */ int delay_base[st_phase_max]; } device_properties_t; typedef struct st_query_result_s { /* Name of peer that sent this result */ char *host; /* Only try peers for non-topology based operations once */ gboolean tried; /* Number of entries in the devices table */ int ndevices; /* Devices available to this host that are capable of fencing the target */ GHashTable *devices; } st_query_result_t; -static GHashTable *remote_op_list = NULL; +GHashTable *stonith_remote_op_list = NULL; void call_remote_stonith(remote_fencing_op_t * op, st_query_result_t * peer); static void remote_op_done(remote_fencing_op_t * op, xmlNode * data, int rc, int dup); extern xmlNode *stonith_create_op(int call_id, const char *token, const char *op, xmlNode * data, int call_options); static void report_timeout_period(remote_fencing_op_t * op, int op_timeout); static int get_op_total_timeout(const remote_fencing_op_t *op, const st_query_result_t *chosen_peer); static gint sort_strings(gconstpointer a, gconstpointer b) { return strcmp(a, b); } static void free_remote_query(gpointer data) { if (data) { st_query_result_t *query = data; crm_trace("Free'ing query result from %s", query->host); g_hash_table_destroy(query->devices); free(query->host); free(query); } } void -free_remote_op_list() +free_stonith_remote_op_list() { - if (remote_op_list != NULL) { - g_hash_table_destroy(remote_op_list); - remote_op_list = NULL; + if (stonith_remote_op_list != NULL) { + g_hash_table_destroy(stonith_remote_op_list); + stonith_remote_op_list = NULL; } } struct peer_count_data { const remote_fencing_op_t *op; gboolean verified_only; int count; }; /*! * \internal * \brief Increment a counter if a device has not been executed yet * * \param[in] key Device ID (ignored) * \param[in] value Device properties * \param[in] user_data Peer count data */ static void count_peer_device(gpointer key, gpointer value, gpointer user_data) { device_properties_t *props = (device_properties_t*)value; struct peer_count_data *data = user_data; if (!props->executed[data->op->phase] && (!data->verified_only || props->verified)) { ++(data->count); } } /*! * \internal * \brief Check the number of available devices in a peer's query results * * \param[in] op Operation that results are for * \param[in] peer Peer to count * \param[in] verified_only Whether to count only verified devices * * \return Number of devices available to peer that were not already executed */ static int count_peer_devices(const remote_fencing_op_t *op, const st_query_result_t *peer, gboolean verified_only) { struct peer_count_data data; data.op = op; data.verified_only = verified_only; data.count = 0; if (peer) { g_hash_table_foreach(peer->devices, count_peer_device, &data); } return data.count; } /*! * \internal * \brief Search for a device in a query result * * \param[in] op Operation that result is for * \param[in] peer Query result for a peer * \param[in] device Device ID to search for * * \return Device properties if found, NULL otherwise */ static device_properties_t * find_peer_device(const remote_fencing_op_t *op, const st_query_result_t *peer, const char *device) { device_properties_t *props = g_hash_table_lookup(peer->devices, device); return (props && !props->executed[op->phase] && !props->disallowed[op->phase])? props : NULL; } /*! * \internal * \brief Find a device in a peer's device list and mark it as executed * * \param[in] op Operation that peer result is for * \param[in,out] peer Peer with results to search * \param[in] device ID of device to mark as done * \param[in] verified_devices_only Only consider verified devices * * \return TRUE if device was found and marked, FALSE otherwise */ static gboolean grab_peer_device(const remote_fencing_op_t *op, st_query_result_t *peer, const char *device, gboolean verified_devices_only) { device_properties_t *props = find_peer_device(op, peer, device); if ((props == NULL) || (verified_devices_only && !props->verified)) { return FALSE; } crm_trace("Removing %s from %s (%d remaining)", device, peer->host, count_peer_devices(op, peer, FALSE)); props->executed[op->phase] = TRUE; return TRUE; } static void clear_remote_op_timers(remote_fencing_op_t * op) { if (op->query_timer) { g_source_remove(op->query_timer); op->query_timer = 0; } if (op->op_timer_total) { g_source_remove(op->op_timer_total); op->op_timer_total = 0; } if (op->op_timer_one) { g_source_remove(op->op_timer_one); op->op_timer_one = 0; } } static void free_remote_op(gpointer data) { remote_fencing_op_t *op = data; crm_trace("Free'ing op %s for %s", op->id, op->target); crm_log_xml_debug(op->request, "Destroying"); clear_remote_op_timers(op); free(op->id); free(op->action); free(op->delegate); free(op->target); free(op->client_id); free(op->client_name); free(op->originator); if (op->query_results) { g_list_free_full(op->query_results, free_remote_query); } if (op->request) { free_xml(op->request); op->request = NULL; } if (op->devices_list) { g_list_free_full(op->devices_list, free); op->devices_list = NULL; } g_list_free_full(op->automatic_list, free); g_list_free(op->duplicates); free(op); } +void +init_stonith_remote_op_hash_table(GHashTable **table) +{ + if (*table == NULL) { + *table = g_hash_table_new_full(crm_str_hash, g_str_equal, NULL, free_remote_op); + } +} + /*! * \internal * \brief Return an operation's originally requested action (before any remap) * * \param[in] op Operation to check * * \return Operation's original action */ static const char * op_requested_action(const remote_fencing_op_t *op) { return ((op->phase > st_phase_requested)? "reboot" : op->action); } /*! * \internal * \brief Remap a "reboot" operation to the "off" phase * * \param[in,out] op Operation to remap */ static void op_phase_off(remote_fencing_op_t *op) { crm_info("Remapping multiple-device reboot of %s (%s) to off", op->target, op->id); op->phase = st_phase_off; /* Happily, "off" and "on" are shorter than "reboot", so we can reuse the * memory allocation at each phase. */ strcpy(op->action, "off"); } /*! * \internal * \brief Advance a remapped reboot operation to the "on" phase * * \param[in,out] op Operation to remap */ static void op_phase_on(remote_fencing_op_t *op) { GListPtr iter = NULL; crm_info("Remapped off of %s complete, remapping to on for %s.%.8s", op->target, op->client_name, op->id); op->phase = st_phase_on; strcpy(op->action, "on"); /* Skip devices with automatic unfencing, because the cluster will handle it * when the node rejoins. */ for (iter = op->automatic_list; iter != NULL; iter = iter->next) { GListPtr match = g_list_find_custom(op->devices_list, iter->data, sort_strings); if (match) { op->devices_list = g_list_remove(op->devices_list, match->data); } } g_list_free_full(op->automatic_list, free); op->automatic_list = NULL; /* Rewind device list pointer */ op->devices = op->devices_list; } /*! * \internal * \brief Reset a remapped reboot operation * * \param[in,out] op Operation to reset */ static void undo_op_remap(remote_fencing_op_t *op) { if (op->phase > 0) { crm_info("Undoing remap of reboot of %s for %s.%.8s", op->target, op->client_name, op->id); op->phase = st_phase_requested; strcpy(op->action, "reboot"); } } static xmlNode * create_op_done_notify(remote_fencing_op_t * op, int rc) { xmlNode *notify_data = create_xml_node(NULL, T_STONITH_NOTIFY_FENCE); crm_xml_add_int(notify_data, "state", op->state); crm_xml_add_int(notify_data, F_STONITH_RC, rc); crm_xml_add(notify_data, F_STONITH_TARGET, op->target); crm_xml_add(notify_data, F_STONITH_ACTION, op->action); crm_xml_add(notify_data, F_STONITH_DELEGATE, op->delegate); crm_xml_add(notify_data, F_STONITH_REMOTE_OP_ID, op->id); crm_xml_add(notify_data, F_STONITH_ORIGIN, op->originator); crm_xml_add(notify_data, F_STONITH_CLIENTID, op->client_id); crm_xml_add(notify_data, F_STONITH_CLIENTNAME, op->client_name); return notify_data; } static void bcast_result_to_peers(remote_fencing_op_t * op, int rc) { static int count = 0; xmlNode *bcast = create_xml_node(NULL, T_STONITH_REPLY); xmlNode *notify_data = create_op_done_notify(op, rc); count++; crm_trace("Broadcasting result to peers"); crm_xml_add(bcast, F_TYPE, T_STONITH_NOTIFY); crm_xml_add(bcast, F_SUBTYPE, "broadcast"); crm_xml_add(bcast, F_STONITH_OPERATION, T_STONITH_NOTIFY); crm_xml_add_int(bcast, "count", count); add_message_xml(bcast, F_STONITH_CALLDATA, notify_data); send_cluster_message(NULL, crm_msg_stonith_ng, bcast, FALSE); free_xml(notify_data); free_xml(bcast); return; } static void handle_local_reply_and_notify(remote_fencing_op_t * op, xmlNode * data, int rc) { xmlNode *notify_data = NULL; xmlNode *reply = NULL; if (op->notify_sent == TRUE) { /* nothing to do */ return; } /* Do notification with a clean data object */ notify_data = create_op_done_notify(op, rc); crm_xml_add_int(data, "state", op->state); crm_xml_add(data, F_STONITH_TARGET, op->target); crm_xml_add(data, F_STONITH_OPERATION, op->action); reply = stonith_construct_reply(op->request, NULL, data, rc); crm_xml_add(reply, F_STONITH_DELEGATE, op->delegate); /* Send fencing OP reply to local client that initiated fencing */ do_local_reply(reply, op->client_id, op->call_options & st_opt_sync_call, FALSE); /* bcast to all local clients that the fencing operation happend */ do_stonith_notify(0, T_STONITH_NOTIFY_FENCE, rc, notify_data); + do_stonith_notify(0, T_STONITH_NOTIFY_HISTORY, 0, NULL); /* mark this op as having notify's already sent */ op->notify_sent = TRUE; free_xml(reply); free_xml(notify_data); } static void handle_duplicates(remote_fencing_op_t * op, xmlNode * data, int rc) { GListPtr iter = NULL; for (iter = op->duplicates; iter != NULL; iter = iter->next) { remote_fencing_op_t *other = iter->data; if (other->state == st_duplicate) { other->state = op->state; crm_debug("Performing duplicate notification for %s@%s.%.8s = %s", other->client_name, other->originator, other->id, pcmk_strerror(rc)); remote_op_done(other, data, rc, TRUE); } else { // Possible if (for example) it timed out already crm_err("Skipping duplicate notification for %s@%s - %d", other->client_name, other->originator, other->state); } } } /*! * \internal * \brief Finalize a remote operation. * * \description This function has two code paths. * * Path 1. This node is the owner of the operation and needs * to notify the cpg group via a broadcast as to the operation's * results. * * Path 2. The cpg broadcast is received. All nodes notify their local * stonith clients the operation results. * * So, The owner of the operation first notifies the cluster of the result, * and once that cpg notify is received back it notifies all the local clients. * * Nodes that are passive watchers of the operation will receive the * broadcast and only need to notify their local clients the operation finished. * * \param op, The fencing operation to finalize * \param data, The xml msg reply (if present) of the last delegated fencing * operation. * \param dup, Is this operation a duplicate, if so treat it a little differently * making sure the broadcast is not sent out. */ static void remote_op_done(remote_fencing_op_t * op, xmlNode * data, int rc, int dup) { int level = LOG_ERR; const char *subt = NULL; xmlNode *local_data = NULL; op->completed = time(NULL); clear_remote_op_timers(op); undo_op_remap(op); if (op->notify_sent == TRUE) { crm_err("Already sent notifications for '%s of %s by %s' (for=%s@%s.%.8s, state=%d): %s", op->action, op->target, op->delegate ? op->delegate : "", op->client_name, op->originator, op->id, op->state, pcmk_strerror(rc)); goto remote_op_done_cleanup; } if (!op->delegate && data && rc != -ENODEV && rc != -EHOSTUNREACH) { xmlNode *ndata = get_xpath_object("//@" F_STONITH_DELEGATE, data, LOG_TRACE); if(ndata) { op->delegate = crm_element_value_copy(ndata, F_STONITH_DELEGATE); } else { op->delegate = crm_element_value_copy(data, F_ORIG); } } if (data == NULL) { data = create_xml_node(NULL, "remote-op"); local_data = data; } /* Tell everyone the operation is done, we will continue * with doing the local notifications once we receive * the broadcast back. */ subt = crm_element_value(data, F_SUBTYPE); if (dup == FALSE && safe_str_neq(subt, "broadcast")) { /* Defer notification until the bcast message arrives */ bcast_result_to_peers(op, rc); goto remote_op_done_cleanup; } if (rc == pcmk_ok || dup) { level = LOG_NOTICE; } else if (safe_str_neq(op->originator, stonith_our_uname)) { level = LOG_NOTICE; } do_crm_log(level, "Operation %s of %s by %s for %s@%s.%.8s: %s", op->action, op->target, op->delegate ? op->delegate : "", op->client_name, op->originator, op->id, pcmk_strerror(rc)); handle_local_reply_and_notify(op, data, rc); if (dup == FALSE) { handle_duplicates(op, data, rc); } /* Free non-essential parts of the record * Keep the record around so we can query the history */ if (op->query_results) { g_list_free_full(op->query_results, free_remote_query); op->query_results = NULL; } if (op->request) { free_xml(op->request); op->request = NULL; } remote_op_done_cleanup: free_xml(local_data); } static gboolean remote_op_watchdog_done(gpointer userdata) { remote_fencing_op_t *op = userdata; op->op_timer_one = 0; crm_notice("Self-fencing (%s) by %s for %s.%8s assumed complete", op->action, op->target, op->client_name, op->id); op->state = st_done; remote_op_done(op, NULL, pcmk_ok, FALSE); return FALSE; } static gboolean remote_op_timeout_one(gpointer userdata) { remote_fencing_op_t *op = userdata; op->op_timer_one = 0; crm_notice("Peer's fencing (%s) of %s for %s timed out" CRM_XS "id=%s", op->action, op->target, op->client_name, op->id); call_remote_stonith(op, NULL); return FALSE; } static gboolean remote_op_timeout(gpointer userdata) { remote_fencing_op_t *op = userdata; op->op_timer_total = 0; if (op->state == st_done) { crm_debug("Action %s (%s) for %s (%s) already completed", op->action, op->id, op->target, op->client_name); return FALSE; } crm_debug("Action %s (%s) for %s (%s) timed out", op->action, op->id, op->target, op->client_name); if (op->phase == st_phase_on) { /* A remapped reboot operation timed out in the "on" phase, but the * "off" phase completed successfully, so quit trying any further * devices, and return success. */ remote_op_done(op, NULL, pcmk_ok, FALSE); return FALSE; } op->state = st_failed; remote_op_done(op, NULL, -ETIME, FALSE); return FALSE; } static gboolean remote_op_query_timeout(gpointer data) { remote_fencing_op_t *op = data; op->query_timer = 0; if (op->state == st_done) { crm_debug("Operation %s for %s already completed", op->id, op->target); } else if (op->state == st_exec) { crm_debug("Operation %s for %s already in progress", op->id, op->target); } else if (op->query_results) { crm_debug("Query %s for %s complete: %d", op->id, op->target, op->state); call_remote_stonith(op, NULL); } else { crm_debug("Query %s for %s timed out: %d", op->id, op->target, op->state); if (op->op_timer_total) { g_source_remove(op->op_timer_total); op->op_timer_total = 0; } remote_op_timeout(op); } return FALSE; } static gboolean topology_is_empty(stonith_topology_t *tp) { int i; if (tp == NULL) { return TRUE; } for (i = 0; i < ST_LEVEL_MAX; i++) { if (tp->levels[i] != NULL) { return FALSE; } } return TRUE; } /*! * \internal * \brief Add a device to an operation's automatic unfencing list * * \param[in,out] op Operation to modify * \param[in] device Device ID to add */ static void add_required_device(remote_fencing_op_t *op, const char *device) { GListPtr match = g_list_find_custom(op->automatic_list, device, sort_strings); if (!match) { op->automatic_list = g_list_prepend(op->automatic_list, strdup(device)); } } /*! * \internal * \brief Remove a device from the automatic unfencing list * * \param[in,out] op Operation to modify * \param[in] device Device ID to remove */ static void remove_required_device(remote_fencing_op_t *op, const char *device) { GListPtr match = g_list_find_custom(op->automatic_list, device, sort_strings); if (match) { op->automatic_list = g_list_remove(op->automatic_list, match->data); } } /* deep copy the device list */ static void set_op_device_list(remote_fencing_op_t * op, GListPtr devices) { GListPtr lpc = NULL; if (op->devices_list) { g_list_free_full(op->devices_list, free); op->devices_list = NULL; } for (lpc = devices; lpc != NULL; lpc = lpc->next) { op->devices_list = g_list_append(op->devices_list, strdup(lpc->data)); } op->devices = op->devices_list; } /*! * \internal * \brief Check whether a node matches a topology target * * \param[in] tp Topology table entry to check * \param[in] node Name of node to check * * \return TRUE if node matches topology target */ static gboolean topology_matches(const stonith_topology_t *tp, const char *node) { regex_t r_patt; CRM_CHECK(node && tp && tp->target, return FALSE); switch(tp->kind) { case 2: /* This level targets by attribute, so tp->target is a NAME=VALUE pair * of a permanent attribute applied to targeted nodes. The test below * relies on the locally cached copy of the CIB, so if fencing needs to * be done before the initial CIB is received or after a malformed CIB * is received, then the topology will be unable to be used. */ if (node_has_attr(node, tp->target_attribute, tp->target_value)) { crm_notice("Matched %s with %s by attribute", node, tp->target); return TRUE; } break; case 1: /* This level targets by name, so tp->target is a regular expression * matching names of nodes to be targeted. */ if (regcomp(&r_patt, tp->target_pattern, REG_EXTENDED|REG_NOSUB)) { crm_info("Bad regex '%s' for fencing level", tp->target); } else { int status = regexec(&r_patt, node, 0, NULL, 0); regfree(&r_patt); if (status == 0) { crm_notice("Matched %s with %s by name", node, tp->target); return TRUE; } } break; case 0: crm_trace("Testing %s against %s", node, tp->target); return safe_str_eq(tp->target, node); } crm_trace("No match for %s with %s", node, tp->target); return FALSE; } stonith_topology_t * find_topology_for_host(const char *host) { GHashTableIter tIter; stonith_topology_t *tp = g_hash_table_lookup(topology, host); if(tp != NULL) { crm_trace("Found %s for %s in %d entries", tp->target, host, g_hash_table_size(topology)); return tp; } g_hash_table_iter_init(&tIter, topology); while (g_hash_table_iter_next(&tIter, NULL, (gpointer *) & tp)) { if (topology_matches(tp, host)) { crm_trace("Found %s for %s in %d entries", tp->target, host, g_hash_table_size(topology)); return tp; } } crm_trace("No matches for %s in %d topology entries", host, g_hash_table_size(topology)); return NULL; } /*! * \internal * \brief Set fencing operation's device list to target's next topology level * * \param[in,out] op Remote fencing operation to modify * * \return pcmk_ok if successful, target was not specified (i.e. queries) or * target has no topology, or -EINVAL if no more topology levels to try */ static int stonith_topology_next(remote_fencing_op_t * op) { stonith_topology_t *tp = NULL; if (op->target) { /* Queries don't have a target set */ tp = find_topology_for_host(op->target); } if (topology_is_empty(tp)) { return pcmk_ok; } set_bit(op->call_options, st_opt_topology); /* This is a new level, so undo any remapping left over from previous */ undo_op_remap(op); do { op->level++; } while (op->level < ST_LEVEL_MAX && tp->levels[op->level] == NULL); if (op->level < ST_LEVEL_MAX) { crm_trace("Attempting fencing level %d for %s (%d devices) - %s@%s.%.8s", op->level, op->target, g_list_length(tp->levels[op->level]), op->client_name, op->originator, op->id); set_op_device_list(op, tp->levels[op->level]); if (g_list_next(op->devices_list) && safe_str_eq(op->action, "reboot")) { /* A reboot has been requested for a topology level with multiple * devices. Instead of rebooting the devices sequentially, we will * turn them all off, then turn them all on again. (Think about * switched power outlets for redundant power supplies.) */ op_phase_off(op); } return pcmk_ok; } crm_notice("All fencing options to fence %s for %s@%s.%.8s failed", op->target, op->client_name, op->originator, op->id); return -EINVAL; } /*! * \brief Check to see if this operation is a duplicate of another in flight * operation. If so merge this operation into the inflight operation, and mark * it as a duplicate. */ static void merge_duplicates(remote_fencing_op_t * op) { GHashTableIter iter; remote_fencing_op_t *other = NULL; time_t now = time(NULL); - g_hash_table_iter_init(&iter, remote_op_list); + g_hash_table_iter_init(&iter, stonith_remote_op_list); while (g_hash_table_iter_next(&iter, NULL, (void **)&other)) { crm_node_t *peer = NULL; const char *other_action = op_requested_action(other); if (other->state > st_exec) { /* Must be in-progress */ continue; } else if (safe_str_neq(op->target, other->target)) { /* Must be for the same node */ continue; } else if (safe_str_neq(op->action, other_action)) { crm_trace("Must be for the same action: %s vs. %s", op->action, other_action); continue; } else if (safe_str_eq(op->client_name, other->client_name)) { crm_trace("Must be for different clients: %s", op->client_name); continue; } else if (safe_str_eq(other->target, other->originator)) { crm_trace("Can't be a suicide operation: %s", other->target); continue; } peer = crm_get_peer(0, other->originator); if(fencing_peer_active(peer) == FALSE) { crm_notice("Failing stonith action %s for node %s originating from %s@%s.%.8s: Originator is dead", other->action, other->target, other->client_name, other->originator, other->id); other->state = st_failed; continue; } else if(other->total_timeout > 0 && now > (other->total_timeout + other->created)) { crm_info("Stonith action %s for node %s originating from %s@%s.%.8s is too old: %ld vs. %ld + %d", other->action, other->target, other->client_name, other->originator, other->id, now, other->created, other->total_timeout); continue; } /* There is another in-flight request to fence the same host * Piggyback on that instead. If it fails, so do we. */ other->duplicates = g_list_append(other->duplicates, op); if (other->total_timeout == 0) { crm_trace("Making a best-guess as to the timeout used"); other->total_timeout = op->total_timeout = TIMEOUT_MULTIPLY_FACTOR * get_op_total_timeout(op, NULL); } crm_notice ("Merging stonith action %s for node %s originating from client %s.%.8s with identical request from %s@%s.%.8s (%ds)", op->action, op->target, op->client_name, op->id, other->client_name, other->originator, other->id, other->total_timeout); report_timeout_period(op, other->total_timeout); op->state = st_duplicate; } } static uint32_t fencing_active_peers(void) { uint32_t count = 0; crm_node_t *entry; GHashTableIter gIter; g_hash_table_iter_init(&gIter, crm_peer_cache); while (g_hash_table_iter_next(&gIter, NULL, (void **)&entry)) { if(fencing_peer_active(entry)) { count++; } } return count; } int stonith_manual_ack(xmlNode * msg, remote_fencing_op_t * op) { xmlNode *dev = get_xpath_object("//@" F_STONITH_TARGET, msg, LOG_ERR); op->state = st_done; op->completed = time(NULL); op->delegate = strdup("a human"); crm_notice("Injecting manual confirmation that %s is safely off/down", crm_element_value(dev, F_STONITH_TARGET)); remote_op_done(op, msg, pcmk_ok, FALSE); /* Replies are sent via done_cb->stonith_send_async_reply()->do_local_reply() */ return -EINPROGRESS; } char * stonith_get_peer_name(unsigned int nodeid) { crm_node_t *node = crm_find_peer(nodeid, NULL); char *nodename = NULL; if (node && node->uname) { return strdup(node->uname); } else if ((nodename = get_node_name(nodeid))) { return nodename; } else { const char *last_known_name = g_hash_table_lookup(known_peer_names, GUINT_TO_POINTER(nodeid)); if (last_known_name) { crm_debug("Use the last known name %s for nodeid %u", last_known_name, nodeid); return strdup(last_known_name); } } return NULL; } /*! * \internal * \brief Create a new remote stonith operation * * \param[in] client ID of local stonith client that initiated the operation * \param[in] request The request from the client that started the operation * \param[in] peer TRUE if this operation is owned by another stonith peer * (an operation owned by one peer is stored on all peers, * but only the owner executes it; all nodes get the results * once the owner finishes execution) */ void * create_remote_stonith_op(const char *client, xmlNode * request, gboolean peer) { remote_fencing_op_t *op = NULL; xmlNode *dev = get_xpath_object("//@" F_STONITH_TARGET, request, LOG_TRACE); int call_options = 0; - if (remote_op_list == NULL) { - remote_op_list = g_hash_table_new_full(crm_str_hash, g_str_equal, NULL, free_remote_op); - } + init_stonith_remote_op_hash_table(&stonith_remote_op_list); /* If this operation is owned by another node, check to make * sure we haven't already created this operation. */ if (peer && dev) { const char *op_id = crm_element_value(dev, F_STONITH_REMOTE_OP_ID); CRM_CHECK(op_id != NULL, return NULL); - op = g_hash_table_lookup(remote_op_list, op_id); + op = g_hash_table_lookup(stonith_remote_op_list, op_id); if (op) { crm_debug("%s already exists", op_id); return op; } } op = calloc(1, sizeof(remote_fencing_op_t)); crm_element_value_int(request, F_STONITH_TIMEOUT, &(op->base_timeout)); if (peer && dev) { op->id = crm_element_value_copy(dev, F_STONITH_REMOTE_OP_ID); } else { op->id = crm_generate_uuid(); } - g_hash_table_replace(remote_op_list, op->id, op); - CRM_LOG_ASSERT(g_hash_table_lookup(remote_op_list, op->id) != NULL); + g_hash_table_replace(stonith_remote_op_list, op->id, op); + CRM_LOG_ASSERT(g_hash_table_lookup(stonith_remote_op_list, op->id) != NULL); crm_trace("Created %s", op->id); op->state = st_query; op->replies_expected = fencing_active_peers(); op->action = crm_element_value_copy(dev, F_STONITH_ACTION); op->originator = crm_element_value_copy(dev, F_STONITH_ORIGIN); op->delegate = crm_element_value_copy(dev, F_STONITH_DELEGATE); /* May not be set */ op->created = time(NULL); if (op->originator == NULL) { /* Local or relayed request */ op->originator = strdup(stonith_our_uname); } CRM_LOG_ASSERT(client != NULL); if (client) { op->client_id = strdup(client); } op->client_name = crm_element_value_copy(request, F_STONITH_CLIENTNAME); op->target = crm_element_value_copy(dev, F_STONITH_TARGET); op->request = copy_xml(request); /* TODO: Figure out how to avoid this */ crm_element_value_int(request, F_STONITH_CALLOPTS, &call_options); op->call_options = call_options; crm_element_value_int(request, F_STONITH_CALLID, &(op->client_callid)); crm_trace("%s new stonith op: %s - %s of %s for %s", (peer && dev) ? "Recorded" : "Generated", op->id, op->action, op->target, op->client_name); if (op->call_options & st_opt_cs_nodeid) { int nodeid = crm_atoi(op->target, NULL); char *nodename = stonith_get_peer_name(nodeid); /* Ensure the conversion only happens once */ op->call_options &= ~st_opt_cs_nodeid; if (nodename) { free(op->target); op->target = nodename; } else { crm_warn("Could not expand nodeid '%s' into a host name", op->target); } } /* check to see if this is a duplicate operation of another in-flight operation */ merge_duplicates(op); + if (op->state != st_duplicate) { + /* kick history readers */ + do_stonith_notify(0, T_STONITH_NOTIFY_HISTORY, 0, NULL); + } + return op; } remote_fencing_op_t * initiate_remote_stonith_op(crm_client_t * client, xmlNode * request, gboolean manual_ack) { int query_timeout = 0; xmlNode *query = NULL; const char *client_id = NULL; remote_fencing_op_t *op = NULL; if (client) { client_id = client->id; } else { client_id = crm_element_value(request, F_STONITH_CLIENTID); } CRM_LOG_ASSERT(client_id != NULL); op = create_remote_stonith_op(client_id, request, FALSE); op->owner = TRUE; if (manual_ack) { crm_notice("Initiating manual confirmation for %s: %s", op->target, op->id); return op; } CRM_CHECK(op->action, return NULL); if (stonith_topology_next(op) != pcmk_ok) { op->state = st_failed; } switch (op->state) { case st_failed: crm_warn("Could not request peer fencing (%s) of %s " CRM_XS " id=%s", op->action, op->target, op->id); remote_op_done(op, NULL, -EINVAL, FALSE); return op; case st_duplicate: crm_info("Requesting peer fencing (%s) of %s (duplicate) " CRM_XS " id=%s", op->action, op->target, op->id); return op; default: crm_notice("Requesting peer fencing (%s) of %s " CRM_XS " id=%s state=%d", op->action, op->target, op->id, op->state); } query = stonith_create_op(op->client_callid, op->id, STONITH_OP_QUERY, NULL, op->call_options); crm_xml_add(query, F_STONITH_REMOTE_OP_ID, op->id); crm_xml_add(query, F_STONITH_TARGET, op->target); crm_xml_add(query, F_STONITH_ACTION, op_requested_action(op)); crm_xml_add(query, F_STONITH_ORIGIN, op->originator); crm_xml_add(query, F_STONITH_CLIENTID, op->client_id); crm_xml_add(query, F_STONITH_CLIENTNAME, op->client_name); crm_xml_add_int(query, F_STONITH_TIMEOUT, op->base_timeout); send_cluster_message(NULL, crm_msg_stonith_ng, query, FALSE); free_xml(query); query_timeout = op->base_timeout * TIMEOUT_MULTIPLY_FACTOR; op->query_timer = g_timeout_add((1000 * query_timeout), remote_op_query_timeout, op); return op; } enum find_best_peer_options { /*! Skip checking the target peer for capable fencing devices */ FIND_PEER_SKIP_TARGET = 0x0001, /*! Only check the target peer for capable fencing devices */ FIND_PEER_TARGET_ONLY = 0x0002, /*! Skip peers and devices that are not verified */ FIND_PEER_VERIFIED_ONLY = 0x0004, }; static st_query_result_t * find_best_peer(const char *device, remote_fencing_op_t * op, enum find_best_peer_options options) { GListPtr iter = NULL; gboolean verified_devices_only = (options & FIND_PEER_VERIFIED_ONLY) ? TRUE : FALSE; if (!device && is_set(op->call_options, st_opt_topology)) { return NULL; } for (iter = op->query_results; iter != NULL; iter = iter->next) { st_query_result_t *peer = iter->data; crm_trace("Testing result from %s for %s with %d devices: %d %x", peer->host, op->target, peer->ndevices, peer->tried, options); if ((options & FIND_PEER_SKIP_TARGET) && safe_str_eq(peer->host, op->target)) { continue; } if ((options & FIND_PEER_TARGET_ONLY) && safe_str_neq(peer->host, op->target)) { continue; } if (is_set(op->call_options, st_opt_topology)) { if (grab_peer_device(op, peer, device, verified_devices_only)) { return peer; } } else if ((peer->tried == FALSE) && count_peer_devices(op, peer, verified_devices_only)) { /* No topology: Use the current best peer */ crm_trace("Simple fencing"); return peer; } } return NULL; } static st_query_result_t * stonith_choose_peer(remote_fencing_op_t * op) { const char *device = NULL; st_query_result_t *peer = NULL; uint32_t active = fencing_active_peers(); do { if (op->devices) { device = op->devices->data; crm_trace("Checking for someone to fence (%s) %s with %s", op->action, op->target, device); } else { crm_trace("Checking for someone to fence (%s) %s", op->action, op->target); } /* Best choice is a peer other than the target with verified access */ peer = find_best_peer(device, op, FIND_PEER_SKIP_TARGET|FIND_PEER_VERIFIED_ONLY); if (peer) { crm_trace("Found verified peer %s for %s", peer->host, device?device:""); return peer; } if(op->query_timer != 0 && op->replies < QB_MIN(op->replies_expected, active)) { crm_trace("Waiting before looking for unverified devices to fence %s", op->target); return NULL; } /* If no other peer has verified access, next best is unverified access */ peer = find_best_peer(device, op, FIND_PEER_SKIP_TARGET); if (peer) { crm_trace("Found best unverified peer %s", peer->host); return peer; } /* If no other peer can do it, last option is self-fencing * (which is never allowed for the "on" phase of a remapped reboot) */ if (op->phase != st_phase_on) { peer = find_best_peer(device, op, FIND_PEER_TARGET_ONLY); if (peer) { crm_trace("%s will fence itself", peer->host); return peer; } } /* Try the next fencing level if there is one (unless we're in the "on" * phase of a remapped "reboot", because we ignore errors in that case) */ } while ((op->phase != st_phase_on) && is_set(op->call_options, st_opt_topology) && stonith_topology_next(op) == pcmk_ok); crm_notice("Couldn't find anyone to fence (%s) %s with %s", op->action, op->target, (device? device : "any device")); return NULL; } static int get_device_timeout(const remote_fencing_op_t *op, const st_query_result_t *peer, const char *device) { device_properties_t *props; if (!peer || !device) { return op->base_timeout; } props = g_hash_table_lookup(peer->devices, device); if (!props) { return op->base_timeout; } return (props->custom_action_timeout[op->phase]? props->custom_action_timeout[op->phase] : op->base_timeout) + props->delay_max[op->phase]; } struct timeout_data { const remote_fencing_op_t *op; const st_query_result_t *peer; int total_timeout; }; /*! * \internal * \brief Add timeout to a total if device has not been executed yet * * \param[in] key GHashTable key (device ID) * \param[in] value GHashTable value (device properties) * \param[in] user_data Timeout data */ static void add_device_timeout(gpointer key, gpointer value, gpointer user_data) { const char *device_id = key; device_properties_t *props = value; struct timeout_data *timeout = user_data; if (!props->executed[timeout->op->phase] && !props->disallowed[timeout->op->phase]) { timeout->total_timeout += get_device_timeout(timeout->op, timeout->peer, device_id); } } static int get_peer_timeout(const remote_fencing_op_t *op, const st_query_result_t *peer) { struct timeout_data timeout; timeout.op = op; timeout.peer = peer; timeout.total_timeout = 0; g_hash_table_foreach(peer->devices, add_device_timeout, &timeout); return (timeout.total_timeout? timeout.total_timeout : op->base_timeout); } static int get_op_total_timeout(const remote_fencing_op_t *op, const st_query_result_t *chosen_peer) { int total_timeout = 0; stonith_topology_t *tp = find_topology_for_host(op->target); if (is_set(op->call_options, st_opt_topology) && tp) { int i; GListPtr device_list = NULL; GListPtr iter = NULL; /* Yep, this looks scary, nested loops all over the place. * Here is what is going on. * Loop1: Iterate through fencing levels. * Loop2: If a fencing level has devices, loop through each device * Loop3: For each device in a fencing level, see what peer owns it * and what that peer has reported the timeout is for the device. */ for (i = 0; i < ST_LEVEL_MAX; i++) { if (!tp->levels[i]) { continue; } for (device_list = tp->levels[i]; device_list; device_list = device_list->next) { for (iter = op->query_results; iter != NULL; iter = iter->next) { const st_query_result_t *peer = iter->data; if (find_peer_device(op, peer, device_list->data)) { total_timeout += get_device_timeout(op, peer, device_list->data); break; } } /* End Loop3: match device with peer that owns device, find device's timeout period */ } /* End Loop2: iterate through devices at a specific level */ } /*End Loop1: iterate through fencing levels */ } else if (chosen_peer) { total_timeout = get_peer_timeout(op, chosen_peer); } else { total_timeout = op->base_timeout; } return total_timeout ? total_timeout : op->base_timeout; } static void report_timeout_period(remote_fencing_op_t * op, int op_timeout) { GListPtr iter = NULL; xmlNode *update = NULL; const char *client_node = NULL; const char *client_id = NULL; const char *call_id = NULL; if (op->call_options & st_opt_sync_call) { /* There is no reason to report the timeout for a synchronous call. It * is impossible to use the reported timeout to do anything when the client * is blocking for the response. This update is only important for * async calls that require a callback to report the results in. */ return; } else if (!op->request) { return; } crm_trace("Reporting timeout for %s.%.8s", op->client_name, op->id); client_node = crm_element_value(op->request, F_STONITH_CLIENTNODE); call_id = crm_element_value(op->request, F_STONITH_CALLID); client_id = crm_element_value(op->request, F_STONITH_CLIENTID); if (!client_node || !call_id || !client_id) { return; } if (safe_str_eq(client_node, stonith_our_uname)) { /* The client is connected to this node, send the update direclty to them */ do_stonith_async_timeout_update(client_id, call_id, op_timeout); return; } /* The client is connected to another node, relay this update to them */ update = stonith_create_op(op->client_callid, op->id, STONITH_OP_TIMEOUT_UPDATE, NULL, 0); crm_xml_add(update, F_STONITH_REMOTE_OP_ID, op->id); crm_xml_add(update, F_STONITH_CLIENTID, client_id); crm_xml_add(update, F_STONITH_CALLID, call_id); crm_xml_add_int(update, F_STONITH_TIMEOUT, op_timeout); send_cluster_message(crm_get_peer(0, client_node), crm_msg_stonith_ng, update, FALSE); free_xml(update); for (iter = op->duplicates; iter != NULL; iter = iter->next) { remote_fencing_op_t *dup = iter->data; crm_trace("Reporting timeout for duplicate %s.%.8s", dup->client_name, dup->id); report_timeout_period(iter->data, op_timeout); } } /*! * \internal * \brief Advance an operation to the next device in its topology * * \param[in,out] op Operation to advance * \param[in] device ID of device just completed * \param[in] msg XML reply that contained device result (if available) * \param[in] rc Return code of device's execution */ static void advance_op_topology(remote_fencing_op_t *op, const char *device, xmlNode *msg, int rc) { /* Advance to the next device at this topology level, if any */ if (op->devices) { op->devices = op->devices->next; } /* Handle automatic unfencing if an "on" action was requested */ if ((op->phase == st_phase_requested) && safe_str_eq(op->action, "on")) { /* If the device we just executed was required, it's not anymore */ remove_required_device(op, device); /* If there are no more devices at this topology level, run through any * remaining devices with automatic unfencing */ if (op->devices == NULL) { op->devices = op->automatic_list; } } if ((op->devices == NULL) && (op->phase == st_phase_off)) { /* We're done with this level and with required devices, but we had * remapped "reboot" to "off", so start over with "on". If any devices * need to be turned back on, op->devices will be non-NULL after this. */ op_phase_on(op); } if (op->devices) { /* Necessary devices remain, so execute the next one */ crm_trace("Next for %s on behalf of %s@%s (rc was %d)", op->target, op->originator, op->client_name, rc); call_remote_stonith(op, NULL); } else { /* We're done with all devices and phases, so finalize operation */ crm_trace("Marking complex fencing op for %s as complete", op->target); op->state = st_done; remote_op_done(op, msg, rc, FALSE); } } void call_remote_stonith(remote_fencing_op_t * op, st_query_result_t * peer) { const char *device = NULL; int timeout = op->base_timeout; crm_trace("State for %s.%.8s: %s %d", op->target, op->client_name, op->id, op->state); if (peer == NULL && !is_set(op->call_options, st_opt_topology)) { peer = stonith_choose_peer(op); } if (!op->op_timer_total) { int total_timeout = get_op_total_timeout(op, peer); op->total_timeout = TIMEOUT_MULTIPLY_FACTOR * total_timeout; op->op_timer_total = g_timeout_add(1000 * op->total_timeout, remote_op_timeout, op); report_timeout_period(op, op->total_timeout); crm_info("Total timeout set to %d for peer's fencing of %s for %s" CRM_XS "id=%s", total_timeout, op->target, op->client_name, op->id); } if (is_set(op->call_options, st_opt_topology) && op->devices) { /* Ignore any peer preference, they might not have the device we need */ /* When using topology, stonith_choose_peer() removes the device from * further consideration, so be sure to calculate timeout beforehand */ peer = stonith_choose_peer(op); device = op->devices->data; timeout = get_device_timeout(op, peer, device); } if (peer) { int timeout_one = 0; xmlNode *remote_op = stonith_create_op(op->client_callid, op->id, STONITH_OP_FENCE, NULL, 0); crm_xml_add(remote_op, F_STONITH_REMOTE_OP_ID, op->id); crm_xml_add(remote_op, F_STONITH_TARGET, op->target); crm_xml_add(remote_op, F_STONITH_ACTION, op->action); crm_xml_add(remote_op, F_STONITH_ORIGIN, op->originator); crm_xml_add(remote_op, F_STONITH_CLIENTID, op->client_id); crm_xml_add(remote_op, F_STONITH_CLIENTNAME, op->client_name); crm_xml_add_int(remote_op, F_STONITH_TIMEOUT, timeout); crm_xml_add_int(remote_op, F_STONITH_CALLOPTS, op->call_options); if (device) { timeout_one = TIMEOUT_MULTIPLY_FACTOR * get_device_timeout(op, peer, device); crm_info("Requesting that '%s' perform op '%s %s' with '%s' for %s (%ds)", peer->host, op->target, op->action, device, op->client_name, timeout_one); crm_xml_add(remote_op, F_STONITH_DEVICE, device); crm_xml_add(remote_op, F_STONITH_MODE, "slave"); } else { timeout_one = TIMEOUT_MULTIPLY_FACTOR * get_peer_timeout(op, peer); crm_info("Requesting that '%s' perform op '%s %s' for %s (%ds, %lds)", peer->host, op->target, op->action, op->client_name, timeout_one, stonith_watchdog_timeout_ms); crm_xml_add(remote_op, F_STONITH_MODE, "smart"); } op->state = st_exec; if (op->op_timer_one) { g_source_remove(op->op_timer_one); } if(stonith_watchdog_timeout_ms > 0 && device && safe_str_eq(device, "watchdog")) { crm_notice("Waiting %lds for %s to self-fence (%s) for %s.%.8s (%p)", stonith_watchdog_timeout_ms/1000, op->target, op->action, op->client_name, op->id, device); op->op_timer_one = g_timeout_add(stonith_watchdog_timeout_ms, remote_op_watchdog_done, op); /* TODO check devices to verify watchdog will be in use */ } else if(stonith_watchdog_timeout_ms > 0 && safe_str_eq(peer->host, op->target) && safe_str_neq(op->action, "on")) { crm_notice("Waiting %lds for %s to self-fence (%s) for %s.%.8s (%p)", stonith_watchdog_timeout_ms/1000, op->target, op->action, op->client_name, op->id, device); op->op_timer_one = g_timeout_add(stonith_watchdog_timeout_ms, remote_op_watchdog_done, op); } else { op->op_timer_one = g_timeout_add((1000 * timeout_one), remote_op_timeout_one, op); } send_cluster_message(crm_get_peer(0, peer->host), crm_msg_stonith_ng, remote_op, FALSE); peer->tried = TRUE; free_xml(remote_op); return; } else if (op->phase == st_phase_on) { /* A remapped "on" cannot be executed, but the node was already * turned off successfully, so ignore the error and continue. */ crm_warn("Ignoring %s 'on' failure (no capable peers) for %s after successful 'off'", device, op->target); advance_op_topology(op, device, NULL, pcmk_ok); return; } else if (op->owner == FALSE) { crm_err("Fencing (%s) of %s for %s is not ours to control", op->action, op->target, op->client_name); } else if (op->query_timer == 0) { /* We've exhausted all available peers */ crm_info("No remaining peers capable of fencing (%s) %s for %s (%d)", op->target, op->action, op->client_name, op->state); CRM_LOG_ASSERT(op->state < st_done); remote_op_timeout(op); } else if(op->replies >= op->replies_expected || op->replies >= fencing_active_peers()) { int rc = -EHOSTUNREACH; /* if the operation never left the query state, * but we have all the expected replies, then no devices * are available to execute the fencing operation. */ if(stonith_watchdog_timeout_ms && (device == NULL || safe_str_eq(device, "watchdog"))) { crm_notice("Waiting %lds for %s to self-fence (%s) for %s.%.8s (%p)", stonith_watchdog_timeout_ms/1000, op->target, op->action, op->client_name, op->id, device); op->op_timer_one = g_timeout_add(stonith_watchdog_timeout_ms, remote_op_watchdog_done, op); return; } if (op->state == st_query) { crm_info("No peers (out of %d) have devices capable of fencing (%s) %s for %s (%d)", op->replies, op->action, op->target, op->client_name, op->state); rc = -ENODEV; } else { crm_info("No peers (out of %d) are capable of fencing (%s) %s for %s (%d)", op->replies, op->action, op->target, op->client_name, op->state); } op->state = st_failed; remote_op_done(op, NULL, rc, FALSE); } else if (device) { crm_info("Waiting for additional peers capable of fencing (%s) %s with %s for %s.%.8s", op->action, op->target, device, op->client_name, op->id); } else { crm_info("Waiting for additional peers capable of fencing (%s) %s for %s%.8s", op->action, op->target, op->client_name, op->id); } } /*! * \internal * \brief Comparison function for sorting query results * * \param[in] a GList item to compare * \param[in] b GList item to compare * * \return Per the glib documentation, "a negative integer if the first value * comes before the second, 0 if they are equal, or a positive integer * if the first value comes after the second." */ static gint sort_peers(gconstpointer a, gconstpointer b) { const st_query_result_t *peer_a = a; const st_query_result_t *peer_b = b; return (peer_b->ndevices - peer_a->ndevices); } /*! * \internal * \brief Determine if all the devices in the topology are found or not */ static gboolean all_topology_devices_found(remote_fencing_op_t * op) { GListPtr device = NULL; GListPtr iter = NULL; device_properties_t *match = NULL; stonith_topology_t *tp = NULL; gboolean skip_target = FALSE; int i; tp = find_topology_for_host(op->target); if (!tp) { return FALSE; } if (safe_str_eq(op->action, "off") || safe_str_eq(op->action, "reboot")) { /* Don't count the devices on the target node if we are killing * the target node. */ skip_target = TRUE; } for (i = 0; i < ST_LEVEL_MAX; i++) { for (device = tp->levels[i]; device; device = device->next) { match = NULL; for (iter = op->query_results; iter && !match; iter = iter->next) { st_query_result_t *peer = iter->data; if (skip_target && safe_str_eq(peer->host, op->target)) { continue; } match = find_peer_device(op, peer, device->data); } if (!match) { return FALSE; } } } return TRUE; } /*! * \internal * \brief Parse action-specific device properties from XML * * \param[in] msg XML element containing the properties * \param[in] peer Name of peer that sent XML (for logs) * \param[in] device Device ID (for logs) * \param[in] action Action the properties relate to (for logs) * \param[in] phase Phase the properties relate to * \param[in,out] props Device properties to update */ static void parse_action_specific(xmlNode *xml, const char *peer, const char *device, const char *action, remote_fencing_op_t *op, enum st_remap_phase phase, device_properties_t *props) { props->custom_action_timeout[phase] = 0; crm_element_value_int(xml, F_STONITH_ACTION_TIMEOUT, &props->custom_action_timeout[phase]); if (props->custom_action_timeout[phase]) { crm_trace("Peer %s with device %s returned %s action timeout %d", peer, device, action, props->custom_action_timeout[phase]); } props->delay_max[phase] = 0; crm_element_value_int(xml, F_STONITH_DELAY_MAX, &props->delay_max[phase]); if (props->delay_max[phase]) { crm_trace("Peer %s with device %s returned maximum of random delay %d for %s", peer, device, props->delay_max[phase], action); } props->delay_base[phase] = 0; crm_element_value_int(xml, F_STONITH_DELAY_BASE, &props->delay_base[phase]); if (props->delay_base[phase]) { crm_trace("Peer %s with device %s returned base delay %d for %s", peer, device, props->delay_base[phase], action); } /* Handle devices with automatic unfencing */ if (safe_str_eq(action, "on")) { int required = 0; crm_element_value_int(xml, F_STONITH_DEVICE_REQUIRED, &required); if (required) { crm_trace("Peer %s requires device %s to execute for action %s", peer, device, action); add_required_device(op, device); } } /* If a reboot is remapped to off+on, it's possible that a node is allowed * to perform one action but not another. */ if (crm_is_true(crm_element_value(xml, F_STONITH_ACTION_DISALLOWED))) { props->disallowed[phase] = TRUE; crm_trace("Peer %s is disallowed from executing %s for device %s", peer, action, device); } } /*! * \internal * \brief Parse one device's properties from peer's XML query reply * * \param[in] xml XML node containing device properties * \param[in,out] op Operation that query and reply relate to * \param[in,out] result Peer's results * \param[in] device ID of device being parsed */ static void add_device_properties(xmlNode *xml, remote_fencing_op_t *op, st_query_result_t *result, const char *device) { xmlNode *child; int verified = 0; device_properties_t *props = calloc(1, sizeof(device_properties_t)); /* Add a new entry to this result's devices list */ CRM_ASSERT(props != NULL); g_hash_table_insert(result->devices, strdup(device), props); /* Peers with verified (monitored) access will be preferred */ crm_element_value_int(xml, F_STONITH_DEVICE_VERIFIED, &verified); if (verified) { crm_trace("Peer %s has confirmed a verified device %s", result->host, device); props->verified = TRUE; } /* Parse action-specific device properties */ parse_action_specific(xml, result->host, device, op_requested_action(op), op, st_phase_requested, props); for (child = __xml_first_child(xml); child != NULL; child = __xml_next(child)) { /* Replies for "reboot" operations will include the action-specific * values for "off" and "on" in child elements, just in case the reboot * winds up getting remapped. */ if (safe_str_eq(ID(child), "off")) { parse_action_specific(child, result->host, device, "off", op, st_phase_off, props); } else if (safe_str_eq(ID(child), "on")) { parse_action_specific(child, result->host, device, "on", op, st_phase_on, props); } } } /*! * \internal * \brief Parse a peer's XML query reply and add it to operation's results * * \param[in,out] op Operation that query and reply relate to * \param[in] host Name of peer that sent this reply * \param[in] ndevices Number of devices expected in reply * \param[in] xml XML node containing device list * * \return Newly allocated result structure with parsed reply */ static st_query_result_t * add_result(remote_fencing_op_t *op, const char *host, int ndevices, xmlNode *xml) { st_query_result_t *result = calloc(1, sizeof(st_query_result_t)); xmlNode *child; CRM_CHECK(result != NULL, return NULL); result->host = strdup(host); result->devices = crm_str_table_new(); /* Each child element describes one capable device available to the peer */ for (child = __xml_first_child(xml); child != NULL; child = __xml_next(child)) { const char *device = ID(child); if (device) { add_device_properties(child, op, result, device); } } result->ndevices = g_hash_table_size(result->devices); CRM_CHECK(ndevices == result->ndevices, crm_err("Query claimed to have %d devices but %d found", ndevices, result->ndevices)); op->query_results = g_list_insert_sorted(op->query_results, result, sort_peers); return result; } /*! * \internal * \brief Handle a peer's reply to our fencing query * * Parse a query result from XML and store it in the remote operation * table, and when enough replies have been received, issue a fencing request. * * \param[in] msg XML reply received * * \return pcmk_ok on success, -errno on error * * \note See initiate_remote_stonith_op() for how the XML query was initially * formed, and stonith_query() for how the peer formed its XML reply. */ int process_remote_stonith_query(xmlNode * msg) { int ndevices = 0; gboolean host_is_target = FALSE; gboolean have_all_replies = FALSE; const char *id = NULL; const char *host = NULL; remote_fencing_op_t *op = NULL; st_query_result_t *result = NULL; uint32_t replies_expected; xmlNode *dev = get_xpath_object("//@" F_STONITH_REMOTE_OP_ID, msg, LOG_ERR); CRM_CHECK(dev != NULL, return -EPROTO); id = crm_element_value(dev, F_STONITH_REMOTE_OP_ID); CRM_CHECK(id != NULL, return -EPROTO); dev = get_xpath_object("//@" F_STONITH_AVAILABLE_DEVICES, msg, LOG_ERR); CRM_CHECK(dev != NULL, return -EPROTO); crm_element_value_int(dev, F_STONITH_AVAILABLE_DEVICES, &ndevices); - op = g_hash_table_lookup(remote_op_list, id); + op = g_hash_table_lookup(stonith_remote_op_list, id); if (op == NULL) { crm_debug("Received query reply for unknown or expired operation %s", id); return -EOPNOTSUPP; } replies_expected = QB_MIN(op->replies_expected, fencing_active_peers()); if ((++op->replies >= replies_expected) && (op->state == st_query)) { have_all_replies = TRUE; } host = crm_element_value(msg, F_ORIG); host_is_target = safe_str_eq(host, op->target); crm_info("Query result %d of %d from %s for %s/%s (%d devices) %s", op->replies, replies_expected, host, op->target, op->action, ndevices, id); if (ndevices > 0) { result = add_result(op, host, ndevices, dev); } if (is_set(op->call_options, st_opt_topology)) { /* If we start the fencing before all the topology results are in, * it is possible fencing levels will be skipped because of the missing * query results. */ if (op->state == st_query && all_topology_devices_found(op)) { /* All the query results are in for the topology, start the fencing ops. */ crm_trace("All topology devices found"); call_remote_stonith(op, result); } else if (have_all_replies) { crm_info("All topology query replies have arrived, continuing (%d expected/%d received) ", replies_expected, op->replies); call_remote_stonith(op, NULL); } } else if (op->state == st_query) { int nverified = count_peer_devices(op, result, TRUE); /* We have a result for a non-topology fencing op that looks promising, * go ahead and start fencing before query timeout */ if (result && (host_is_target == FALSE) && nverified) { /* we have a verified device living on a peer that is not the target */ crm_trace("Found %d verified devices", nverified); call_remote_stonith(op, result); } else if (have_all_replies) { crm_info("All query replies have arrived, continuing (%d expected/%d received) ", replies_expected, op->replies); call_remote_stonith(op, NULL); } else { crm_trace("Waiting for more peer results before launching fencing operation"); } } else if (result && (op->state == st_done)) { crm_info("Discarding query result from %s (%d devices): Operation is in state %d", result->host, result->ndevices, op->state); } return pcmk_ok; } /*! * \internal * \brief Handle a peer's reply to a fencing request * * Parse a fencing reply from XML, and either finalize the operation * or attempt another device as appropriate. * * \param[in] msg XML reply received * * \return pcmk_ok on success, -errno on error */ int process_remote_stonith_exec(xmlNode * msg) { int rc = 0; const char *id = NULL; const char *device = NULL; remote_fencing_op_t *op = NULL; xmlNode *dev = get_xpath_object("//@" F_STONITH_REMOTE_OP_ID, msg, LOG_ERR); CRM_CHECK(dev != NULL, return -EPROTO); id = crm_element_value(dev, F_STONITH_REMOTE_OP_ID); CRM_CHECK(id != NULL, return -EPROTO); dev = get_xpath_object("//@" F_STONITH_RC, msg, LOG_ERR); CRM_CHECK(dev != NULL, return -EPROTO); crm_element_value_int(dev, F_STONITH_RC, &rc); device = crm_element_value(dev, F_STONITH_DEVICE); - if (remote_op_list) { - op = g_hash_table_lookup(remote_op_list, id); + if (stonith_remote_op_list) { + op = g_hash_table_lookup(stonith_remote_op_list, id); } if (op == NULL && rc == pcmk_ok) { /* Record successful fencing operations */ const char *client_id = crm_element_value(dev, F_STONITH_CLIENTID); op = create_remote_stonith_op(client_id, dev, TRUE); } if (op == NULL) { /* Could be for an event that began before we started */ /* TODO: Record the op for later querying */ crm_info("Received peer result of unknown or expired operation %s", id); return -EOPNOTSUPP; } if (op->devices && device && safe_str_neq(op->devices->data, device)) { crm_err("Received outdated reply for device %s (instead of %s) to " "fence (%s) %s. Operation already timed out at peer level.", device, (const char *) op->devices->data, op->action, op->target); return rc; } if (safe_str_eq(crm_element_value(msg, F_SUBTYPE), "broadcast")) { crm_debug("Marking call to %s for %s on behalf of %s@%s.%.8s: %s (%d)", op->action, op->target, op->client_name, op->id, op->originator, pcmk_strerror(rc), rc); if (rc == pcmk_ok) { op->state = st_done; } else { op->state = st_failed; } remote_op_done(op, msg, rc, FALSE); return pcmk_ok; } else if (safe_str_neq(op->originator, stonith_our_uname)) { /* If this isn't a remote level broadcast, and we are not the * originator of the operation, we should not be receiving this msg. */ crm_err ("%s received non-broadcast fencing result for operation it does not own (device %s targeting %s)", stonith_our_uname, device, op->target); return rc; } if (is_set(op->call_options, st_opt_topology)) { const char *device = crm_element_value(msg, F_STONITH_DEVICE); crm_notice("Call to %s for '%s %s' on behalf of %s@%s: %s (%d)", device, op->target, op->action, op->client_name, op->originator, pcmk_strerror(rc), rc); /* We own the op, and it is complete. broadcast the result to all nodes * and notify our local clients. */ if (op->state == st_done) { remote_op_done(op, msg, rc, FALSE); return rc; } if ((op->phase == 2) && (rc != pcmk_ok)) { /* A remapped "on" failed, but the node was already turned off * successfully, so ignore the error and continue. */ crm_warn("Ignoring %s 'on' failure (exit code %d) for %s after successful 'off'", device, rc, op->target); rc = pcmk_ok; } if (rc == pcmk_ok) { /* An operation completed successfully. Try another device if * necessary, otherwise mark the operation as done. */ advance_op_topology(op, device, msg, rc); return rc; } else { /* This device failed, time to try another topology level. If no other * levels are available, mark this operation as failed and report results. */ if (stonith_topology_next(op) != pcmk_ok) { op->state = st_failed; remote_op_done(op, msg, rc, FALSE); return rc; } } } else if (rc == pcmk_ok && op->devices == NULL) { crm_trace("All done for %s", op->target); op->state = st_done; remote_op_done(op, msg, rc, FALSE); return rc; } else if (rc == -ETIME && op->devices == NULL) { /* If the operation timed out don't bother retrying other peers. */ op->state = st_failed; remote_op_done(op, msg, rc, FALSE); return rc; } else { /* fall-through and attempt other fencing action using another peer */ } /* Retry on failure */ crm_trace("Next for %s on behalf of %s@%s (rc was %d)", op->target, op->originator, op->client_name, rc); call_remote_stonith(op, NULL); return rc; } -int -stonith_fence_history(xmlNode * msg, xmlNode ** output) -{ - int rc = 0; - const char *target = NULL; - xmlNode *dev = get_xpath_object("//@" F_STONITH_TARGET, msg, LOG_TRACE); - char *nodename = NULL; - - if (dev) { - int options = 0; - - target = crm_element_value(dev, F_STONITH_TARGET); - crm_element_value_int(msg, F_STONITH_CALLOPTS, &options); - if (target && (options & st_opt_cs_nodeid)) { - int nodeid = crm_atoi(target, NULL); - - nodename = stonith_get_peer_name(nodeid); - if (nodename) { - target = nodename; - } - } - } - - crm_trace("Looking for operations on %s in %p", target, remote_op_list); - - *output = create_xml_node(NULL, F_STONITH_HISTORY_LIST); - if (remote_op_list) { - GHashTableIter iter; - remote_fencing_op_t *op = NULL; - - g_hash_table_iter_init(&iter, remote_op_list); - while (g_hash_table_iter_next(&iter, NULL, (void **)&op)) { - xmlNode *entry = NULL; - - if (target && strcmp(op->target, target) != 0) { - continue; - } - - rc = 0; - crm_trace("Attaching op %s", op->id); - entry = create_xml_node(*output, STONITH_OP_EXEC); - crm_xml_add(entry, F_STONITH_TARGET, op->target); - crm_xml_add(entry, F_STONITH_ACTION, op->action); - crm_xml_add(entry, F_STONITH_ORIGIN, op->originator); - crm_xml_add(entry, F_STONITH_DELEGATE, op->delegate); - crm_xml_add(entry, F_STONITH_CLIENTNAME, op->client_name); - crm_xml_add_int(entry, F_STONITH_DATE, (int) op->completed); - crm_xml_add_int(entry, F_STONITH_STATE, op->state); - } - } - - free(nodename); - return rc; -} - gboolean stonith_check_fence_tolerance(int tolerance, const char *target, const char *action) { GHashTableIter iter; time_t now = time(NULL); remote_fencing_op_t *rop = NULL; - crm_trace("tolerance=%d, remote_op_list=%p", tolerance, remote_op_list); + crm_trace("tolerance=%d, stonith_remote_op_list=%p", tolerance, + stonith_remote_op_list); - if (tolerance <= 0 || !remote_op_list || target == NULL || action == NULL) { + if (tolerance <= 0 || !stonith_remote_op_list || target == NULL || + action == NULL) { return FALSE; } - g_hash_table_iter_init(&iter, remote_op_list); + g_hash_table_iter_init(&iter, stonith_remote_op_list); while (g_hash_table_iter_next(&iter, NULL, (void **)&rop)) { if (strcmp(rop->target, target) != 0) { continue; } else if (rop->state != st_done) { continue; /* We don't have to worry about remapped reboots here * because if state is done, any remapping has been undone */ } else if (strcmp(rop->action, action) != 0) { continue; } else if ((rop->completed + tolerance) < now) { continue; } crm_notice("Target %s was fenced (%s) less than %ds ago by %s on behalf of %s", target, action, tolerance, rop->delegate, rop->originator); return TRUE; } return FALSE; } diff --git a/daemons/fenced/pacemaker-fenced.c b/daemons/fenced/pacemaker-fenced.c index 5ccd83b9aa..0eaf0491f5 100644 --- a/daemons/fenced/pacemaker-fenced.c +++ b/daemons/fenced/pacemaker-fenced.c @@ -1,1493 +1,1497 @@ /* * Copyright 2009-2018 Andrew Beekhof * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include char *stonith_our_uname = NULL; char *stonith_our_uuid = NULL; long stonith_watchdog_timeout_ms = 0; static GMainLoop *mainloop = NULL; gboolean stand_alone = FALSE; static gboolean no_cib_connect = FALSE; static gboolean stonith_shutdown_flag = FALSE; static qb_ipcs_service_t *ipcs = NULL; static xmlNode *local_cib = NULL; GHashTable *known_peer_names = NULL; static cib_t *cib_api = NULL; static void *cib_library = NULL; static void stonith_shutdown(int nsig); static void stonith_cleanup(void); static int32_t st_ipc_accept(qb_ipcs_connection_t * c, uid_t uid, gid_t gid) { if (stonith_shutdown_flag) { crm_info("Ignoring new client [%d] during shutdown", crm_ipcs_client_pid(c)); return -EPERM; } if (crm_client_new(c, uid, gid) == NULL) { return -EIO; } return 0; } static void st_ipc_created(qb_ipcs_connection_t * c) { crm_trace("Connection created for %p", c); } /* Exit code means? */ static int32_t st_ipc_dispatch(qb_ipcs_connection_t * qbc, void *data, size_t size) { uint32_t id = 0; uint32_t flags = 0; int call_options = 0; xmlNode *request = NULL; crm_client_t *c = crm_client_get(qbc); const char *op = NULL; if (c == NULL) { crm_info("Invalid client: %p", qbc); return 0; } request = crm_ipcs_recv(c, data, size, &id, &flags); if (request == NULL) { crm_ipcs_send_ack(c, id, flags, "nack", __FUNCTION__, __LINE__); return 0; } op = crm_element_value(request, F_CRM_TASK); if(safe_str_eq(op, CRM_OP_RM_NODE_CACHE)) { crm_xml_add(request, F_TYPE, T_STONITH_NG); crm_xml_add(request, F_STONITH_OPERATION, op); crm_xml_add(request, F_STONITH_CLIENTID, c->id); crm_xml_add(request, F_STONITH_CLIENTNAME, crm_client_name(c)); crm_xml_add(request, F_STONITH_CLIENTNODE, stonith_our_uname); send_cluster_message(NULL, crm_msg_stonith_ng, request, FALSE); free_xml(request); return 0; } if (c->name == NULL) { const char *value = crm_element_value(request, F_STONITH_CLIENTNAME); if (value == NULL) { value = "unknown"; } c->name = crm_strdup_printf("%s.%u", value, c->pid); } crm_element_value_int(request, F_STONITH_CALLOPTS, &call_options); crm_trace("Flags %u/%u for command %u from %s", flags, call_options, id, crm_client_name(c)); if (is_set(call_options, st_opt_sync_call)) { CRM_ASSERT(flags & crm_ipc_client_response); CRM_LOG_ASSERT(c->request_id == 0); /* This means the client has two synchronous events in-flight */ c->request_id = id; /* Reply only to the last one */ } crm_xml_add(request, F_STONITH_CLIENTID, c->id); crm_xml_add(request, F_STONITH_CLIENTNAME, crm_client_name(c)); crm_xml_add(request, F_STONITH_CLIENTNODE, stonith_our_uname); crm_log_xml_trace(request, "Client[inbound]"); stonith_command(c, id, flags, request, NULL); free_xml(request); return 0; } /* Error code means? */ static int32_t st_ipc_closed(qb_ipcs_connection_t * c) { crm_client_t *client = crm_client_get(c); if (client == NULL) { return 0; } crm_trace("Connection %p closed", c); crm_client_destroy(client); /* 0 means: yes, go ahead and destroy the connection */ return 0; } static void st_ipc_destroy(qb_ipcs_connection_t * c) { crm_trace("Connection %p destroyed", c); st_ipc_closed(c); } static void stonith_peer_callback(xmlNode * msg, void *private_data) { const char *remote_peer = crm_element_value(msg, F_ORIG); const char *op = crm_element_value(msg, F_STONITH_OPERATION); if (crm_str_eq(op, "poke", TRUE)) { return; } crm_log_xml_trace(msg, "Peer[inbound]"); stonith_command(NULL, 0, 0, msg, remote_peer); } #if SUPPORT_COROSYNC static void stonith_peer_ais_callback(cpg_handle_t handle, const struct cpg_name *groupName, uint32_t nodeid, uint32_t pid, void *msg, size_t msg_len) { uint32_t kind = 0; xmlNode *xml = NULL; const char *from = NULL; char *data = pcmk_message_common_cs(handle, nodeid, pid, msg, &kind, &from); if(data == NULL) { return; } if (kind == crm_class_cluster) { xml = string2xml(data); if (xml == NULL) { crm_err("Invalid XML: '%.120s'", data); free(data); return; } crm_xml_add(xml, F_ORIG, from); /* crm_xml_add_int(xml, F_SEQ, wrapper->id); */ stonith_peer_callback(xml, NULL); } free_xml(xml); free(data); return; } static void stonith_peer_cs_destroy(gpointer user_data) { crm_crit("Lost connection to cluster layer, shutting down"); stonith_shutdown(0); } #endif void do_local_reply(xmlNode * notify_src, const char *client_id, gboolean sync_reply, gboolean from_peer) { /* send callback to originating child */ crm_client_t *client_obj = NULL; int local_rc = pcmk_ok; crm_trace("Sending response"); client_obj = crm_client_get_by_id(client_id); crm_trace("Sending callback to request originator"); if (client_obj == NULL) { local_rc = -1; crm_trace("No client to sent the response to. F_STONITH_CLIENTID not set."); } else { int rid = 0; if (sync_reply) { CRM_LOG_ASSERT(client_obj->request_id); rid = client_obj->request_id; client_obj->request_id = 0; crm_trace("Sending response %d to %s %s", rid, client_obj->name, from_peer ? "(originator of delegated request)" : ""); } else { crm_trace("Sending an event to %s %s", client_obj->name, from_peer ? "(originator of delegated request)" : ""); } local_rc = crm_ipcs_send(client_obj, rid, notify_src, sync_reply?crm_ipc_flags_none:crm_ipc_server_event); } if (local_rc < pcmk_ok && client_obj != NULL) { crm_warn("%sSync reply to %s failed: %s", sync_reply ? "" : "A-", client_obj ? client_obj->name : "", pcmk_strerror(local_rc)); } } long long get_stonith_flag(const char *name) { if (safe_str_eq(name, T_STONITH_NOTIFY_FENCE)) { - return 0x01; + return st_callback_notify_fence; } else if (safe_str_eq(name, STONITH_OP_DEVICE_ADD)) { - return 0x04; + return st_callback_device_add; } else if (safe_str_eq(name, STONITH_OP_DEVICE_DEL)) { - return 0x10; + return st_callback_device_del; + + } else if (safe_str_eq(name, T_STONITH_NOTIFY_HISTORY)) { + return st_callback_notify_history; + } - return 0; + return st_callback_unknown; } static void stonith_notify_client(gpointer key, gpointer value, gpointer user_data) { xmlNode *update_msg = user_data; crm_client_t *client = value; const char *type = NULL; CRM_CHECK(client != NULL, return); CRM_CHECK(update_msg != NULL, return); type = crm_element_value(update_msg, F_SUBTYPE); CRM_CHECK(type != NULL, crm_log_xml_err(update_msg, "notify"); return); if (client->ipcs == NULL) { crm_trace("Skipping client with NULL channel"); return; } if (client->options & get_stonith_flag(type)) { int rc = crm_ipcs_send(client, 0, update_msg, crm_ipc_server_event | crm_ipc_server_error); if (rc <= 0) { crm_warn("%s notification of client %s.%.6s failed: %s (%d)", type, crm_client_name(client), client->id, pcmk_strerror(rc), rc); } else { crm_trace("Sent %s notification to client %s.%.6s", type, crm_client_name(client), client->id); } } } void do_stonith_async_timeout_update(const char *client_id, const char *call_id, int timeout) { crm_client_t *client = NULL; xmlNode *notify_data = NULL; if (!timeout || !call_id || !client_id) { return; } client = crm_client_get_by_id(client_id); if (!client) { return; } notify_data = create_xml_node(NULL, T_STONITH_TIMEOUT_VALUE); crm_xml_add(notify_data, F_TYPE, T_STONITH_TIMEOUT_VALUE); crm_xml_add(notify_data, F_STONITH_CALLID, call_id); crm_xml_add_int(notify_data, F_STONITH_TIMEOUT, timeout); crm_trace("timeout update is %d for client %s and call id %s", timeout, client_id, call_id); if (client) { crm_ipcs_send(client, 0, notify_data, crm_ipc_server_event); } free_xml(notify_data); } void do_stonith_notify(int options, const char *type, int result, xmlNode * data) { /* TODO: Standardize the contents of data */ xmlNode *update_msg = create_xml_node(NULL, "notify"); CRM_CHECK(type != NULL,;); crm_xml_add(update_msg, F_TYPE, T_STONITH_NOTIFY); crm_xml_add(update_msg, F_SUBTYPE, type); crm_xml_add(update_msg, F_STONITH_OPERATION, type); crm_xml_add_int(update_msg, F_STONITH_RC, result); if (data != NULL) { add_message_xml(update_msg, F_STONITH_CALLDATA, data); } crm_trace("Notifying clients"); g_hash_table_foreach(client_connections, stonith_notify_client, update_msg); free_xml(update_msg); crm_trace("Notify complete"); } static void do_stonith_notify_config(int options, const char *op, int rc, const char *desc, int active) { xmlNode *notify_data = create_xml_node(NULL, op); CRM_CHECK(notify_data != NULL, return); crm_xml_add(notify_data, F_STONITH_DEVICE, desc); crm_xml_add_int(notify_data, F_STONITH_ACTIVE, active); do_stonith_notify(options, op, rc, notify_data); free_xml(notify_data); } void do_stonith_notify_device(int options, const char *op, int rc, const char *desc) { do_stonith_notify_config(options, op, rc, desc, g_hash_table_size(device_list)); } void do_stonith_notify_level(int options, const char *op, int rc, const char *desc) { do_stonith_notify_config(options, op, rc, desc, g_hash_table_size(topology)); } static void topology_remove_helper(const char *node, int level) { int rc; char *desc = NULL; xmlNode *data = create_xml_node(NULL, XML_TAG_FENCING_LEVEL); crm_xml_add(data, F_STONITH_ORIGIN, __FUNCTION__); crm_xml_add_int(data, XML_ATTR_STONITH_INDEX, level); crm_xml_add(data, XML_ATTR_STONITH_TARGET, node); rc = stonith_level_remove(data, &desc); do_stonith_notify_level(0, STONITH_OP_LEVEL_DEL, rc, desc); free_xml(data); free(desc); } static void remove_cib_device(xmlXPathObjectPtr xpathObj) { int max = numXpathResults(xpathObj), lpc = 0; for (lpc = 0; lpc < max; lpc++) { const char *rsc_id = NULL; const char *standard = NULL; xmlNode *match = getXpathResult(xpathObj, lpc); CRM_LOG_ASSERT(match != NULL); if(match != NULL) { standard = crm_element_value(match, XML_AGENT_ATTR_CLASS); } if (safe_str_neq(standard, PCMK_RESOURCE_CLASS_STONITH)) { continue; } rsc_id = crm_element_value(match, XML_ATTR_ID); stonith_device_remove(rsc_id, TRUE); } } static void handle_topology_change(xmlNode *match, bool remove) { int rc; char *desc = NULL; CRM_CHECK(match != NULL, return); crm_trace("Updating %s", ID(match)); if(remove) { int index = 0; char *key = stonith_level_key(match, -1); crm_element_value_int(match, XML_ATTR_STONITH_INDEX, &index); topology_remove_helper(key, index); free(key); } rc = stonith_level_register(match, &desc); do_stonith_notify_level(0, STONITH_OP_LEVEL_ADD, rc, desc); free(desc); } static void remove_fencing_topology(xmlXPathObjectPtr xpathObj) { int max = numXpathResults(xpathObj), lpc = 0; for (lpc = 0; lpc < max; lpc++) { xmlNode *match = getXpathResult(xpathObj, lpc); CRM_LOG_ASSERT(match != NULL); if (match && crm_element_value(match, XML_DIFF_MARKER)) { /* Deletion */ int index = 0; char *target = stonith_level_key(match, -1); crm_element_value_int(match, XML_ATTR_STONITH_INDEX, &index); if (target == NULL) { crm_err("Invalid fencing target in element %s", ID(match)); } else if (index <= 0) { crm_err("Invalid level for %s in element %s", target, ID(match)); } else { topology_remove_helper(target, index); } /* } else { Deal with modifications during the 'addition' stage */ } } } static void register_fencing_topology(xmlXPathObjectPtr xpathObj) { int max = numXpathResults(xpathObj), lpc = 0; for (lpc = 0; lpc < max; lpc++) { xmlNode *match = getXpathResult(xpathObj, lpc); handle_topology_change(match, TRUE); } } /* Fencing */ static void fencing_topology_init() { xmlXPathObjectPtr xpathObj = NULL; const char *xpath = "//" XML_TAG_FENCING_LEVEL; crm_trace("Full topology refresh"); free_topology_list(); init_topology_list(); /* Grab everything */ xpathObj = xpath_search(local_cib, xpath); register_fencing_topology(xpathObj); freeXpathObject(xpathObj); } #define rsc_name(x) x->clone_name?x->clone_name:x->id /*! * \internal * \brief Check whether our uname is in a resource's allowed node list * * \param[in] rsc Resource to check * * \return Pointer to node object if found, NULL otherwise */ static node_t * our_node_allowed_for(resource_t *rsc) { GHashTableIter iter; node_t *node = NULL; if (rsc && stonith_our_uname) { g_hash_table_iter_init(&iter, rsc->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (void **)&node)) { if (node && strcmp(node->details->uname, stonith_our_uname) == 0) { break; } node = NULL; } } return node; } /*! * \internal * \brief If a resource or any of its children are STONITH devices, update their * definitions given a cluster working set. * * \param[in] rsc Resource to check * \param[in] data_set Cluster working set with device information */ static void cib_device_update(resource_t *rsc, pe_working_set_t *data_set) { node_t *node = NULL; const char *value = NULL; const char *rclass = NULL; node_t *parent = NULL; gboolean remove = TRUE; /* If this is a complex resource, check children rather than this resource itself. * TODO: Mark each installed device and remove if untouched when this process finishes. */ if(rsc->children) { GListPtr gIter = NULL; for (gIter = rsc->children; gIter != NULL; gIter = gIter->next) { cib_device_update(gIter->data, data_set); if(pe_rsc_is_clone(rsc)) { crm_trace("Only processing one copy of the clone %s", rsc->id); break; } } return; } /* We only care about STONITH resources. */ rclass = crm_element_value(rsc->xml, XML_AGENT_ATTR_CLASS); if (safe_str_neq(rclass, PCMK_RESOURCE_CLASS_STONITH)) { return; } /* If this STONITH resource is disabled, just remove it. */ value = g_hash_table_lookup(rsc->meta, XML_RSC_ATTR_TARGET_ROLE); if (safe_str_eq(value, RSC_STOPPED)) { crm_info("Device %s has been disabled", rsc->id); goto update_done; } /* Check whether our node is allowed for this resource (and its parent if in a group) */ node = our_node_allowed_for(rsc); if (rsc->parent && (rsc->parent->variant == pe_group)) { parent = our_node_allowed_for(rsc->parent); } if(node == NULL) { /* Our node is disallowed, so remove the device */ GHashTableIter iter; crm_info("Device %s has been disabled on %s: unknown", rsc->id, stonith_our_uname); g_hash_table_iter_init(&iter, rsc->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (void **)&node)) { crm_trace("Available: %s = %d", node->details->uname, node->weight); } goto update_done; } else if(node->weight < 0 || (parent && parent->weight < 0)) { /* Our node (or its group) is disallowed by score, so remove the device */ char *score = score2char((node->weight < 0) ? node->weight : parent->weight); crm_info("Device %s has been disabled on %s: score=%s", rsc->id, stonith_our_uname, score); free(score); goto update_done; } else { /* Our node is allowed, so update the device information */ xmlNode *data; GHashTableIter gIter; stonith_key_value_t *params = NULL; const char *name = NULL; const char *agent = crm_element_value(rsc->xml, XML_EXPR_ATTR_TYPE); const char *rsc_provides = NULL; crm_debug("Device %s is allowed on %s: score=%d", rsc->id, stonith_our_uname, node->weight); get_rsc_attributes(rsc->parameters, rsc, node, data_set); get_meta_attributes(rsc->meta, rsc, node, data_set); rsc_provides = g_hash_table_lookup(rsc->meta, XML_RSC_ATTR_PROVIDES); g_hash_table_iter_init(&gIter, rsc->parameters); while (g_hash_table_iter_next(&gIter, (gpointer *) & name, (gpointer *) & value)) { if (!name || !value) { continue; } params = stonith_key_value_add(params, name, value); crm_trace(" %s=%s", name, value); } remove = FALSE; data = create_device_registration_xml(rsc_name(rsc), st_namespace_any, agent, params, rsc_provides); stonith_device_register(data, NULL, TRUE); stonith_key_value_freeall(params, 1, 1); free_xml(data); } update_done: if(remove && g_hash_table_lookup(device_list, rsc_name(rsc))) { stonith_device_remove(rsc_name(rsc), TRUE); } } extern xmlNode *do_calculations(pe_working_set_t * data_set, xmlNode * xml_input, crm_time_t * now); /*! * \internal * \brief Update all STONITH device definitions based on current CIB */ static void cib_devices_update(void) { GListPtr gIter = NULL; pe_working_set_t data_set; crm_info("Updating devices to version %s.%s.%s", crm_element_value(local_cib, XML_ATTR_GENERATION_ADMIN), crm_element_value(local_cib, XML_ATTR_GENERATION), crm_element_value(local_cib, XML_ATTR_NUMUPDATES)); set_working_set_defaults(&data_set); data_set.input = local_cib; data_set.now = crm_time_new(NULL); data_set.flags |= pe_flag_quick_location; data_set.localhost = stonith_our_uname; cluster_status(&data_set); do_calculations(&data_set, NULL, NULL); for (gIter = data_set.resources; gIter != NULL; gIter = gIter->next) { cib_device_update(gIter->data, &data_set); } data_set.input = NULL; /* Wasn't a copy */ cleanup_alloc_calculations(&data_set); } static void update_cib_stonith_devices_v2(const char *event, xmlNode * msg) { xmlNode *change = NULL; char *reason = NULL; bool needs_update = FALSE; xmlNode *patchset = get_message_xml(msg, F_CIB_UPDATE_RESULT); for (change = __xml_first_child(patchset); change != NULL; change = __xml_next(change)) { const char *op = crm_element_value(change, XML_DIFF_OP); const char *xpath = crm_element_value(change, XML_DIFF_PATH); const char *shortpath = NULL; if(op == NULL || strcmp(op, "move") == 0) { continue; } else if(safe_str_eq(op, "delete") && strstr(xpath, XML_CIB_TAG_RESOURCE)) { const char *rsc_id = NULL; char *search = NULL; char *mutable = NULL; if (strstr(xpath, XML_TAG_ATTR_SETS)) { needs_update = TRUE; break; } mutable = strdup(xpath); rsc_id = strstr(mutable, "primitive[@id=\'"); if (rsc_id != NULL) { rsc_id += strlen("primitive[@id=\'"); search = strchr(rsc_id, '\''); } if (search != NULL) { *search = 0; stonith_device_remove(rsc_id, TRUE); } else { crm_warn("Ignoring malformed CIB update (resource deletion)"); } free(mutable); } else if(strstr(xpath, XML_CIB_TAG_RESOURCES)) { shortpath = strrchr(xpath, '/'); CRM_ASSERT(shortpath); reason = crm_strdup_printf("%s %s", op, shortpath+1); needs_update = TRUE; break; } else if(strstr(xpath, XML_CIB_TAG_CONSTRAINTS)) { shortpath = strrchr(xpath, '/'); CRM_ASSERT(shortpath); reason = crm_strdup_printf("%s %s", op, shortpath+1); needs_update = TRUE; break; } } if(needs_update) { crm_info("Updating device list from the cib: %s", reason); cib_devices_update(); } else { crm_trace("No updates for device list found in cib"); } free(reason); } static void update_cib_stonith_devices_v1(const char *event, xmlNode * msg) { const char *reason = "none"; gboolean needs_update = FALSE; xmlXPathObjectPtr xpath_obj = NULL; /* process new constraints */ xpath_obj = xpath_search(msg, "//" F_CIB_UPDATE_RESULT "//" XML_CONS_TAG_RSC_LOCATION); if (numXpathResults(xpath_obj) > 0) { int max = numXpathResults(xpath_obj), lpc = 0; /* Safest and simplest to always recompute */ needs_update = TRUE; reason = "new location constraint"; for (lpc = 0; lpc < max; lpc++) { xmlNode *match = getXpathResult(xpath_obj, lpc); crm_log_xml_trace(match, "new constraint"); } } freeXpathObject(xpath_obj); /* process deletions */ xpath_obj = xpath_search(msg, "//" F_CIB_UPDATE_RESULT "//" XML_TAG_DIFF_REMOVED "//" XML_CIB_TAG_RESOURCE); if (numXpathResults(xpath_obj) > 0) { remove_cib_device(xpath_obj); } freeXpathObject(xpath_obj); /* process additions */ xpath_obj = xpath_search(msg, "//" F_CIB_UPDATE_RESULT "//" XML_TAG_DIFF_ADDED "//" XML_CIB_TAG_RESOURCE); if (numXpathResults(xpath_obj) > 0) { int max = numXpathResults(xpath_obj), lpc = 0; for (lpc = 0; lpc < max; lpc++) { const char *rsc_id = NULL; const char *standard = NULL; xmlNode *match = getXpathResult(xpath_obj, lpc); rsc_id = crm_element_value(match, XML_ATTR_ID); standard = crm_element_value(match, XML_AGENT_ATTR_CLASS); if (safe_str_neq(standard, PCMK_RESOURCE_CLASS_STONITH)) { continue; } crm_trace("Fencing resource %s was added or modified", rsc_id); reason = "new resource"; needs_update = TRUE; } } freeXpathObject(xpath_obj); if(needs_update) { crm_info("Updating device list from the cib: %s", reason); cib_devices_update(); } } static void update_cib_stonith_devices(const char *event, xmlNode * msg) { int format = 1; xmlNode *patchset = get_message_xml(msg, F_CIB_UPDATE_RESULT); CRM_ASSERT(patchset); crm_element_value_int(patchset, "format", &format); switch(format) { case 1: update_cib_stonith_devices_v1(event, msg); break; case 2: update_cib_stonith_devices_v2(event, msg); break; default: crm_warn("Unknown patch format: %d", format); } } /* Needs to hold node name + attribute name + attribute value + 75 */ #define XPATH_MAX 512 /*! * \internal * \brief Check whether a node has a specific attribute name/value * * \param[in] node Name of node to check * \param[in] name Name of an attribute to look for * \param[in] value The value the named attribute needs to be set to in order to be considered a match * * \return TRUE if the locally cached CIB has the specified node attribute */ gboolean node_has_attr(const char *node, const char *name, const char *value) { char xpath[XPATH_MAX]; xmlNode *match; int n; CRM_CHECK(local_cib != NULL, return FALSE); /* Search for the node's attributes in the CIB. While the schema allows * multiple sets of instance attributes, and allows instance attributes to * use id-ref to reference values elsewhere, that is intended for resources, * so we ignore that here. */ n = snprintf(xpath, XPATH_MAX, "//" XML_CIB_TAG_NODES "/" XML_CIB_TAG_NODE "[@uname='%s']/" XML_TAG_ATTR_SETS "/" XML_CIB_TAG_NVPAIR "[@name='%s' and @value='%s']", node, name, value); match = get_xpath_object(xpath, local_cib, LOG_TRACE); CRM_CHECK(n < XPATH_MAX, return FALSE); return (match != NULL); } static void update_fencing_topology(const char *event, xmlNode * msg) { int format = 1; const char *xpath; xmlXPathObjectPtr xpathObj = NULL; xmlNode *patchset = get_message_xml(msg, F_CIB_UPDATE_RESULT); CRM_ASSERT(patchset); crm_element_value_int(patchset, "format", &format); if(format == 1) { /* Process deletions (only) */ xpath = "//" F_CIB_UPDATE_RESULT "//" XML_TAG_DIFF_REMOVED "//" XML_TAG_FENCING_LEVEL; xpathObj = xpath_search(msg, xpath); remove_fencing_topology(xpathObj); freeXpathObject(xpathObj); /* Process additions and changes */ xpath = "//" F_CIB_UPDATE_RESULT "//" XML_TAG_DIFF_ADDED "//" XML_TAG_FENCING_LEVEL; xpathObj = xpath_search(msg, xpath); register_fencing_topology(xpathObj); freeXpathObject(xpathObj); } else if(format == 2) { xmlNode *change = NULL; int add[] = { 0, 0, 0 }; int del[] = { 0, 0, 0 }; xml_patch_versions(patchset, add, del); for (change = __xml_first_child(patchset); change != NULL; change = __xml_next(change)) { const char *op = crm_element_value(change, XML_DIFF_OP); const char *xpath = crm_element_value(change, XML_DIFF_PATH); if(op == NULL) { continue; } else if(strstr(xpath, "/" XML_TAG_FENCING_LEVEL) != NULL) { /* Change to a specific entry */ crm_trace("Handling %s operation %d.%d.%d for %s", op, add[0], add[1], add[2], xpath); if(strcmp(op, "move") == 0) { continue; } else if(strcmp(op, "create") == 0) { handle_topology_change(change->children, FALSE); } else if(strcmp(op, "modify") == 0) { xmlNode *match = first_named_child(change, XML_DIFF_RESULT); if(match) { handle_topology_change(match->children, TRUE); } } else if(strcmp(op, "delete") == 0) { /* Nuclear option, all we have is the path and an id... not enough to remove a specific entry */ crm_info("Re-initializing fencing topology after %s operation %d.%d.%d for %s", op, add[0], add[1], add[2], xpath); fencing_topology_init(); return; } } else if (strstr(xpath, "/" XML_TAG_FENCING_TOPOLOGY) != NULL) { /* Change to the topology in general */ crm_info("Re-initializing fencing topology after top-level %s operation %d.%d.%d for %s", op, add[0], add[1], add[2], xpath); fencing_topology_init(); return; } else if (strstr(xpath, "/" XML_CIB_TAG_CONFIGURATION)) { /* Changes to the whole config section, possibly including the topology as a whild */ if(first_named_child(change, XML_TAG_FENCING_TOPOLOGY) == NULL) { crm_trace("Nothing for us in %s operation %d.%d.%d for %s.", op, add[0], add[1], add[2], xpath); } else if(strcmp(op, "delete") == 0 || strcmp(op, "create") == 0) { crm_info("Re-initializing fencing topology after top-level %s operation %d.%d.%d for %s.", op, add[0], add[1], add[2], xpath); fencing_topology_init(); return; } } else { crm_trace("Nothing for us in %s operation %d.%d.%d for %s", op, add[0], add[1], add[2], xpath); } } } else { crm_warn("Unknown patch format: %d", format); } } static bool have_cib_devices = FALSE; static void update_cib_cache_cb(const char *event, xmlNode * msg) { int rc = pcmk_ok; xmlNode *stonith_enabled_xml = NULL; xmlNode *stonith_watchdog_xml = NULL; const char *stonith_enabled_s = NULL; static gboolean stonith_enabled_saved = TRUE; if(!have_cib_devices) { crm_trace("Skipping updates until we get a full dump"); return; } else if(msg == NULL) { crm_trace("Missing %s update", event); return; } /* Maintain a local copy of the CIB so that we have full access * to device definitions, location constraints, and node attributes */ if (local_cib != NULL) { int rc = pcmk_ok; xmlNode *patchset = NULL; crm_element_value_int(msg, F_CIB_RC, &rc); if (rc != pcmk_ok) { return; } patchset = get_message_xml(msg, F_CIB_UPDATE_RESULT); xml_log_patchset(LOG_TRACE, "Config update", patchset); rc = xml_apply_patchset(local_cib, patchset, TRUE); switch (rc) { case pcmk_ok: case -pcmk_err_old_data: break; case -pcmk_err_diff_resync: case -pcmk_err_diff_failed: crm_notice("[%s] Patch aborted: %s (%d)", event, pcmk_strerror(rc), rc); free_xml(local_cib); local_cib = NULL; break; default: crm_warn("[%s] ABORTED: %s (%d)", event, pcmk_strerror(rc), rc); free_xml(local_cib); local_cib = NULL; } } if (local_cib == NULL) { crm_trace("Re-requesting the full cib"); rc = cib_api->cmds->query(cib_api, NULL, &local_cib, cib_scope_local | cib_sync_call); if(rc != pcmk_ok) { crm_err("Couldn't retrieve the CIB: %s (%d)", pcmk_strerror(rc), rc); return; } CRM_ASSERT(local_cib != NULL); stonith_enabled_saved = FALSE; /* Trigger a full refresh below */ } stonith_enabled_xml = get_xpath_object("//nvpair[@name='stonith-enabled']", local_cib, LOG_TRACE); if (stonith_enabled_xml) { stonith_enabled_s = crm_element_value(stonith_enabled_xml, XML_NVPAIR_ATTR_VALUE); } if (stonith_enabled_s == NULL || crm_is_true(stonith_enabled_s)) { long timeout_ms = 0; const char *value = NULL; stonith_watchdog_xml = get_xpath_object("//nvpair[@name='stonith-watchdog-timeout']", local_cib, LOG_TRACE); if (stonith_watchdog_xml) { value = crm_element_value(stonith_watchdog_xml, XML_NVPAIR_ATTR_VALUE); } if(value) { timeout_ms = crm_get_msec(value); } if (timeout_ms < 0) { timeout_ms = crm_auto_watchdog_timeout(); } if(timeout_ms != stonith_watchdog_timeout_ms) { crm_notice("New watchdog timeout %lds (was %lds)", timeout_ms/1000, stonith_watchdog_timeout_ms/1000); stonith_watchdog_timeout_ms = timeout_ms; } } else { stonith_watchdog_timeout_ms = 0; } if (stonith_enabled_s && crm_is_true(stonith_enabled_s) == FALSE) { crm_trace("Ignoring cib updates while stonith is disabled"); stonith_enabled_saved = FALSE; return; } else if (stonith_enabled_saved == FALSE) { crm_info("Updating stonith device and topology lists now that stonith is enabled"); stonith_enabled_saved = TRUE; fencing_topology_init(); cib_devices_update(); } else { update_fencing_topology(event, msg); update_cib_stonith_devices(event, msg); } } static void init_cib_cache_cb(xmlNode * msg, int call_id, int rc, xmlNode * output, void *user_data) { crm_info("Updating device list from the cib: init"); have_cib_devices = TRUE; local_cib = copy_xml(output); fencing_topology_init(); cib_devices_update(); } static void stonith_shutdown(int nsig) { stonith_shutdown_flag = TRUE; crm_info("Terminating with %d clients", crm_hash_table_size(client_connections)); if (mainloop != NULL && g_main_is_running(mainloop)) { g_main_loop_quit(mainloop); } else { stonith_cleanup(); crm_exit(CRM_EX_OK); } } static void cib_connection_destroy(gpointer user_data) { if (stonith_shutdown_flag) { crm_info("Connection to the CIB manager closed"); return; } else { crm_crit("Lost connection to the CIB manager, shutting down"); } if (cib_api) { cib_api->cmds->signoff(cib_api); } stonith_shutdown(0); } static void stonith_cleanup(void) { if (cib_api) { cib_api->cmds->signoff(cib_api); } if (ipcs) { qb_ipcs_destroy(ipcs); } if (known_peer_names != NULL) { g_hash_table_destroy(known_peer_names); known_peer_names = NULL; } crm_peer_destroy(); crm_client_cleanup(); - free_remote_op_list(); + free_stonith_remote_op_list(); free_topology_list(); free_device_list(); free_metadata_cache(); free(stonith_our_uname); stonith_our_uname = NULL; free_xml(local_cib); local_cib = NULL; } /* *INDENT-OFF* */ static struct crm_option long_options[] = { {"stand-alone", 0, 0, 's'}, {"stand-alone-w-cpg", 0, 0, 'c'}, {"logfile", 1, 0, 'l'}, {"verbose", 0, 0, 'V'}, {"version", 0, 0, '$'}, {"help", 0, 0, '?'}, {0, 0, 0, 0} }; /* *INDENT-ON* */ static void setup_cib(void) { int rc, retries = 0; static cib_t *(*cib_new_fn) (void) = NULL; if (cib_new_fn == NULL) { cib_new_fn = find_library_function(&cib_library, CIB_LIBRARY, "cib_new", TRUE); } if (cib_new_fn != NULL) { cib_api = (*cib_new_fn) (); } if (cib_api == NULL) { crm_err("No connection to the CIB manager"); return; } do { sleep(retries); rc = cib_api->cmds->signon(cib_api, CRM_SYSTEM_STONITHD, cib_command); } while (rc == -ENOTCONN && ++retries < 5); if (rc != pcmk_ok) { crm_err("Could not connect to the CIB manager: %s (%d)", pcmk_strerror(rc), rc); } else if (pcmk_ok != cib_api->cmds->add_notify_callback(cib_api, T_CIB_DIFF_NOTIFY, update_cib_cache_cb)) { crm_err("Could not set CIB notification callback"); } else { rc = cib_api->cmds->query(cib_api, NULL, NULL, cib_scope_local); cib_api->cmds->register_callback(cib_api, rc, 120, FALSE, NULL, "init_cib_cache_cb", init_cib_cache_cb); cib_api->cmds->set_connection_dnotify(cib_api, cib_connection_destroy); crm_info("Watching for stonith topology changes"); } } struct qb_ipcs_service_handlers ipc_callbacks = { .connection_accept = st_ipc_accept, .connection_created = st_ipc_created, .msg_process = st_ipc_dispatch, .connection_closed = st_ipc_closed, .connection_destroyed = st_ipc_destroy }; /*! * \internal * \brief Callback for peer status changes * * \param[in] type What changed * \param[in] node What peer had the change * \param[in] data Previous value of what changed */ static void st_peer_update_callback(enum crm_status_type type, crm_node_t * node, const void *data) { if ((type != crm_status_processes) && !is_set(node->flags, crm_remote_node)) { xmlNode *query = NULL; if (node->id && node->uname) { g_hash_table_insert(known_peer_names, GUINT_TO_POINTER(node->id), strdup(node->uname)); } /* * This is a hack until we can send to a nodeid and/or we fix node name lookups * These messages are ignored in stonith_peer_callback() */ query = create_xml_node(NULL, "stonith_command"); crm_xml_add(query, F_XML_TAGNAME, "stonith_command"); crm_xml_add(query, F_TYPE, T_STONITH_NG); crm_xml_add(query, F_STONITH_OPERATION, "poke"); crm_debug("Broadcasting our uname because of node %u", node->id); send_cluster_message(NULL, crm_msg_stonith_ng, query, FALSE); free_xml(query); } } int main(int argc, char **argv) { int flag; int lpc = 0; int argerr = 0; int option_index = 0; crm_cluster_t cluster; const char *actions[] = { "reboot", "off", "on", "list", "monitor", "status" }; crm_log_preinit(NULL, argc, argv); crm_set_options(NULL, "mode [options]", long_options, "Provides a summary of cluster's current state." "\n\nOutputs varying levels of detail in a number of different formats.\n"); while (1) { flag = crm_get_option(argc, argv, &option_index); if (flag == -1) { break; } switch (flag) { case 'V': crm_bump_log_level(argc, argv); break; case 'l': crm_add_logfile(optarg); break; case 's': stand_alone = TRUE; break; case 'c': stand_alone = FALSE; no_cib_connect = TRUE; break; case '$': case '?': crm_help(flag, CRM_EX_OK); break; default: ++argerr; break; } } if (argc - optind == 1 && safe_str_eq("metadata", argv[optind])) { printf("\n"); printf("\n"); printf(" 1.0\n"); printf(" Instance attributes available for all \"stonith\"-class resources" " and used by Pacemaker's fence daemon, formerly known as stonithd\n"); printf(" Instance attributes available for all \"stonith\"-class resources\n"); printf(" \n"); printf(" \n"); printf (" The priority of the stonith resource. Devices are tried in order of highest priority to lowest.\n"); printf(" \n"); printf(" \n"); printf(" \n", STONITH_ATTR_HOSTARG); printf (" Advanced use only: An alternate parameter to supply instead of 'port'\n"); printf (" Some devices do not support the standard 'port' parameter or may provide additional ones.\n" "Use this to specify an alternate, device-specific, parameter that should indicate the machine to be fenced.\n" "A value of 'none' can be used to tell the cluster not to supply any additional parameters.\n" " \n"); printf(" \n"); printf(" \n"); printf(" \n", STONITH_ATTR_HOSTMAP); printf (" A mapping of host names to ports numbers for devices that do not support host names.\n"); printf (" Eg. node1:1;node2:2,3 would tell the cluster to use port 1 for node1 and ports 2 and 3 for node2\n"); printf(" \n"); printf(" \n"); printf(" \n", STONITH_ATTR_HOSTLIST); printf (" A list of machines controlled by this device (Optional unless %s=static-list).\n", STONITH_ATTR_HOSTCHECK); printf(" \n"); printf(" \n"); printf(" \n", STONITH_ATTR_HOSTCHECK); printf (" How to determine which machines are controlled by the device.\n"); printf (" Allowed values: dynamic-list (query the device), static-list (check the %s attribute), none (assume every device can fence every machine)\n", STONITH_ATTR_HOSTLIST); printf(" \n"); printf(" \n"); printf(" \n", STONITH_ATTR_DELAY_MAX); printf (" Enable a random delay for stonith actions and specify the maximum of random delay.\n"); printf (" This prevents double fencing when using slow devices such as sbd.\n" "Use this to enable a random delay for stonith actions.\n" "The overall delay is derived from this random delay value adding a static delay so that the sum is kept below the maximum delay.\n"); printf(" \n"); printf(" \n"); printf(" \n", STONITH_ATTR_DELAY_BASE); printf (" Enable a base delay for stonith actions and specify base delay value.\n"); printf (" This prevents double fencing when different delays are configured on the nodes.\n" "Use this to enable a static delay for stonith actions.\n" "The overall delay is derived from a random delay value adding this static delay so that the sum is kept below the maximum delay.\n"); printf(" \n"); printf(" \n"); printf(" \n", STONITH_ATTR_ACTION_LIMIT); printf (" The maximum number of actions can be performed in parallel on this device\n"); printf (" Cluster property concurrent-fencing=true needs to be configured first.\n" "Then use this to specify the maximum number of actions can be performed in parallel on this device. -1 is unlimited.\n"); printf(" \n"); printf(" \n"); for (lpc = 0; lpc < DIMOF(actions); lpc++) { printf(" \n", actions[lpc]); printf (" Advanced use only: An alternate command to run instead of '%s'\n", actions[lpc]); printf (" Some devices do not support the standard commands or may provide additional ones.\n" "Use this to specify an alternate, device-specific, command that implements the '%s' action.\n", actions[lpc]); printf(" \n", actions[lpc]); printf(" \n"); printf(" \n", actions[lpc]); printf (" Advanced use only: Specify an alternate timeout to use for %s actions instead of stonith-timeout\n", actions[lpc]); printf (" Some devices need much more/less time to complete than normal.\n" "Use this to specify an alternate, device-specific, timeout for '%s' actions.\n", actions[lpc]); printf(" \n"); printf(" \n"); printf(" \n", actions[lpc]); printf (" Advanced use only: The maximum number of times to retry the '%s' command within the timeout period\n", actions[lpc]); printf(" Some devices do not support multiple connections." " Operations may 'fail' if the device is busy with another task so Pacemaker will automatically retry the operation, if there is time remaining." " Use this option to alter the number of times Pacemaker retries '%s' actions before giving up." "\n", actions[lpc]); printf(" \n"); printf(" \n"); } printf(" \n"); printf("\n"); return CRM_EX_OK; } if (optind != argc) { ++argerr; } if (argerr) { crm_help('?', CRM_EX_USAGE); } crm_log_init(NULL, LOG_INFO, TRUE, FALSE, argc, argv, FALSE); mainloop_add_signal(SIGTERM, stonith_shutdown); crm_peer_init(); known_peer_names = g_hash_table_new_full(g_direct_hash, g_direct_equal, NULL, free); if (stand_alone == FALSE) { if (is_corosync_cluster()) { #if SUPPORT_COROSYNC cluster.destroy = stonith_peer_cs_destroy; cluster.cpg.cpg_deliver_fn = stonith_peer_ais_callback; cluster.cpg.cpg_confchg_fn = pcmk_cpg_membership; #endif } crm_set_status_callback(&st_peer_update_callback); if (crm_cluster_connect(&cluster) == FALSE) { crm_crit("Cannot sign in to the cluster... terminating"); crm_exit(CRM_EX_FATAL); } stonith_our_uname = cluster.uname; stonith_our_uuid = cluster.uuid; if (no_cib_connect == FALSE) { setup_cib(); } } else { stonith_our_uname = strdup("localhost"); } init_device_list(); init_topology_list(); if(stonith_watchdog_timeout_ms > 0) { xmlNode *xml; stonith_key_value_t *params = NULL; params = stonith_key_value_add(params, STONITH_ATTR_HOSTLIST, stonith_our_uname); xml = create_device_registration_xml("watchdog", st_namespace_internal, STONITH_WATCHDOG_AGENT, params, NULL); stonith_device_register(xml, NULL, FALSE); stonith_key_value_freeall(params, 1, 1); free_xml(xml); } stonith_ipc_server_init(&ipcs, &ipc_callbacks); /* Create the mainloop and run it... */ mainloop = g_main_loop_new(NULL, FALSE); crm_info("Starting %s mainloop", crm_system_name); g_main_loop_run(mainloop); stonith_cleanup(); crm_info("Done"); return crm_exit(CRM_EX_OK); } diff --git a/daemons/fenced/pacemaker-fenced.h b/daemons/fenced/pacemaker-fenced.h index 34d82c901c..67ea542310 100644 --- a/daemons/fenced/pacemaker-fenced.h +++ b/daemons/fenced/pacemaker-fenced.h @@ -1,254 +1,266 @@ /* * Copyright 2009-2018 Andrew Beekhof * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include /*! * \internal * \brief Check to see if target was fenced in the last few seconds. * \param tolerance, The number of seconds to look back in time * \param target, The node to search for * \param action, The action we want to match. * * \retval FALSE, not match * \retval TRUE, fencing operation took place in the last 'tolerance' number of seconds. */ gboolean stonith_check_fence_tolerance(int tolerance, const char *target, const char *action); enum st_device_flags { st_device_supports_list = 0x0001, st_device_supports_status = 0x0002, st_device_supports_reboot = 0x0004, }; typedef struct stonith_device_s { char *id; char *agent; char *namespace; /*! list of actions that must execute on the target node. Used for unfencing */ char *on_target_actions; GListPtr targets; time_t targets_age; gboolean has_attr_map; /* should nodeid parameter for victim be included in agent arguments */ gboolean include_nodeid; /* whether the cluster should automatically unfence nodes with the device */ gboolean automatic_unfencing; guint priority; enum st_device_flags flags; GHashTable *params; GHashTable *aliases; GList *pending_ops; crm_trigger_t *work; xmlNode *agent_metadata; /*! A verified device is one that has contacted the * agent successfully to perform a monitor operation */ gboolean verified; gboolean cib_registered; gboolean api_registered; } stonith_device_t; /* These values are used to index certain arrays by "phase". Usually an * operation has only one "phase", so phase is always zero. However, some * reboots are remapped to "off" then "on", in which case "reboot" will be * phase 0, "off" will be phase 1 and "on" will be phase 2. */ enum st_remap_phase { st_phase_requested = 0, st_phase_off = 1, st_phase_on = 2, st_phase_max = 3 }; typedef struct remote_fencing_op_s { /* The unique id associated with this operation */ char *id; /*! The node this operation will fence */ char *target; /*! The fencing action to perform on the target. (reboot, on, off) */ char *action; /*! When was the fencing action recorded (seconds since epoch) */ time_t created; /*! Marks if the final notifications have been sent to local stonith clients. */ gboolean notify_sent; /*! The number of query replies received */ guint replies; /*! The number of query replies expected */ guint replies_expected; /*! Does this node own control of this operation */ gboolean owner; /*! After query is complete, This the high level timer that expires the entire operation */ guint op_timer_total; /*! This timer expires the current fencing request. Many fencing * requests may exist in a single operation */ guint op_timer_one; /*! This timer expires the query request sent out to determine * what nodes are contain what devices, and who those devices can fence */ guint query_timer; /*! This is the default timeout to use for each fencing device if no * custom timeout is received in the query. */ gint base_timeout; /*! This is the calculated total timeout an operation can take before * expiring. This is calculated by adding together all the timeout * values associated with the devices this fencing operation may call */ gint total_timeout; /*! Delegate is the node being asked to perform a fencing action * on behalf of the node that owns the remote operation. Some operations * will involve multiple delegates. This value represents the final delegate * that is used. */ char *delegate; /*! The point at which the remote operation completed */ time_t completed; /*! The stonith_call_options associated with this remote operation */ long long call_options; /*! The current state of the remote operation. This indicates * what stage the op is in, query, exec, done, duplicate, failed. */ enum op_state state; /*! The node that owns the remote operation */ char *originator; /*! The local client id that initiated the fencing request */ char *client_id; /*! The client's call_id that initiated the fencing request */ int client_callid; /*! The name of client that initiated the fencing request */ char *client_name; /*! List of the received query results for all the nodes in the cpg group */ GListPtr query_results; /*! The original request that initiated the remote stonith operation */ xmlNode *request; /*! The current topology level being executed */ guint level; /*! The current operation phase being executed */ enum st_remap_phase phase; /*! Devices with automatic unfencing (always run if "on" requested, never if remapped) */ GListPtr automatic_list; /*! List of all devices at the currently executing topology level */ GListPtr devices_list; /*! Current entry in the topology device list */ GListPtr devices; /*! List of duplicate operations attached to this operation. Once this operation * completes, the duplicate operations will be closed out as well. */ GListPtr duplicates; } remote_fencing_op_t; +enum st_callback_flags { + st_callback_unknown = 0x0000, + st_callback_notify_fence = 0x0001, + st_callback_device_add = 0x0004, + st_callback_device_del = 0x0010, + st_callback_notify_history = 0x0020 +}; + /* * Complex fencing requirements are specified via fencing topologies. * A topology consists of levels; each level is a list of fencing devices. * Topologies are stored in a hash table by node name. When a node needs to be * fenced, if it has an entry in the topology table, the levels are tried * sequentially, and the devices in each level are tried sequentially. * Fencing is considered successful as soon as any level succeeds; * a level is considered successful if all its devices succeed. * Essentially, all devices at a given level are "and-ed" and the * levels are "or-ed". * * This structure is used for the topology table entries. * Topology levels start from 1, so levels[0] is unused and always NULL. */ typedef struct stonith_topology_s { int kind; /*! Node name regex or attribute name=value for which topology applies */ char *target; char *target_value; char *target_pattern; char *target_attribute; /*! Names of fencing devices at each topology level */ GListPtr levels[ST_LEVEL_MAX]; } stonith_topology_t; void init_device_list(void); void free_device_list(void); void init_topology_list(void); void free_topology_list(void); -void free_remote_op_list(void); +void free_stonith_remote_op_list(void); +void init_stonith_remote_op_hash_table(GHashTable **table); void free_metadata_cache(void); long long get_stonith_flag(const char *name); void stonith_command(crm_client_t * client, uint32_t id, uint32_t flags, xmlNode * op_request, const char *remote_peer); int stonith_device_register(xmlNode * msg, const char **desc, gboolean from_cib); int stonith_device_remove(const char *id, gboolean from_cib); char *stonith_level_key(xmlNode * msg, int mode); int stonith_level_kind(xmlNode * msg); int stonith_level_register(xmlNode * msg, char **desc); int stonith_level_remove(xmlNode * msg, char **desc); stonith_topology_t *find_topology_for_host(const char *host); void do_local_reply(xmlNode * notify_src, const char *client_id, gboolean sync_reply, gboolean from_peer); xmlNode *stonith_construct_reply(xmlNode * request, const char *output, xmlNode * data, int rc); void do_stonith_async_timeout_update(const char *client, const char *call_id, int timeout); void do_stonith_notify(int options, const char *type, int result, xmlNode * data); void do_stonith_notify_device(int options, const char *op, int rc, const char *desc); void do_stonith_notify_level(int options, const char *op, int rc, const char *desc); remote_fencing_op_t *initiate_remote_stonith_op(crm_client_t * client, xmlNode * request, gboolean manual_ack); int process_remote_stonith_exec(xmlNode * msg); int process_remote_stonith_query(xmlNode * msg); void *create_remote_stonith_op(const char *client, xmlNode * request, gboolean peer); -int stonith_fence_history(xmlNode * msg, xmlNode ** output); +int stonith_fence_history(xmlNode *msg, xmlNode **output, + const char *remote_peer, int options); bool fencing_peer_active(crm_node_t *peer); int stonith_manual_ack(xmlNode * msg, remote_fencing_op_t * op); gboolean string_in_list(GListPtr list, const char *item); gboolean node_has_attr(const char *node, const char *name, const char *value); void schedule_internal_command(const char *origin, stonith_device_t * device, const char *action, const char *victim, int timeout, void *internal_user_data, void (*done_cb) (GPid pid, int rc, const char *output, gpointer user_data)); char *stonith_get_peer_name(unsigned int nodeid); extern char *stonith_our_uname; extern gboolean stand_alone; extern GHashTable *device_list; extern GHashTable *topology; extern long stonith_watchdog_timeout_ms; extern GHashTable *known_peer_names; + +extern GHashTable *stonith_remote_op_list; diff --git a/include/crm/fencing/internal.h b/include/crm/fencing/internal.h index ad63faf110..0a090f28fa 100644 --- a/include/crm/fencing/internal.h +++ b/include/crm/fencing/internal.h @@ -1,143 +1,144 @@ /* * Copyright 2011-2018 Andrew Beekhof * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #ifndef STONITH_NG_INTERNAL__H # define STONITH_NG_INTERNAL__H # include # include struct stonith_action_s; typedef struct stonith_action_s stonith_action_t; stonith_action_t *stonith_action_create(const char *agent, const char *_action, const char *victim, uint32_t victim_nodeid, int timeout, GHashTable * device_args, GHashTable * port_map); void stonith__destroy_action(stonith_action_t *action); void stonith__action_result(stonith_action_t *action, int *rc, char **output, char **error_output); GPid stonith_action_execute_async(stonith_action_t * action, void *userdata, void (*done) (GPid pid, int rc, const char *output, gpointer user_data)); int stonith__execute(stonith_action_t *action); xmlNode *create_level_registration_xml(const char *node, const char *pattern, const char *attr, const char *value, int level, stonith_key_value_t *device_list); xmlNode *create_device_registration_xml(const char *id, enum stonith_namespace namespace, const char *agent, stonith_key_value_t *params, const char *rsc_provides); # define ST_LEVEL_MAX 10 # define F_STONITH_CLIENTID "st_clientid" # define F_STONITH_CALLOPTS "st_callopt" # define F_STONITH_CALLID "st_callid" # define F_STONITH_CALLDATA "st_calldata" # define F_STONITH_OPERATION "st_op" # define F_STONITH_TARGET "st_target" # define F_STONITH_REMOTE_OP_ID "st_remote_op" # define F_STONITH_RC "st_rc" /*! Timeout period per a device execution */ # define F_STONITH_TIMEOUT "st_timeout" # define F_STONITH_TOLERANCE "st_tolerance" /*! Action specific timeout period returned in query of fencing devices. */ # define F_STONITH_ACTION_TIMEOUT "st_action_timeout" /*! Host in query result is not allowed to run this action */ # define F_STONITH_ACTION_DISALLOWED "st_action_disallowed" /*! Maximum of random fencing delay for a device */ # define F_STONITH_DELAY_MAX "st_delay_max" /*! Base delay used for a fencing delay */ # define F_STONITH_DELAY_BASE "st_delay_base" /*! Has this device been verified using a monitor type * operation (monitor, list, status) */ # define F_STONITH_DEVICE_VERIFIED "st_monitor_verified" /*! device is required for this action */ # define F_STONITH_DEVICE_REQUIRED "st_required" /*! number of available devices in query result */ # define F_STONITH_AVAILABLE_DEVICES "st-available-devices" # define F_STONITH_CALLBACK_TOKEN "st_async_id" # define F_STONITH_CLIENTNAME "st_clientname" # define F_STONITH_CLIENTNODE "st_clientnode" # define F_STONITH_NOTIFY_ACTIVATE "st_notify_activate" # define F_STONITH_NOTIFY_DEACTIVATE "st_notify_deactivate" # define F_STONITH_DELEGATE "st_delegate" /*! The node initiating the stonith operation. If an operation * is relayed, this is the last node the operation lands on. When * in standalone mode, origin is the client's id that originated the * operation. */ # define F_STONITH_ORIGIN "st_origin" # define F_STONITH_HISTORY_LIST "st_history" # define F_STONITH_DATE "st_date" # define F_STONITH_STATE "st_state" # define F_STONITH_ACTIVE "st_active" +# define F_STONITH_DIFFERENTIAL "st_differential" # define F_STONITH_DEVICE "st_device_id" # define F_STONITH_ACTION "st_device_action" # define F_STONITH_MODE "st_mode" # define T_STONITH_NG "stonith-ng" # define T_STONITH_REPLY "st-reply" /*! For async operations, an event from the server containing * the total amount of time the server is allowing for the operation * to take place is returned to the client. */ # define T_STONITH_TIMEOUT_VALUE "st-async-timeout-value" # define T_STONITH_NOTIFY "st_notify" # define STONITH_ATTR_HOSTARG "pcmk_host_argument" # define STONITH_ATTR_HOSTMAP "pcmk_host_map" # define STONITH_ATTR_HOSTLIST "pcmk_host_list" # define STONITH_ATTR_HOSTCHECK "pcmk_host_check" # define STONITH_ATTR_DELAY_MAX "pcmk_delay_max" # define STONITH_ATTR_DELAY_BASE "pcmk_delay_base" # define STONITH_ATTR_ACTION_LIMIT "pcmk_action_limit" # define STONITH_ATTR_ACTION_OP "action" # define STONITH_OP_EXEC "st_execute" # define STONITH_OP_TIMEOUT_UPDATE "st_timeout_update" # define STONITH_OP_QUERY "st_query" # define STONITH_OP_FENCE "st_fence" # define STONITH_OP_RELAY "st_relay" # define STONITH_OP_DEVICE_ADD "st_device_register" # define STONITH_OP_DEVICE_DEL "st_device_remove" # define STONITH_OP_FENCE_HISTORY "st_fence_history" # define STONITH_OP_LEVEL_ADD "st_level_add" # define STONITH_OP_LEVEL_DEL "st_level_remove" # define STONITH_WATCHDOG_AGENT "#watchdog" # ifdef HAVE_STONITH_STONITH_H // utilities from st_lha.c int stonith__list_lha_agents(stonith_key_value_t **devices); int stonith__lha_metadata(const char *agent, int timeout, char **output); bool stonith__agent_is_lha(const char *agent); int stonith__lha_validate(stonith_t *st, int call_options, const char *target, const char *agent, GHashTable *params, int timeout, char **output, char **error_output); # endif // utilities from st_rhcs.c int stonith__list_rhcs_agents(stonith_key_value_t **devices); int stonith__rhcs_metadata(const char *agent, int timeout, char **output); bool stonith__agent_is_rhcs(const char *agent); int stonith__rhcs_validate(stonith_t *st, int call_options, const char *target, const char *agent, GHashTable *params, int timeout, char **output, char **error_output); #endif diff --git a/include/crm/stonith-ng.h b/include/crm/stonith-ng.h index 5ca6d07d47..aa225e2c9d 100644 --- a/include/crm/stonith-ng.h +++ b/include/crm/stonith-ng.h @@ -1,523 +1,528 @@ /* * Copyright 2004-2018 Andrew Beekhof * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #ifdef __cplusplus extern "C" { #endif /** * \file * \brief Fencing aka. STONITH * \ingroup fencing */ #ifndef STONITH_NG__H # define STONITH_NG__H # include # include # include /* TO-DO: Work out how to drop this requirement */ # include # define T_STONITH_NOTIFY_DISCONNECT "st_notify_disconnect" # define T_STONITH_NOTIFY_FENCE "st_notify_fence" +# define T_STONITH_NOTIFY_HISTORY "st_notify_history" /* *INDENT-OFF* */ enum stonith_state { stonith_connected_command, stonith_connected_query, stonith_disconnected, }; enum stonith_call_options { st_opt_none = 0x00000000, st_opt_verbose = 0x00000001, st_opt_allow_suicide = 0x00000002, st_opt_manual_ack = 0x00000008, st_opt_discard_reply = 0x00000010, /* st_opt_all_replies = 0x00000020, */ st_opt_topology = 0x00000040, st_opt_scope_local = 0x00000100, st_opt_cs_nodeid = 0x00000200, st_opt_sync_call = 0x00001000, /*! Allow the timeout period for a callback to be adjusted * based on the time the server reports the operation will take. */ st_opt_timeout_updates = 0x00002000, /*! Only report back if operation is a success in callback */ st_opt_report_only_success = 0x00004000, + /* used where ever apropriate - e.g. cleanup of history */ + st_opt_cleanup = 0x000080000, + /* used where ever apropriate - e.g. send out a history query to all nodes */ + st_opt_broadcast = 0x000100000, }; /*! Order matters here, do not change values */ enum op_state { st_query, st_exec, st_done, st_duplicate, st_failed, }; // Supported fence agent interface standards enum stonith_namespace { st_namespace_invalid, st_namespace_any, st_namespace_internal, // Implemented internally by Pacemaker /* Neither of these projects are active any longer, but the fence agent * interfaces they created are still in use and supported by Pacemaker. */ st_namespace_rhcs, // Red Hat Cluster Suite compatible st_namespace_lha, // Linux-HA compatible }; enum stonith_namespace stonith_text2namespace(const char *namespace_s); const char *stonith_namespace2text(enum stonith_namespace namespace); enum stonith_namespace stonith_get_namespace(const char *agent, const char *namespace_s); typedef struct stonith_key_value_s { char *key; char *value; struct stonith_key_value_s *next; } stonith_key_value_t; typedef struct stonith_history_s { char *target; char *action; char *origin; char *delegate; char *client; int state; time_t completed; struct stonith_history_s *next; } stonith_history_t; typedef struct stonith_s stonith_t; typedef struct stonith_event_s { char *id; char *type; char *message; char *operation; int result; char *origin; char *target; char *action; char *executioner; char *device; /*! The name of the client that initiated the action. */ char *client_origin; } stonith_event_t; typedef struct stonith_callback_data_s { int rc; int call_id; void *userdata; } stonith_callback_data_t; typedef struct stonith_api_operations_s { /*! * \brief Destroy the stonith api structure. */ int (*free) (stonith_t *st); /*! * \brief Connect to the local stonith daemon. * * \retval 0, success * \retval negative error code on failure */ int (*connect) (stonith_t *st, const char *name, int *stonith_fd); /*! * \brief Disconnect from the local stonith daemon. * * \retval 0, success * \retval negative error code on failure */ int (*disconnect)(stonith_t *st); /*! * \brief Remove a registered stonith device with the local stonith daemon. * * \note Synchronous, guaranteed to occur in daemon before function returns. * * \retval 0, success * \retval negative error code on failure */ int (*remove_device)( stonith_t *st, int options, const char *name); /*! * \brief Register a stonith device with the local stonith daemon. * * \note Synchronous, guaranteed to occur in daemon before function returns. * * \retval 0, success * \retval negative error code on failure */ int (*register_device)( stonith_t *st, int options, const char *id, const char *provider, const char *agent, stonith_key_value_t *params); /*! * \brief Remove a fencing level for a specific node. * * \note This feature is not available when stonith is in standalone mode. * * \retval 0, success * \retval negative error code on failure */ int (*remove_level)( stonith_t *st, int options, const char *node, int level); /*! * \brief Register a fencing level containing the fencing devices to be used * at that level for a specific node. * * \note This feature is not available when stonith is in standalone mode. * * \retval 0, success * \retval negative error code on failure */ int (*register_level)( stonith_t *st, int options, const char *node, int level, stonith_key_value_t *device_list); /*! * \brief Get the metadata documentation for a resource. * * \note Value is returned in output. Output must be freed when set. * * \retval 0 success * \retval negative error code on failure */ int (*metadata)(stonith_t *st, int options, const char *device, const char *provider, char **output, int timeout); /*! * \brief Retrieve a list of installed stonith agents * * \note if provider is not provided, all known agents will be returned * \note list must be freed using stonith_key_value_freeall() * \note call_options parameter is not used, it is reserved for future use. * * \retval num items in list on success * \retval negative error code on failure */ int (*list_agents)(stonith_t *stonith, int call_options, const char *provider, stonith_key_value_t **devices, int timeout); /*! * \brief Retrieve string listing hosts and port assignments from a local stonith device. * * \retval 0 on success * \retval negative error code on failure */ int (*list)(stonith_t *st, int options, const char *id, char **list_output, int timeout); /*! * \brief Check to see if a local stonith device is reachable * * \retval 0 on success * \retval negative error code on failure */ int (*monitor)(stonith_t *st, int options, const char *id, int timeout); /*! * \brief Check to see if a local stonith device's port is reachable * * \retval 0 on success * \retval negative error code on failure */ int (*status)(stonith_t *st, int options, const char *id, const char *port, int timeout); /*! * \brief Retrieve a list of registered stonith devices. * * \note If node is provided, only devices that can fence the node id * will be returned. * * \retval num items in list on success * \retval negative error code on failure */ int (*query)(stonith_t *st, int options, const char *node, stonith_key_value_t **devices, int timeout); /*! * \brief Issue a fencing action against a node. * * \note Possible actions are, 'on', 'off', and 'reboot'. * * \param st, stonith connection * \param options, call options * \param node, The target node to fence * \param action, The fencing action to take * \param timeout, The default per device timeout to use with each device * capable of fencing the target. * * \retval 0 success * \retval negative error code on failure. */ int (*fence)(stonith_t *st, int options, const char *node, const char *action, int timeout, int tolerance); /*! * \brief Manually confirm that a node is down. * * \retval 0 success * \retval negative error code on failure. */ int (*confirm)(stonith_t *st, int options, const char *node); /*! * \brief Retrieve a list of fencing operations that have occurred for a specific node. * * \note History is not available in standalone mode. * * \retval 0 success * \retval negative error code on failure. */ int (*history)(stonith_t *st, int options, const char *node, stonith_history_t **output, int timeout); int (*register_notification)( stonith_t *st, const char *event, void (*notify)(stonith_t *st, stonith_event_t *e)); int (*remove_notification)(stonith_t *st, const char *event); /*! * \brief Register a callback to receive the result of an async call id * * \param call_id, The call id to register the callback for. * \param timeout, The default timeout period to wait until this callback expires * \param options, Option flags, st_opt_timeout_updates and st_opt_report_only_success are the * only valid options for this function. * \param userdate, A pointer that will be handed back in the callback. * \param callback_name, Unique name given to callback * \param callback, The callback function * * \retval 0 success * \retval negative error code on failure. */ int (*register_callback)(stonith_t *st, int call_id, int timeout, int options, void *userdata, const char *callback_name, void (*callback)(stonith_t *st, stonith_callback_data_t *data)); /*! * \brief Remove a registered callback for a given call id. */ int (*remove_callback)(stonith_t *st, int call_id, bool all_callbacks); /*! * \brief Remove fencing level for specific node, node regex or attribute * * \param[in] st Fencer connection to use * \param[in] options Bitmask of stonith_call_options to pass to the fencer * \param[in] node If not NULL, target level by this node name * \param[in] pattern If not NULL, target by node name using this regex * \param[in] attr If not NULL, target by this node attribute * \param[in] value If not NULL, target by this node attribute value * \param[in] level Index number of level to remove * * \return 0 on success, negative error code otherwise * * \note This feature is not available when stonith is in standalone mode. * The caller should set only one of node, pattern or attr/value. */ int (*remove_level_full)(stonith_t *st, int options, const char *node, const char *pattern, const char *attr, const char *value, int level); /*! * \brief Register fencing level for specific node, node regex or attribute * * \param[in] st Fencer connection to use * \param[in] options Bitmask of stonith_call_options to pass to fencer * \param[in] node If not NULL, target level by this node name * \param[in] pattern If not NULL, target by node name using this regex * \param[in] attr If not NULL, target by this node attribute * \param[in] value If not NULL, target by this node attribute value * \param[in] level Index number of level to add * \param[in] device_list Devices to use in level * * \return 0 on success, negative error code otherwise * * \note This feature is not available when stonith is in standalone mode. * The caller should set only one of node, pattern or attr/value. */ int (*register_level_full)(stonith_t *st, int options, const char *node, const char *pattern, const char *attr, const char *value, int level, stonith_key_value_t *device_list); /*! * \brief Validate an arbitrary stonith device configuration * * \param[in] st Stonithd connection to use * \param[in] call_options Bitmask of stonith_call_options to use with fencer * \param[in] rsc_id ID used to replace CIB secrets in params * \param[in] namespace_s Namespace of fence agent to validate (optional) * \param[in] agent Fence agent to validate * \param[in] params Configuration parameters to pass to fence agent * \param[in] timeout Fail if no response within this many seconds * \param[out] output If non-NULL, where to store any agent output * \param[out] error_output If non-NULL, where to store agent error output * * \return pcmk_ok if validation succeeds, -errno otherwise * * \note If pcmk_ok is returned, the caller is responsible for freeing * the output (if requested). */ int (*validate)(stonith_t *st, int call_options, const char *rsc_id, const char *namespace_s, const char *agent, stonith_key_value_t *params, int timeout, char **output, char **error_output); } stonith_api_operations_t; struct stonith_s { enum stonith_state state; int call_id; int call_timeout; void *st_private; stonith_api_operations_t *cmds; }; /* *INDENT-ON* */ /* Core functions */ stonith_t *stonith_api_new(void); void stonith_api_delete(stonith_t * st); void stonith_dump_pending_callbacks(stonith_t * st); // deprecated (use stonith_get_namespace() instead) const char *get_stonith_provider(const char *agent, const char *provider); bool stonith_dispatch(stonith_t * st); stonith_key_value_t *stonith_key_value_add(stonith_key_value_t * kvp, const char *key, const char *value); void stonith_key_value_freeall(stonith_key_value_t * kvp, int keys, int values); void stonith_history_free(stonith_history_t *history); /* Basic helpers that allows nodes to be fenced and the history to be * queried without mainloop or the caller understanding the full API * * At least one of nodeid and uname are required */ int stonith_api_kick(uint32_t nodeid, const char *uname, int timeout, bool off); time_t stonith_api_time(uint32_t nodeid, const char *uname, bool in_progress); /* * Helpers for using the above functions without install-time dependencies * * Usage: * #include * * To turn a node off by corosync nodeid: * stonith_api_kick_helper(nodeid, 120, 1); * * To check the last fence date/time (also by nodeid): * last = stonith_api_time_helper(nodeid, 0); * * To check if fencing is in progress: * if(stonith_api_time_helper(nodeid, 1) > 0) { ... } * * eg. #include #include #include int main(int argc, char ** argv) { int rc = 0; int nodeid = 102; rc = stonith_api_time_helper(nodeid, 0); printf("%d last fenced at %s\n", nodeid, ctime(rc)); rc = stonith_api_kick_helper(nodeid, 120, 1); printf("%d fence result: %d\n", nodeid, rc); rc = stonith_api_time_helper(nodeid, 0); printf("%d last fenced at %s\n", nodeid, ctime(rc)); return 0; } */ # define STONITH_LIBRARY "libstonithd.so.26" typedef int (*st_api_kick_fn) (int nodeid, const char *uname, int timeout, bool off); typedef time_t (*st_api_time_fn) (int nodeid, const char *uname, bool in_progress); static inline int stonith_api_kick_helper(uint32_t nodeid, int timeout, bool off) { static void *st_library = NULL; static st_api_kick_fn st_kick_fn; if (st_library == NULL) { st_library = dlopen(STONITH_LIBRARY, RTLD_LAZY); } if (st_library && st_kick_fn == NULL) { st_kick_fn = (st_api_kick_fn) dlsym(st_library, "stonith_api_kick"); } if (st_kick_fn == NULL) { #ifdef ELIBACC return -ELIBACC; #else return -ENOSYS; #endif } return (*st_kick_fn) (nodeid, NULL, timeout, off); } static inline time_t stonith_api_time_helper(uint32_t nodeid, bool in_progress) { static void *st_library = NULL; static st_api_time_fn st_time_fn; if (st_library == NULL) { st_library = dlopen(STONITH_LIBRARY, RTLD_LAZY); } if (st_library && st_time_fn == NULL) { st_time_fn = (st_api_time_fn) dlsym(st_library, "stonith_api_time"); } if (st_time_fn == NULL) { return 0; } return (*st_time_fn) (nodeid, NULL, in_progress); } #ifdef __cplusplus } #endif #endif diff --git a/tools/crm_mon.c b/tools/crm_mon.c index 8f71642a3d..8cc3bd918e 100644 --- a/tools/crm_mon.c +++ b/tools/crm_mon.c @@ -1,4296 +1,4294 @@ /* * Copyright 2004-2018 Andrew Beekhof * * 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 /* crm_ends_with_ext */ #include #include #include #include #include #include #include #include <../lib/pengine/unpack.h> #include #include extern void cleanup_alloc_calculations(pe_working_set_t * data_set); static void clean_up(crm_exit_t exit_code); void crm_diff_update(const char *event, xmlNode * msg); gboolean mon_refresh_display(gpointer user_data); int cib_connect(gboolean full); void mon_st_callback_event(stonith_t * st, stonith_event_t * e); void mon_st_callback_display(stonith_t * st, stonith_event_t * e); void kick_refresh(gboolean data_updated); static char *get_node_display_name(node_t *node); /* * Definitions indicating which items to print */ #define mon_show_times (0x0001U) #define mon_show_stack (0x0002U) #define mon_show_dc (0x0004U) #define mon_show_count (0x0008U) #define mon_show_nodes (0x0010U) #define mon_show_resources (0x0020U) #define mon_show_attributes (0x0040U) #define mon_show_failcounts (0x0080U) #define mon_show_operations (0x0100U) #define mon_show_tickets (0x0200U) #define mon_show_bans (0x0400U) #define mon_show_fence_history (0x0800U) #define mon_show_headers (mon_show_times | mon_show_stack | mon_show_dc \ | mon_show_count) #define mon_show_default (mon_show_headers | mon_show_nodes \ | mon_show_resources) #define mon_show_all (mon_show_default | mon_show_attributes \ | mon_show_failcounts | mon_show_operations \ | mon_show_tickets | mon_show_bans \ | mon_show_fence_history) static unsigned int show = mon_show_default; /* * Definitions indicating how to output */ enum mon_output_format_e { mon_output_none, mon_output_monitor, mon_output_plain, mon_output_console, mon_output_xml, mon_output_html, mon_output_cgi } output_format = mon_output_console; static char *output_filename = NULL; /* if sending output to a file, its name */ /* other globals */ static char *xml_file = NULL; static char *pid_file = NULL; static gboolean group_by_node = FALSE; static gboolean inactive_resources = FALSE; static int reconnect_msec = 5000; static gboolean daemonize = FALSE; static GMainLoop *mainloop = NULL; static guint timer_id = 0; static mainloop_timer_t *refresh_timer = NULL; static GList *attr_list = NULL; static const char *external_agent = NULL; static const char *external_recipient = NULL; static cib_t *cib = NULL; static stonith_t *st = NULL; static xmlNode *current_cib = NULL; static gboolean one_shot = FALSE; static gboolean has_warnings = FALSE; static gboolean print_timing = FALSE; static gboolean watch_fencing = FALSE; static gboolean fence_history = FALSE; static gboolean fence_full_history = FALSE; static gboolean fence_connect = FALSE; -static int fence_history_level = 0; +static int fence_history_level = 1; static gboolean print_brief = FALSE; static gboolean print_pending = TRUE; static gboolean print_clone_detail = FALSE; /* FIXME allow, detect, and correctly interpret glob pattern or regex? */ const char *print_neg_location_prefix = ""; /* Never display node attributes whose name starts with one of these prefixes */ #define FILTER_STR { CRM_FAIL_COUNT_PREFIX, CRM_LAST_FAILURE_PREFIX, \ "shutdown", "terminate", "standby", "probe_complete", \ "#", NULL } long last_refresh = 0; crm_trigger_t *refresh_trigger = NULL; /* Define exit codes for monitoring-compatible output */ #define MON_STATUS_WARN CRM_EX_ERROR /* Convenience macro for prettifying output (e.g. "node" vs "nodes") */ #define s_if_plural(i) (((i) == 1)? "" : "s") #if CURSES_ENABLED # define print_dot() if (output_format == mon_output_console) { \ printw("."); \ clrtoeol(); \ refresh(); \ } else { \ fprintf(stdout, "."); \ } #else # define print_dot() fprintf(stdout, "."); #endif #if CURSES_ENABLED # define print_as(fmt, args...) if (output_format == mon_output_console) { \ printw(fmt, ##args); \ clrtoeol(); \ refresh(); \ } else { \ fprintf(stdout, fmt, ##args); \ } #else # define print_as(fmt, args...) fprintf(stdout, fmt, ##args); #endif static void blank_screen(void) { #if CURSES_ENABLED int lpc = 0; for (lpc = 0; lpc < LINES; lpc++) { move(lpc, 0); clrtoeol(); } move(0, 0); refresh(); #endif } static gboolean mon_timer_popped(gpointer data) { int rc = pcmk_ok; #if CURSES_ENABLED if (output_format == mon_output_console) { clear(); refresh(); } #endif if (timer_id > 0) { g_source_remove(timer_id); timer_id = 0; } print_as("Reconnecting...\n"); rc = cib_connect(TRUE); if (rc != pcmk_ok) { timer_id = g_timeout_add(reconnect_msec, mon_timer_popped, NULL); } return FALSE; } static void mon_cib_connection_destroy(gpointer user_data) { print_as("Connection to the cluster-daemons terminated\n"); if (refresh_timer != NULL) { /* we'll trigger a refresh after reconnect */ mainloop_timer_stop(refresh_timer); } if (timer_id) { /* we'll trigger a new reconnect-timeout at the end */ g_source_remove(timer_id); timer_id = 0; } if (st) { /* the client API won't properly reconnect notifications * if they are still in the table - so remove them */ st->cmds->remove_notification(st, T_STONITH_NOTIFY_DISCONNECT); st->cmds->remove_notification(st, T_STONITH_NOTIFY_FENCE); + st->cmds->remove_notification(st, T_STONITH_NOTIFY_HISTORY); if (st->state != stonith_disconnected) { st->cmds->disconnect(st); } } if (cib) { cib->cmds->signoff(cib); timer_id = g_timeout_add(reconnect_msec, mon_timer_popped, NULL); } return; } /* * Mainloop signal handler. */ static void mon_shutdown(int nsig) { clean_up(CRM_EX_OK); } #if ON_DARWIN # define sighandler_t sig_t #endif #if CURSES_ENABLED # ifndef HAVE_SIGHANDLER_T typedef void (*sighandler_t) (int); # endif static sighandler_t ncurses_winch_handler; static void mon_winresize(int nsig) { static int not_done; int lines = 0, cols = 0; if (!not_done++) { if (ncurses_winch_handler) /* the original ncurses WINCH signal handler does the * magic of retrieving the new window size; * otherwise, we'd have to use ioctl or tgetent */ (*ncurses_winch_handler) (SIGWINCH); getmaxyx(stdscr, lines, cols); resizeterm(lines, cols); mainloop_set_trigger(refresh_trigger); } not_done--; } #endif int cib_connect(gboolean full) { int rc = pcmk_ok; static gboolean need_pass = TRUE; CRM_CHECK(cib != NULL, return -EINVAL); if (getenv("CIB_passwd") != NULL) { need_pass = FALSE; } if ((fence_connect) && (st == NULL)) { st = stonith_api_new(); } if ((fence_connect) && (st->state == stonith_disconnected)) { crm_trace("Connecting to stonith"); rc = st->cmds->connect(st, crm_system_name, NULL); if (rc == pcmk_ok) { crm_trace("Setting up stonith callbacks"); if (watch_fencing) { st->cmds->register_notification(st, T_STONITH_NOTIFY_DISCONNECT, mon_st_callback_event); st->cmds->register_notification(st, T_STONITH_NOTIFY_FENCE, mon_st_callback_event); } else { st->cmds->register_notification(st, T_STONITH_NOTIFY_DISCONNECT, mon_st_callback_display); - st->cmds->register_notification(st, T_STONITH_NOTIFY_FENCE, mon_st_callback_display); + st->cmds->register_notification(st, T_STONITH_NOTIFY_HISTORY, mon_st_callback_display); } } } if (cib->state != cib_connected_query && cib->state != cib_connected_command) { crm_trace("Connecting to the CIB"); if ((output_format == mon_output_console) && need_pass && (cib->variant == cib_remote)) { need_pass = FALSE; print_as("Password:"); } rc = cib->cmds->signon(cib, crm_system_name, cib_query); if (rc != pcmk_ok) { return rc; } rc = cib->cmds->query(cib, NULL, ¤t_cib, cib_scope_local | cib_sync_call); if (rc == pcmk_ok) { mon_refresh_display(NULL); } if (rc == pcmk_ok && full) { if (rc == pcmk_ok) { rc = cib->cmds->set_connection_dnotify(cib, mon_cib_connection_destroy); if (rc == -EPROTONOSUPPORT) { print_as ("Notification setup not supported, won't be able to reconnect after failure"); if (output_format == mon_output_console) { sleep(2); } rc = pcmk_ok; } } if (rc == pcmk_ok) { cib->cmds->del_notify_callback(cib, T_CIB_DIFF_NOTIFY, crm_diff_update); rc = cib->cmds->add_notify_callback(cib, T_CIB_DIFF_NOTIFY, crm_diff_update); } if (rc != pcmk_ok) { print_as("Notification setup failed, could not monitor CIB actions"); if (output_format == mon_output_console) { sleep(2); } clean_up(crm_errno2exit(rc)); } } } return rc; } /* *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"}, {"quiet", 0, 0, 'Q', "\tDisplay only essential output" }, {"-spacer-", 1, 0, '-', "\nModes (mutually exclusive):"}, {"as-html", 1, 0, 'h', "\tWrite cluster status to the named html file"}, {"as-xml", 0, 0, 'X', "\t\tWrite cluster status as xml to stdout. This will enable one-shot mode."}, {"web-cgi", 0, 0, 'w', "\t\tWeb mode with output suitable for CGI (preselected when run as *.cgi)"}, {"simple-status", 0, 0, 's', "\tDisplay the cluster status once as a simple one line output (suitable for nagios)"}, {"-spacer-", 1, 0, '-', "\nDisplay Options:"}, {"group-by-node", 0, 0, 'n', "\tGroup resources by node" }, {"inactive", 0, 0, 'r', "\t\tDisplay inactive resources" }, {"failcounts", 0, 0, 'f', "\tDisplay resource fail counts"}, {"operations", 0, 0, 'o', "\tDisplay resource operation history" }, {"timing-details", 0, 0, 't', "\tDisplay resource operation history with timing details" }, {"tickets", 0, 0, 'c', "\t\tDisplay cluster tickets"}, {"watch-fencing", 0, 0, 'W', "\tListen for fencing events. For use with --external-agent"}, {"fence-history", 2, 0, 'm', "Show fence history\n" "\t\t\t\t\t0=off, 1=failures, 2=add successes and pending (default without value),\n" "\t\t\t\t\t3=show full history without reduction to most recent of each flavor"}, {"neg-locations", 2, 0, 'L', "Display negative location constraints [optionally filtered by id prefix]"}, {"show-node-attributes", 0, 0, 'A', "Display node attributes" }, {"hide-headers", 0, 0, 'D', "\tHide all headers" }, {"show-detail", 0, 0, 'R', "\tShow more details (node IDs, individual clone instances)" }, {"brief", 0, 0, 'b', "\t\tBrief output" }, {"pending", 0, 0, 'j', "\t\tDisplay pending state if 'record-pending' is enabled", pcmk_option_hidden}, {"-spacer-", 1, 0, '-', "\nAdditional Options:"}, {"interval", 1, 0, 'i', "\tUpdate frequency in seconds" }, {"one-shot", 0, 0, '1', "\t\tDisplay the cluster status once on the console and exit"}, {"disable-ncurses",0, 0, 'N', "\tDisable the use of ncurses", !CURSES_ENABLED}, {"daemonize", 0, 0, 'd', "\tRun in the background as a daemon"}, {"pid-file", 1, 0, 'p', "\t(Advanced) Daemon pid file location"}, {"external-agent", 1, 0, 'E', "A program to run when resource operations take place."}, {"external-recipient",1, 0, 'e', "A recipient for your program (assuming you want the program to send something to someone)."}, {"xml-file", 1, 0, 'x', NULL, pcmk_option_hidden}, {"-spacer-", 1, 0, '-', "\nExamples:", pcmk_option_paragraph}, {"-spacer-", 1, 0, '-', "Display the cluster status on the console with updates as they occur:", pcmk_option_paragraph}, {"-spacer-", 1, 0, '-', " crm_mon", pcmk_option_example}, {"-spacer-", 1, 0, '-', "Display the cluster status on the console just once then exit:", pcmk_option_paragraph}, {"-spacer-", 1, 0, '-', " crm_mon -1", pcmk_option_example}, {"-spacer-", 1, 0, '-', "Display your cluster status, group resources by node, and include inactive resources in the list:", pcmk_option_paragraph}, {"-spacer-", 1, 0, '-', " crm_mon --group-by-node --inactive", pcmk_option_example}, {"-spacer-", 1, 0, '-', "Start crm_mon as a background daemon and have it write the cluster status to an HTML file:", pcmk_option_paragraph}, {"-spacer-", 1, 0, '-', " crm_mon --daemonize --as-html /path/to/docroot/filename.html", pcmk_option_example}, {"-spacer-", 1, 0, '-', "Start crm_mon and export the current cluster status as xml to stdout, then exit.:", pcmk_option_paragraph}, {"-spacer-", 1, 0, '-', " crm_mon --as-xml", pcmk_option_example}, {NULL, 0, 0, 0} }; /* *INDENT-ON* */ #if CURSES_ENABLED static const char * get_option_desc(char c) { int lpc; for (lpc = 0; long_options[lpc].name != NULL; lpc++) { if (long_options[lpc].name[0] == '-') continue; if (long_options[lpc].val == c) { static char *buf = NULL; const char *rv; char *nl; /* chop off tabs and cut at newline */ free(buf); /* free string from last usage */ buf = strdup(long_options[lpc].desc); rv = buf; /* make a copy to keep buf pointer unaltered for freeing when we come by next time. Like this the result stays valid until the next call. */ while(isspace(rv[0])) { rv++; } nl = strchr(rv, '\n'); if (nl) { *nl = '\0'; } return rv; } } return NULL; } #define print_option_help(option, condition) \ print_as("%c %c: \t%s\n", ((condition)? '*': ' '), option, get_option_desc(option)); static gboolean detect_user_input(GIOChannel *channel, GIOCondition condition, gpointer unused) { int c; gboolean config_mode = FALSE; while (1) { /* Get user input */ c = getchar(); switch (c) { case 'm': if (!fence_history_level) { fence_history = TRUE; fence_connect = TRUE; if (st == NULL) { mon_cib_connection_destroy(NULL); } } show ^= mon_show_fence_history; break; case 'c': show ^= mon_show_tickets; break; case 'f': show ^= mon_show_failcounts; break; case 'n': group_by_node = ! group_by_node; break; case 'o': show ^= mon_show_operations; if ((show & mon_show_operations) == 0) { print_timing = 0; } break; case 'r': inactive_resources = ! inactive_resources; break; case 'R': print_clone_detail = ! print_clone_detail; break; case 't': print_timing = ! print_timing; if (print_timing) { show |= mon_show_operations; } break; case 'A': show ^= mon_show_attributes; break; case 'L': show ^= mon_show_bans; break; case 'D': /* If any header is shown, clear them all, otherwise set them all */ if (show & mon_show_headers) { show &= ~mon_show_headers; } else { show |= mon_show_headers; } break; case 'b': print_brief = ! print_brief; break; case 'j': print_pending = ! print_pending; break; case '?': config_mode = TRUE; break; default: goto refresh; } if (!config_mode) goto refresh; blank_screen(); print_as("Display option change mode\n"); print_as("\n"); print_option_help('c', show & mon_show_tickets); print_option_help('f', show & mon_show_failcounts); print_option_help('n', group_by_node); print_option_help('o', show & mon_show_operations); print_option_help('r', inactive_resources); print_option_help('t', print_timing); print_option_help('A', show & mon_show_attributes); print_option_help('L', show & mon_show_bans); print_option_help('D', (show & mon_show_headers) == 0); print_option_help('R', print_clone_detail); print_option_help('b', print_brief); print_option_help('j', print_pending); print_option_help('m', (show & mon_show_fence_history)); print_as("\n"); print_as("Toggle fields via field letter, type any other key to return"); } refresh: mon_refresh_display(NULL); return TRUE; } #endif int main(int argc, char **argv) { int flag; int argerr = 0; int option_index = 0; pid_file = strdup("/tmp/ClusterMon.pid"); crm_log_cli_init("crm_mon"); crm_set_options(NULL, "mode [options]", long_options, "Provides a summary of cluster's current state." "\n\nOutputs varying levels of detail in a number of different formats.\n"); #if !defined (ON_DARWIN) && !defined (ON_BSD) /* prevent zombies */ signal(SIGCLD, SIG_IGN); #endif if (crm_ends_with_ext(argv[0], ".cgi") == TRUE) { output_format = mon_output_cgi; one_shot = TRUE; } /* to enable stonith-connection when called via some application like pcs * set environment-variable FENCE_HISTORY to desired level * so you don't have to modify this application */ /* fence_history_level = crm_atoi(getenv("FENCE_HISTORY"), "0"); */ while (1) { flag = crm_get_option(argc, argv, &option_index); if (flag == -1) break; switch (flag) { case 'V': crm_bump_log_level(argc, argv); break; case 'Q': show &= ~mon_show_times; break; case 'i': reconnect_msec = crm_get_msec(optarg); break; case 'n': group_by_node = TRUE; break; case 'r': inactive_resources = TRUE; break; case 'W': watch_fencing = TRUE; fence_connect = TRUE; break; case 'm': fence_history_level = crm_atoi(optarg, "2"); break; case 'd': daemonize = TRUE; break; case 't': print_timing = TRUE; show |= mon_show_operations; break; case 'o': show |= mon_show_operations; break; case 'f': show |= mon_show_failcounts; break; case 'A': show |= mon_show_attributes; break; case 'L': show |= mon_show_bans; print_neg_location_prefix = optarg? optarg : ""; break; case 'D': show &= ~mon_show_headers; break; case 'b': print_brief = TRUE; break; case 'j': print_pending = TRUE; break; case 'R': print_clone_detail = TRUE; break; case 'c': show |= mon_show_tickets; break; case 'p': free(pid_file); if(optarg == NULL) { crm_help(flag, CRM_EX_USAGE); } pid_file = strdup(optarg); break; case 'x': if(optarg == NULL) { crm_help(flag, CRM_EX_USAGE); } xml_file = strdup(optarg); one_shot = TRUE; break; case 'h': if(optarg == NULL) { crm_help(flag, CRM_EX_USAGE); } argerr += (output_format != mon_output_console); output_format = mon_output_html; output_filename = strdup(optarg); umask(S_IWGRP | S_IWOTH); break; case 'X': argerr += (output_format != mon_output_console); output_format = mon_output_xml; one_shot = TRUE; break; case 'w': /* do not allow argv[0] and argv[1...] redundancy */ argerr += (output_format != mon_output_console); output_format = mon_output_cgi; one_shot = TRUE; break; case 's': argerr += (output_format != mon_output_console); output_format = mon_output_monitor; one_shot = TRUE; break; case 'E': external_agent = optarg; break; case 'e': external_recipient = optarg; break; case '1': one_shot = TRUE; break; case 'N': if (output_format == mon_output_console) { output_format = mon_output_plain; } break; case '$': case '?': crm_help(flag, CRM_EX_OK); break; default: printf("Argument code 0%o (%c) is not (?yet?) supported\n", flag, flag); ++argerr; break; } } + if (watch_fencing) { + /* don't moan as fence_history_level == 1 is default */ + fence_history_level = 0; + } + switch (fence_history_level) { case 3: fence_full_history = TRUE; /* fall through to next lower level */ case 2: show |= mon_show_fence_history; /* fall through to next lower level */ case 1: fence_history = TRUE; fence_connect = TRUE; break; default: break; } /* Extra sanity checks when in CGI mode */ if (output_format == mon_output_cgi) { argerr += (optind < argc); argerr += (output_filename != NULL); argerr += (xml_file != NULL); argerr += (external_agent != NULL); argerr += (daemonize == TRUE); /* paranoia */ } else if (optind < argc) { printf("non-option ARGV-elements: "); while (optind < argc) printf("%s ", argv[optind++]); printf("\n"); } if (argerr) { if (output_format == mon_output_cgi) { fprintf(stdout, "Content-Type: text/plain\n" "Status: 500\n\n"); return CRM_EX_USAGE; } crm_help('?', CRM_EX_USAGE); } /* XML output always prints everything */ if (output_format == mon_output_xml) { show = mon_show_all; print_timing = TRUE; } if (one_shot) { if (output_format == mon_output_console) { output_format = mon_output_plain; } } else if (daemonize) { if ((output_format == mon_output_console) || (output_format == mon_output_plain)) { output_format = mon_output_none; } crm_enable_stderr(FALSE); if ((output_format != mon_output_html) && (output_format != mon_output_xml) && !external_agent) { printf ("Looks like you forgot to specify one or more of: --as-html, --as-xml, --external-agent\n"); crm_help('?', CRM_EX_USAGE); } crm_make_daemon(crm_system_name, TRUE, pid_file); } else if (output_format == mon_output_console) { #if CURSES_ENABLED initscr(); cbreak(); noecho(); crm_enable_stderr(FALSE); #else one_shot = TRUE; output_format = mon_output_plain; printf("Defaulting to one-shot mode\n"); printf("You need to have curses available at compile time to enable console mode\n"); #endif } crm_info("Starting %s", crm_system_name); if (xml_file != NULL) { current_cib = filename2xml(xml_file); mon_refresh_display(NULL); return CRM_EX_OK; } if (current_cib == NULL) { int rc = pcmk_ok; cib = cib_new(); do { if (!one_shot) { print_as("Attempting connection to the cluster...\n"); } rc = cib_connect(!one_shot); if (one_shot) { break; } else if (rc != pcmk_ok) { sleep(reconnect_msec / 1000); #if CURSES_ENABLED if (output_format == mon_output_console) { clear(); refresh(); } #endif } } while (rc == -ENOTCONN); if (rc != pcmk_ok) { if (output_format == mon_output_monitor) { printf("CLUSTER WARN: Connection to cluster failed: %s\n", pcmk_strerror(rc)); clean_up(MON_STATUS_WARN); } else { print_as("\nConnection to cluster failed: %s\n", pcmk_strerror(rc)); } if (output_format == mon_output_console) { sleep(2); } clean_up(crm_errno2exit(rc)); } } if (one_shot) { return CRM_EX_OK; } mainloop = g_main_loop_new(NULL, FALSE); mainloop_add_signal(SIGTERM, mon_shutdown); mainloop_add_signal(SIGINT, mon_shutdown); #if CURSES_ENABLED if (output_format == mon_output_console) { ncurses_winch_handler = signal(SIGWINCH, mon_winresize); if (ncurses_winch_handler == SIG_DFL || ncurses_winch_handler == SIG_IGN || ncurses_winch_handler == SIG_ERR) ncurses_winch_handler = NULL; g_io_add_watch(g_io_channel_unix_new(STDIN_FILENO), G_IO_IN, detect_user_input, NULL); } #endif refresh_trigger = mainloop_add_trigger(G_PRIORITY_LOW, mon_refresh_display, NULL); g_main_loop_run(mainloop); g_main_destroy(mainloop); crm_info("Exiting %s", crm_system_name); clean_up(CRM_EX_OK); return CRM_EX_OK; // Should never be reached } #define mon_warn(fmt...) do { \ if (!has_warnings) { \ print_as("CLUSTER WARN:"); \ } else { \ print_as(","); \ } \ print_as(fmt); \ has_warnings = TRUE; \ } while(0) static int count_resources(pe_working_set_t * data_set, resource_t * rsc) { int count = 0; GListPtr gIter = NULL; if (rsc == NULL) { gIter = data_set->resources; } else if (rsc->children) { gIter = rsc->children; } else { return is_not_set(rsc->flags, pe_rsc_orphan); } for (; gIter != NULL; gIter = gIter->next) { count += count_resources(data_set, gIter->data); } return count; } /*! * \internal * \brief Print one-line status suitable for use with monitoring software * * \param[in] data_set Working set of CIB state * \param[in] history List of stonith actions * * \note This function's output (and the return code when the program exits) * should conform to https://www.monitoring-plugins.org/doc/guidelines.html */ static void print_simple_status(pe_working_set_t * data_set, stonith_history_t *history) { GListPtr gIter = NULL; int nodes_online = 0; int nodes_standby = 0; int nodes_maintenance = 0; if (data_set->dc_node == NULL) { mon_warn(" No DC"); } for (gIter = data_set->nodes; gIter != NULL; gIter = gIter->next) { node_t *node = (node_t *) gIter->data; if (node->details->standby && node->details->online) { nodes_standby++; } else if (node->details->maintenance && node->details->online) { nodes_maintenance++; } else if (node->details->online) { nodes_online++; } else { mon_warn(" offline node: %s", node->details->uname); } } if (!has_warnings) { int nresources = count_resources(data_set, NULL); print_as("CLUSTER OK: %d node%s online", nodes_online, s_if_plural(nodes_online)); if (nodes_standby > 0) { print_as(", %d standby node%s", nodes_standby, s_if_plural(nodes_standby)); } if (nodes_maintenance > 0) { print_as(", %d maintenance node%s", nodes_maintenance, s_if_plural(nodes_maintenance)); } print_as(", %d resource%s configured", nresources, s_if_plural(nresources)); } print_as("\n"); } /*! * \internal * \brief Print a [name]=[value][units] pair, optionally using time string * * \param[in] stream File stream to display output to * \param[in] name Name to display * \param[in] value Value to display (or NULL to convert time instead) * \param[in] units Units to display (or NULL for no units) * \param[in] epoch_time Epoch time to convert if value is NULL */ static void print_nvpair(FILE *stream, const char *name, const char *value, const char *units, time_t epoch_time) { /* print name= */ switch (output_format) { case mon_output_plain: case mon_output_console: print_as(" %s=", name); break; case mon_output_html: case mon_output_cgi: case mon_output_xml: fprintf(stream, " %s=", name); break; default: break; } /* If we have a value (and optionally units), print it */ if (value) { switch (output_format) { case mon_output_plain: case mon_output_console: print_as("%s%s", value, (units? units : "")); break; case mon_output_html: case mon_output_cgi: fprintf(stream, "%s%s", value, (units? units : "")); break; case mon_output_xml: fprintf(stream, "\"%s%s\"", value, (units? units : "")); break; default: break; } /* Otherwise print user-friendly time string */ } else { static char empty_str[] = ""; char *c, *date_str = asctime(localtime(&epoch_time)); for (c = (date_str != NULL) ? date_str : empty_str; *c != '\0'; ++c) { if (*c == '\n') { *c = '\0'; break; } } switch (output_format) { case mon_output_plain: case mon_output_console: print_as("'%s'", date_str); break; case mon_output_html: case mon_output_cgi: case mon_output_xml: fprintf(stream, "\"%s\"", date_str); break; default: break; } } } /*! * \internal * \brief Print whatever is needed to start a node section * * \param[in] stream File stream to display output to * \param[in] node Node to print */ static void print_node_start(FILE *stream, node_t *node) { char *node_name; switch (output_format) { case mon_output_plain: case mon_output_console: node_name = get_node_display_name(node); print_as("* Node %s:\n", node_name); free(node_name); break; case mon_output_html: case mon_output_cgi: node_name = get_node_display_name(node); fprintf(stream, "

Node: %s

\n
    \n", node_name); free(node_name); break; case mon_output_xml: fprintf(stream, " \n", node->details->uname); break; default: break; } } /*! * \internal * \brief Print whatever is needed to end a node section * * \param[in] stream File stream to display output to */ static void print_node_end(FILE *stream) { switch (output_format) { case mon_output_html: case mon_output_cgi: fprintf(stream, "
\n"); break; case mon_output_xml: fprintf(stream, " \n"); break; default: break; } } /*! * \internal * \brief Print resources section heading appropriate to options * * \param[in] stream File stream to display output to */ static void print_resources_heading(FILE *stream) { const char *heading; if (group_by_node) { /* Active resources have already been printed by node */ heading = (inactive_resources? "Inactive resources" : NULL); } else if (inactive_resources) { heading = "Full list of resources"; } else { heading = "Active resources"; } /* Print section heading */ switch (output_format) { case mon_output_plain: case mon_output_console: print_as("\n%s:\n\n", heading); break; case mon_output_html: case mon_output_cgi: fprintf(stream, "
\n

%s

\n", heading); break; case mon_output_xml: fprintf(stream, " \n"); break; default: break; } } /*! * \internal * \brief Print whatever resource section closing is appropriate * * \param[in] stream File stream to display output to */ static void print_resources_closing(FILE *stream, gboolean printed_heading) { const char *heading; /* What type of resources we did or did not display */ if (group_by_node) { heading = "inactive "; } else if (inactive_resources) { heading = ""; } else { heading = "active "; } switch (output_format) { case mon_output_plain: case mon_output_console: if (!printed_heading) { print_as("\nNo %sresources\n\n", heading); } break; case mon_output_html: case mon_output_cgi: if (!printed_heading) { fprintf(stream, "
\n

No %sresources

\n", heading); } break; case mon_output_xml: fprintf(stream, " %s\n", (printed_heading? "" : "")); break; default: break; } } /*! * \internal * \brief Print whatever resource section(s) are appropriate * * \param[in] stream File stream to display output to * \param[in] data_set Cluster state to display * \param[in] print_opts Bitmask of pe_print_options */ static void print_resources(FILE *stream, pe_working_set_t *data_set, int print_opts) { GListPtr rsc_iter; const char *prefix = NULL; gboolean printed_heading = FALSE; gboolean brief_output = print_brief; /* If we already showed active resources by node, and * we're not showing inactive resources, we have nothing to do */ if (group_by_node && !inactive_resources) { return; } /* XML uses an indent, and ignores brief option for resources */ if (output_format == mon_output_xml) { prefix = " "; brief_output = FALSE; } /* If we haven't already printed resources grouped by node, * and brief output was requested, print resource summary */ if (brief_output && !group_by_node) { print_resources_heading(stream); printed_heading = TRUE; print_rscs_brief(data_set->resources, NULL, print_opts, stream, inactive_resources); } /* For each resource, display it if appropriate */ for (rsc_iter = data_set->resources; rsc_iter != NULL; rsc_iter = rsc_iter->next) { resource_t *rsc = (resource_t *) rsc_iter->data; /* Complex resources may have some sub-resources active and some inactive */ gboolean is_active = rsc->fns->active(rsc, TRUE); gboolean partially_active = rsc->fns->active(rsc, FALSE); /* Skip inactive orphans (deleted but still in CIB) */ if (is_set(rsc->flags, pe_rsc_orphan) && !is_active) { continue; /* Skip active resources if we already displayed them by node */ } else if (group_by_node) { if (is_active) { continue; } /* Skip primitives already counted in a brief summary */ } else if (brief_output && (rsc->variant == pe_native)) { continue; /* Skip resources that aren't at least partially active, * unless we're displaying inactive resources */ } else if (!partially_active && !inactive_resources) { continue; } /* Print this resource */ if (printed_heading == FALSE) { print_resources_heading(stream); printed_heading = TRUE; } rsc->fns->print(rsc, prefix, print_opts, stream); } print_resources_closing(stream, printed_heading); } /*! * \internal * \brief Print heading for resource history * * \param[in] stream File stream to display output to * \param[in] data_set Current state of CIB * \param[in] node Node that ran this resource * \param[in] rsc Resource to print * \param[in] rsc_id ID of resource to print * \param[in] all Whether to print every resource or just failed ones */ static void print_rsc_history_start(FILE *stream, pe_working_set_t *data_set, node_t *node, resource_t *rsc, const char *rsc_id, gboolean all) { time_t last_failure = 0; int failcount = rsc? pe_get_failcount(node, rsc, &last_failure, pe_fc_default, NULL, data_set) : 0; if (!all && !failcount && (last_failure <= 0)) { return; } /* Print resource ID */ switch (output_format) { case mon_output_plain: case mon_output_console: print_as(" %s:", rsc_id); break; case mon_output_html: case mon_output_cgi: fprintf(stream, "
  • %s:", rsc_id); break; case mon_output_xml: fprintf(stream, " 0)) { /* Print migration threshold */ switch (output_format) { case mon_output_plain: case mon_output_console: print_as(" migration-threshold=%d", rsc->migration_threshold); break; case mon_output_html: case mon_output_cgi: fprintf(stream, " migration-threshold=%d", rsc->migration_threshold); break; case mon_output_xml: fprintf(stream, " orphan=\"false\" migration-threshold=\"%d\"", rsc->migration_threshold); break; default: break; } /* Print fail count if any */ if (failcount > 0) { switch (output_format) { case mon_output_plain: case mon_output_console: print_as(" " CRM_FAIL_COUNT_PREFIX "=%d", failcount); break; case mon_output_html: case mon_output_cgi: fprintf(stream, " " CRM_FAIL_COUNT_PREFIX "=%d", failcount); break; case mon_output_xml: fprintf(stream, " " CRM_FAIL_COUNT_PREFIX "=\"%d\"", failcount); break; default: break; } } /* Print last failure time if any */ if (last_failure > 0) { print_nvpair(stream, CRM_LAST_FAILURE_PREFIX, NULL, NULL, last_failure); } } /* End the heading */ switch (output_format) { case mon_output_plain: case mon_output_console: print_as("\n"); break; case mon_output_html: case mon_output_cgi: fprintf(stream, "\n
      \n"); break; case mon_output_xml: fprintf(stream, ">\n"); break; default: break; } } /*! * \internal * \brief Print closing for resource history * * \param[in] stream File stream to display output to */ static void print_rsc_history_end(FILE *stream) { switch (output_format) { case mon_output_html: case mon_output_cgi: fprintf(stream, "
    \n
    • \n"); break; case mon_output_xml: fprintf(stream, " \n"); break; default: break; } } /*! * \internal * \brief Print operation history * * \param[in] stream File stream to display output to * \param[in] data_set Current state of CIB * \param[in] node Node this operation is for * \param[in] xml_op Root of XML tree describing this operation * \param[in] task Task parsed from this operation's XML * \param[in] interval_ms_s Interval parsed from this operation's XML * \param[in] rc Return code parsed from this operation's XML */ static void print_op_history(FILE *stream, pe_working_set_t *data_set, node_t *node, xmlNode *xml_op, const char *task, const char *interval_ms_s, int rc) { const char *value = NULL; const char *call = crm_element_value(xml_op, XML_LRM_ATTR_CALLID); /* Begin the operation description */ switch (output_format) { case mon_output_plain: case mon_output_console: print_as(" + (%s) %s:", call, task); break; case mon_output_html: case mon_output_cgi: fprintf(stream, "
  • (%s) %s:", call, task); break; case mon_output_xml: fprintf(stream, " 0) { print_nvpair(stream, attr, NULL, NULL, int_value); } } attr = XML_RSC_OP_LAST_RUN; value = crm_element_value(xml_op, attr); if (value) { int_value = crm_parse_int(value, NULL); if (int_value > 0) { print_nvpair(stream, attr, NULL, NULL, int_value); } } attr = XML_RSC_OP_T_EXEC; value = crm_element_value(xml_op, attr); if (value) { print_nvpair(stream, attr, value, "ms", 0); } attr = XML_RSC_OP_T_QUEUE; value = crm_element_value(xml_op, attr); if (value) { print_nvpair(stream, attr, value, "ms", 0); } } /* End the operation description */ switch (output_format) { case mon_output_plain: case mon_output_console: print_as(" rc=%d (%s)\n", rc, services_ocf_exitcode_str(rc)); break; case mon_output_html: case mon_output_cgi: fprintf(stream, " rc=%d (%s)
    • \n", rc, services_ocf_exitcode_str(rc)); break; case mon_output_xml: fprintf(stream, " rc=\"%d\" rc_text=\"%s\" />\n", rc, services_ocf_exitcode_str(rc)); break; default: break; } } /*! * \internal * \brief Print resource operation/failure history * * \param[in] stream File stream to display output to * \param[in] data_set Current state of CIB * \param[in] node Node that ran this resource * \param[in] rsc_entry Root of XML tree describing resource status * \param[in] operations Whether to print operations or just failcounts */ static void print_rsc_history(FILE *stream, pe_working_set_t *data_set, node_t *node, xmlNode *rsc_entry, gboolean operations) { GListPtr gIter = NULL; GListPtr op_list = NULL; gboolean printed = FALSE; const char *rsc_id = crm_element_value(rsc_entry, XML_ATTR_ID); resource_t *rsc = pe_find_resource(data_set->resources, rsc_id); xmlNode *rsc_op = NULL; /* If we're not showing operations, just print the resource failure summary */ if (operations == FALSE) { print_rsc_history_start(stream, data_set, node, rsc, rsc_id, FALSE); print_rsc_history_end(stream); return; } /* Create a list of this resource's operations */ for (rsc_op = __xml_first_child(rsc_entry); rsc_op != NULL; rsc_op = __xml_next(rsc_op)) { if (crm_str_eq((const char *)rsc_op->name, XML_LRM_TAG_RSC_OP, TRUE)) { op_list = g_list_append(op_list, rsc_op); } } op_list = g_list_sort(op_list, sort_op_by_callid); /* Print each operation */ for (gIter = op_list; gIter != NULL; gIter = gIter->next) { xmlNode *xml_op = (xmlNode *) gIter->data; const char *task = crm_element_value(xml_op, XML_LRM_ATTR_TASK); const char *interval_ms_s = crm_element_value(xml_op, XML_LRM_ATTR_INTERVAL_MS); const char *op_rc = crm_element_value(xml_op, XML_LRM_ATTR_RC); int rc = crm_parse_int(op_rc, "0"); /* Display 0-interval monitors as "probe" */ if (safe_str_eq(task, CRMD_ACTION_STATUS) && ((interval_ms_s == NULL) || safe_str_eq(interval_ms_s, "0"))) { task = "probe"; } /* Ignore notifies and some probes */ if (safe_str_eq(task, CRMD_ACTION_NOTIFY) || (safe_str_eq(task, "probe") && (rc == 7))) { continue; } /* If this is the first printed operation, print heading for resource */ if (printed == FALSE) { printed = TRUE; print_rsc_history_start(stream, data_set, node, rsc, rsc_id, TRUE); } /* Print the operation */ print_op_history(stream, data_set, node, xml_op, task, interval_ms_s, rc); } /* Free the list we created (no need to free the individual items) */ g_list_free(op_list); /* If we printed anything, close the resource */ if (printed) { print_rsc_history_end(stream); } } /*! * \internal * \brief Print node operation/failure history * * \param[in] stream File stream to display output to * \param[in] data_set Current state of CIB * \param[in] node_state Root of XML tree describing node status * \param[in] operations Whether to print operations or just failcounts */ static void print_node_history(FILE *stream, pe_working_set_t *data_set, xmlNode *node_state, gboolean operations) { node_t *node = pe_find_node_id(data_set->nodes, ID(node_state)); xmlNode *lrm_rsc = NULL; xmlNode *rsc_entry = NULL; if (node && node->details && node->details->online) { print_node_start(stream, node); lrm_rsc = find_xml_node(node_state, XML_CIB_TAG_LRM, FALSE); lrm_rsc = find_xml_node(lrm_rsc, XML_LRM_TAG_RESOURCES, FALSE); /* Print history of each of the node's resources */ for (rsc_entry = __xml_first_child(lrm_rsc); rsc_entry != NULL; rsc_entry = __xml_next(rsc_entry)) { if (crm_str_eq((const char *)rsc_entry->name, XML_LRM_TAG_RESOURCE, TRUE)) { print_rsc_history(stream, data_set, node, rsc_entry, operations); } } print_node_end(stream); } } /*! * \internal * \brief Print extended information about an attribute if appropriate * * \param[in] data_set Working set of CIB state * * \return TRUE if extended information was printed, FALSE otherwise * \note Currently, extended information is only supported for ping/pingd * resources, for which a message will be printed if connectivity is lost * or degraded. */ static gboolean print_attr_msg(FILE *stream, node_t * node, GListPtr rsc_list, const char *attrname, const char *attrvalue) { GListPtr gIter = NULL; for (gIter = rsc_list; gIter != NULL; gIter = gIter->next) { resource_t *rsc = (resource_t *) gIter->data; const char *type = g_hash_table_lookup(rsc->meta, "type"); if (rsc->children != NULL) { if (print_attr_msg(stream, node, rsc->children, attrname, attrvalue)) { return TRUE; } } if (safe_str_eq(type, "ping") || safe_str_eq(type, "pingd")) { const char *name = g_hash_table_lookup(rsc->parameters, "name"); if (name == NULL) { name = "pingd"; } /* To identify the resource with the attribute name. */ if (safe_str_eq(name, attrname)) { int host_list_num = 0; int expected_score = 0; int value = crm_parse_int(attrvalue, "0"); const char *hosts = g_hash_table_lookup(rsc->parameters, "host_list"); const char *multiplier = g_hash_table_lookup(rsc->parameters, "multiplier"); if(hosts) { char **host_list = g_strsplit(hosts, " ", 0); host_list_num = g_strv_length(host_list); g_strfreev(host_list); } /* pingd multiplier is the same as the default value. */ expected_score = host_list_num * crm_parse_int(multiplier, "1"); switch (output_format) { case mon_output_plain: case mon_output_console: if (value <= 0) { print_as("\t: Connectivity is lost"); } else if (value < expected_score) { print_as("\t: Connectivity is degraded (Expected=%d)", expected_score); } break; case mon_output_html: case mon_output_cgi: if (value <= 0) { fprintf(stream, " (connectivity is lost)"); } else if (value < expected_score) { fprintf(stream, " (connectivity is degraded -- expected %d)", expected_score); } break; case mon_output_xml: fprintf(stream, " expected=\"%d\"", expected_score); break; default: break; } return TRUE; } } } return FALSE; } static int compare_attribute(gconstpointer a, gconstpointer b) { int rc; rc = strcmp((const char *)a, (const char *)b); return rc; } static void create_attr_list(gpointer name, gpointer value, gpointer data) { int i; const char *filt_str[] = FILTER_STR; CRM_CHECK(name != NULL, return); /* filtering automatic attributes */ for (i = 0; filt_str[i] != NULL; i++) { if (g_str_has_prefix(name, filt_str[i])) { return; } } attr_list = g_list_insert_sorted(attr_list, name, compare_attribute); } /* structure for passing multiple user data to g_list_foreach() */ struct mon_attr_data { FILE *stream; node_t *node; }; static void print_node_attribute(gpointer name, gpointer user_data) { const char *value = NULL; struct mon_attr_data *data = (struct mon_attr_data *) user_data; value = pe_node_attribute_raw(data->node, name); /* Print attribute name and value */ switch (output_format) { case mon_output_plain: case mon_output_console: print_as(" + %-32s\t: %-10s", (char *)name, value); break; case mon_output_html: case mon_output_cgi: fprintf(data->stream, "
  • %s: %s", (char *)name, value); break; case mon_output_xml: fprintf(data->stream, " stream, data->node, data->node->details->running_rsc, name, value); /* Close out the attribute */ switch (output_format) { case mon_output_plain: case mon_output_console: print_as("\n"); break; case mon_output_html: case mon_output_cgi: fprintf(data->stream, "
    • \n"); break; case mon_output_xml: fprintf(data->stream, " />\n"); break; default: break; } } static void print_node_summary(FILE *stream, pe_working_set_t * data_set, gboolean operations) { xmlNode *node_state = NULL; xmlNode *cib_status = get_object_root(XML_CIB_TAG_STATUS, data_set->input); /* Print heading */ switch (output_format) { case mon_output_plain: case mon_output_console: if (operations) { print_as("\nOperations:\n"); } else { print_as("\nMigration Summary:\n"); } break; case mon_output_html: case mon_output_cgi: if (operations) { fprintf(stream, "
    \n

    Operations

    \n"); } else { fprintf(stream, "
    \n

    Migration Summary

    \n"); } break; case mon_output_xml: fprintf(stream, " \n"); break; default: break; } /* Print each node in the CIB status */ for (node_state = __xml_first_child(cib_status); node_state != NULL; node_state = __xml_next(node_state)) { if (crm_str_eq((const char *)node_state->name, XML_CIB_TAG_STATE, TRUE)) { print_node_history(stream, data_set, node_state, operations); } } /* Close section */ switch (output_format) { case mon_output_xml: fprintf(stream, " \n"); break; default: break; } } static void print_ticket(gpointer name, gpointer value, gpointer data) { ticket_t *ticket = (ticket_t *) value; FILE *stream = (FILE *) data; switch (output_format) { case mon_output_plain: case mon_output_console: print_as("* %s:\t%s%s", ticket->id, (ticket->granted? "granted" : "revoked"), (ticket->standby? " [standby]" : "")); break; case mon_output_html: case mon_output_cgi: fprintf(stream, "
  • %s: %s%s", ticket->id, (ticket->granted? "granted" : "revoked"), (ticket->standby? " [standby]" : "")); break; case mon_output_xml: fprintf(stream, " id, (ticket->granted? "granted" : "revoked"), (ticket->standby? "true" : "false")); break; default: break; } if (ticket->last_granted > -1) { print_nvpair(stdout, "last-granted", NULL, NULL, ticket->last_granted); } switch (output_format) { case mon_output_plain: case mon_output_console: print_as("\n"); break; case mon_output_html: case mon_output_cgi: fprintf(stream, "
    • \n"); break; case mon_output_xml: fprintf(stream, " />\n"); break; default: break; } } static void print_cluster_tickets(FILE *stream, pe_working_set_t * data_set) { /* Print section heading */ switch (output_format) { case mon_output_plain: case mon_output_console: print_as("\nTickets:\n"); break; case mon_output_html: case mon_output_cgi: fprintf(stream, "
    \n

    Tickets

    \n
      \n"); break; case mon_output_xml: fprintf(stream, " \n"); break; default: break; } /* Print each ticket */ g_hash_table_foreach(data_set->tickets, print_ticket, stream); /* Close section */ switch (output_format) { case mon_output_html: case mon_output_cgi: fprintf(stream, "
    \n"); break; case mon_output_xml: fprintf(stream, " \n"); break; default: break; } } /*! * \internal * \brief Return human-friendly string representing node name * * The returned string will be in the format * uname[@hostUname] [(nodeID)] * "@hostUname" will be printed if the node is a guest node. * "(nodeID)" will be printed if the node ID is different from the node uname, * and detailed output has been requested. * * \param[in] node Node to represent * \return Newly allocated string with representation of node name * \note It is the caller's responsibility to free the result with free(). */ static char * get_node_display_name(node_t *node) { char *node_name; const char *node_host = NULL; const char *node_id = NULL; int name_len; CRM_ASSERT((node != NULL) && (node->details != NULL) && (node->details->uname != NULL)); /* Host is displayed only if this is a guest node */ if (is_container_remote_node(node)) { node_t *host_node = pe__current_node(node->details->remote_rsc); if (host_node && host_node->details) { node_host = host_node->details->uname; } if (node_host == NULL) { node_host = ""; /* so we at least get "uname@" to indicate guest */ } } /* Node ID is displayed if different from uname and detail is requested */ if (print_clone_detail && safe_str_neq(node->details->uname, node->details->id)) { node_id = node->details->id; } /* Determine name length */ name_len = strlen(node->details->uname) + 1; if (node_host) { name_len += strlen(node_host) + 1; /* "@node_host" */ } if (node_id) { name_len += strlen(node_id) + 3; /* + " (node_id)" */ } /* Allocate and populate display name */ node_name = malloc(name_len); CRM_ASSERT(node_name != NULL); strcpy(node_name, node->details->uname); if (node_host) { strcat(node_name, "@"); strcat(node_name, node_host); } if (node_id) { strcat(node_name, " ("); strcat(node_name, node_id); strcat(node_name, ")"); } return node_name; } /*! * \internal * \brief Print a negative location constraint * * \param[in] stream File stream to display output to * \param[in] node Node affected by constraint * \param[in] location Constraint to print */ static void print_ban(FILE *stream, node_t *node, rsc_to_node_t *location) { char *node_name = NULL; switch (output_format) { case mon_output_plain: case mon_output_console: node_name = get_node_display_name(node); print_as(" %s\tprevents %s from running %son %s\n", location->id, location->rsc_lh->id, ((location->role_filter == RSC_ROLE_MASTER)? "as Master " : ""), node_name); break; case mon_output_html: case mon_output_cgi: node_name = get_node_display_name(node); fprintf(stream, "
  • %s prevents %s from running %son %s
    • \n", location->id, location->rsc_lh->id, ((location->role_filter == RSC_ROLE_MASTER)? "as Master " : ""), node_name); break; case mon_output_xml: fprintf(stream, " \n", location->id, location->rsc_lh->id, node->details->uname, node->weight, ((location->role_filter == RSC_ROLE_MASTER)? "true" : "false")); break; default: break; } free(node_name); } /*! * \internal * \brief Print section for negative location constraints * * \param[in] stream File stream to display output to * \param[in] data_set Working set corresponding to CIB status to display */ static void print_neg_locations(FILE *stream, pe_working_set_t *data_set) { GListPtr gIter, gIter2; /* Print section heading */ switch (output_format) { case mon_output_plain: case mon_output_console: print_as("\nNegative Location Constraints:\n"); break; case mon_output_html: case mon_output_cgi: fprintf(stream, "
    \n

    Negative Location Constraints

    \n
      \n"); break; case mon_output_xml: fprintf(stream, " \n"); break; default: break; } /* Print each ban */ for (gIter = data_set->placement_constraints; gIter != NULL; gIter = gIter->next) { rsc_to_node_t *location = (rsc_to_node_t *) gIter->data; if (!g_str_has_prefix(location->id, print_neg_location_prefix)) continue; for (gIter2 = location->node_list_rh; gIter2 != NULL; gIter2 = gIter2->next) { node_t *node = (node_t *) gIter2->data; if (node->weight < 0) { print_ban(stream, node, location); } } } /* Close section */ switch (output_format) { case mon_output_cgi: case mon_output_html: fprintf(stream, "
    \n"); break; case mon_output_xml: fprintf(stream, " \n"); break; default: break; } } static void crm_mon_get_parameters(resource_t *rsc, pe_working_set_t * data_set) { get_rsc_attributes(rsc->parameters, rsc, NULL, data_set); crm_trace("Beekhof: unpacked params for %s (%d)", rsc->id, g_hash_table_size(rsc->parameters)); if(rsc->children) { GListPtr gIter = NULL; for (gIter = rsc->children; gIter != NULL; gIter = gIter->next) { crm_mon_get_parameters(gIter->data, data_set); } } } /*! * \internal * \brief Print node attributes section * * \param[in] stream File stream to display output to * \param[in] data_set Working set of CIB state */ static void print_node_attributes(FILE *stream, pe_working_set_t *data_set) { GListPtr gIter = NULL; /* Print section heading */ switch (output_format) { case mon_output_plain: case mon_output_console: print_as("\nNode Attributes:\n"); break; case mon_output_html: case mon_output_cgi: fprintf(stream, "
    \n

    Node Attributes

    \n"); break; case mon_output_xml: fprintf(stream, " \n"); break; default: break; } /* Unpack all resource parameters (it would be more efficient to do this * only when needed for the first time in print_attr_msg()) */ for (gIter = data_set->resources; gIter != NULL; gIter = gIter->next) { crm_mon_get_parameters(gIter->data, data_set); } /* Display each node's attributes */ for (gIter = data_set->nodes; gIter != NULL; gIter = gIter->next) { struct mon_attr_data data; data.stream = stream; data.node = (node_t *) gIter->data; if (data.node && data.node->details && data.node->details->online) { print_node_start(stream, data.node); g_hash_table_foreach(data.node->details->attrs, create_attr_list, NULL); g_list_foreach(attr_list, print_node_attribute, &data); g_list_free(attr_list); attr_list = NULL; print_node_end(stream); } } /* Print section footer */ switch (output_format) { case mon_output_xml: fprintf(stream, " \n"); break; default: break; } } /*! * \internal * \brief Return resource display options corresponding to command-line choices * * \return Bitmask of pe_print_options suitable for resource print functions */ static int get_resource_display_options(void) { int print_opts; /* Determine basic output format */ switch (output_format) { case mon_output_console: print_opts = pe_print_ncurses; break; case mon_output_html: case mon_output_cgi: print_opts = pe_print_html; break; case mon_output_xml: print_opts = pe_print_xml; break; default: print_opts = pe_print_printf; break; } /* Add optional display elements */ if (print_pending) { print_opts |= pe_print_pending; } if (print_clone_detail) { print_opts |= pe_print_clone_details|pe_print_implicit; } if (!inactive_resources) { print_opts |= pe_print_clone_active; } if (print_brief) { print_opts |= pe_print_brief; } return print_opts; } /*! * \internal * \brief Return human-friendly string representing current time * * \return Current time as string (as by ctime() but without newline) on success * or "Could not determine current time" on error * \note The return value points to a statically allocated string which might be * overwritten by subsequent calls to any of the C library date and time functions. */ static const char * crm_now_string(void) { time_t a_time = time(NULL); char *since_epoch = ctime(&a_time); if ((a_time == (time_t) -1) || (since_epoch == NULL)) { return "Could not determine current time"; } since_epoch[strlen(since_epoch) - 1] = EOS; /* trim newline */ return (since_epoch); } /*! * \internal * \brief Print header for cluster summary if needed * * \param[in] stream File stream to display output to */ static void print_cluster_summary_header(FILE *stream) { switch (output_format) { case mon_output_html: case mon_output_cgi: fprintf(stream, "

    Cluster Summary

    \n

    \n"); break; case mon_output_xml: fprintf(stream, "

    \n"); break; default: break; } } /*! * \internal * \brief Print footer for cluster summary if needed * * \param[in] stream File stream to display output to */ static void print_cluster_summary_footer(FILE *stream) { switch (output_format) { case mon_output_cgi: case mon_output_html: fprintf(stream, "

    \n"); break; case mon_output_xml: fprintf(stream, "     \n"); break; default: break; } } /*! * \internal * \brief Print times the display was last updated and CIB last changed * * \param[in] stream File stream to display output to * \param[in] data_set Working set of CIB state */ static void print_cluster_times(FILE *stream, pe_working_set_t *data_set) { const char *last_written = crm_element_value(data_set->input, XML_CIB_ATTR_WRITTEN); const char *user = crm_element_value(data_set->input, XML_ATTR_UPDATE_USER); const char *client = crm_element_value(data_set->input, XML_ATTR_UPDATE_CLIENT); const char *origin = crm_element_value(data_set->input, XML_ATTR_UPDATE_ORIG); switch (output_format) { case mon_output_plain: case mon_output_console: print_as("Last updated: %s", crm_now_string()); print_as((user || client || origin)? "\n" : "\t\t"); print_as("Last change: %s", last_written ? last_written : ""); if (user) { print_as(" by %s", user); } if (client) { print_as(" via %s", client); } if (origin) { print_as(" on %s", origin); } print_as("\n"); break; case mon_output_html: case mon_output_cgi: fprintf(stream, " Last updated: %s
    \n", crm_now_string()); fprintf(stream, " Last change: %s", last_written ? last_written : ""); if (user) { fprintf(stream, " by %s", user); } if (client) { fprintf(stream, " via %s", client); } if (origin) { fprintf(stream, " on %s", origin); } fprintf(stream, "
    \n"); break; case mon_output_xml: fprintf(stream, " \n", crm_now_string()); fprintf(stream, " \n", last_written ? last_written : "", user ? user : "", client ? client : "", origin ? origin : ""); break; default: break; } } /*! * \internal * \brief Print cluster stack * * \param[in] stream File stream to display output to * \param[in] stack_s Stack name */ static void print_cluster_stack(FILE *stream, const char *stack_s) { switch (output_format) { case mon_output_plain: case mon_output_console: print_as("Stack: %s\n", stack_s); break; case mon_output_html: case mon_output_cgi: fprintf(stream, " Stack: %s
    \n", stack_s); break; case mon_output_xml: fprintf(stream, " \n", stack_s); break; default: break; } } /*! * \internal * \brief Print current DC and its version * * \param[in] stream File stream to display output to * \param[in] data_set Working set of CIB state */ static void print_cluster_dc(FILE *stream, pe_working_set_t *data_set) { node_t *dc = data_set->dc_node; xmlNode *dc_version = get_xpath_object("//nvpair[@name='dc-version']", data_set->input, LOG_DEBUG); const char *dc_version_s = dc_version? crm_element_value(dc_version, XML_NVPAIR_ATTR_VALUE) : NULL; const char *quorum = crm_element_value(data_set->input, XML_ATTR_HAVE_QUORUM); char *dc_name = dc? get_node_display_name(dc) : NULL; switch (output_format) { case mon_output_plain: case mon_output_console: print_as("Current DC: "); if (dc) { print_as("%s (version %s) - partition %s quorum\n", dc_name, (dc_version_s? dc_version_s : "unknown"), (crm_is_true(quorum) ? "with" : "WITHOUT")); } else { print_as("NONE\n"); } break; case mon_output_html: case mon_output_cgi: fprintf(stream, " Current DC: "); if (dc) { fprintf(stream, "%s (version %s) - partition %s quorum", dc_name, (dc_version_s? dc_version_s : "unknown"), (crm_is_true(quorum)? "with" : "WITHOUT")); } else { fprintf(stream, "NONE"); } fprintf(stream, "
    \n"); break; case mon_output_xml: fprintf(stream, " details->uname, dc->details->id, (crm_is_true(quorum) ? "true" : "false")); } else { fprintf(stream, "present=\"false\""); } fprintf(stream, " />\n"); break; default: break; } free(dc_name); } /*! * \internal * \brief Print counts of configured nodes and resources * * \param[in] stream File stream to display output to * \param[in] data_set Working set of CIB state * \param[in] stack_s Stack name */ static void print_cluster_counts(FILE *stream, pe_working_set_t *data_set, const char *stack_s) { int nnodes = g_list_length(data_set->nodes); int nresources = count_resources(data_set, NULL); switch (output_format) { case mon_output_plain: case mon_output_console: print_as("\n%d node%s configured\n", nnodes, s_if_plural(nnodes)); print_as("%d resource%s configured", nresources, s_if_plural(nresources)); if(data_set->disabled_resources || data_set->blocked_resources) { print_as(" ("); if (data_set->disabled_resources) { print_as("%d DISABLED", data_set->disabled_resources); } if (data_set->disabled_resources && data_set->blocked_resources) { print_as(", "); } if (data_set->blocked_resources) { print_as("%d BLOCKED from starting due to failure", data_set->blocked_resources); } print_as(")"); } print_as("\n"); break; case mon_output_html: case mon_output_cgi: fprintf(stream, " %d node%s configured
    \n", nnodes, s_if_plural(nnodes)); fprintf(stream, " %d resource%s configured", nresources, s_if_plural(nresources)); if (data_set->disabled_resources || data_set->blocked_resources) { fprintf(stream, " ("); if (data_set->disabled_resources) { fprintf(stream, "%d DISABLED", data_set->disabled_resources); } if (data_set->disabled_resources && data_set->blocked_resources) { fprintf(stream, ", "); } if (data_set->blocked_resources) { fprintf(stream, "%d BLOCKED from starting due to failure", data_set->blocked_resources); } fprintf(stream, ")"); } fprintf(stream, "
    \n"); break; case mon_output_xml: fprintf(stream, " \n", g_list_length(data_set->nodes)); fprintf(stream, " \n", count_resources(data_set, NULL), data_set->disabled_resources, data_set->blocked_resources); break; default: break; } } /*! * \internal * \brief Print cluster-wide options * * \param[in] stream File stream to display output to * \param[in] data_set Working set of CIB state * * \note Currently this is only implemented for HTML and XML output, and * prints only a few options. If there is demand, more could be added. */ static void print_cluster_options(FILE *stream, pe_working_set_t *data_set) { switch (output_format) { case mon_output_plain: case mon_output_console: if (is_set(data_set->flags, pe_flag_maintenance_mode)) { print_as("\n *** Resource management is DISABLED ***"); print_as("\n The cluster will not attempt to start, stop or recover services"); print_as("\n"); } break; case mon_output_html: fprintf(stream, "

    \n

    Config Options

    \n"); fprintf(stream, " \n"); fprintf(stream, " \n", is_set(data_set->flags, pe_flag_stonith_enabled)? "enabled" : "disabled"); fprintf(stream, " \n", is_set(data_set->flags, pe_flag_symmetric_cluster)? "" : "a"); fprintf(stream, " \n"); fprintf(stream, " \n"); fprintf(stream, "
    STONITH of failed nodes%s
    Cluster is%ssymmetric
    No Quorum Policy"); switch (data_set->no_quorum_policy) { case no_quorum_freeze: fprintf(stream, "Freeze resources"); break; case no_quorum_stop: fprintf(stream, "Stop ALL resources"); break; case no_quorum_ignore: fprintf(stream, "Ignore"); break; case no_quorum_suicide: fprintf(stream, "Suicide"); break; } fprintf(stream, "
    Resource management"); if (is_set(data_set->flags, pe_flag_maintenance_mode)) { fprintf(stream, "DISABLED (the cluster will " "not attempt to start, stop or recover services)"); } else { fprintf(stream, "enabled"); } fprintf(stream, "
    \n

    \n"); break; case mon_output_xml: fprintf(stream, " flags, pe_flag_stonith_enabled)? "true" : "false"); fprintf(stream, " symmetric-cluster=\"%s\"", is_set(data_set->flags, pe_flag_symmetric_cluster)? "true" : "false"); fprintf(stream, " no-quorum-policy=\""); switch (data_set->no_quorum_policy) { case no_quorum_freeze: fprintf(stream, "freeze"); break; case no_quorum_stop: fprintf(stream, "stop"); break; case no_quorum_ignore: fprintf(stream, "ignore"); break; case no_quorum_suicide: fprintf(stream, "suicide"); break; } fprintf(stream, "\""); fprintf(stream, " maintenance-mode=\"%s\"", is_set(data_set->flags, pe_flag_maintenance_mode)? "true" : "false"); fprintf(stream, " />\n"); break; default: break; } } /*! * \internal * \brief Get the name of the stack in use (or "unknown" if not available) * * \param[in] data_set Working set of CIB state * * \return String representing stack name */ static const char * get_cluster_stack(pe_working_set_t *data_set) { xmlNode *stack = get_xpath_object("//nvpair[@name='cluster-infrastructure']", data_set->input, LOG_DEBUG); return stack? crm_element_value(stack, XML_NVPAIR_ATTR_VALUE) : "unknown"; } /*! * \internal * \brief Print a summary of cluster-wide information * * \param[in] stream File stream to display output to * \param[in] data_set Working set of CIB state */ static void print_cluster_summary(FILE *stream, pe_working_set_t *data_set) { const char *stack_s = get_cluster_stack(data_set); gboolean header_printed = FALSE; if (show & mon_show_stack) { if (header_printed == FALSE) { print_cluster_summary_header(stream); header_printed = TRUE; } print_cluster_stack(stream, stack_s); } /* Always print DC if none, even if not requested */ if ((data_set->dc_node == NULL) || (show & mon_show_dc)) { if (header_printed == FALSE) { print_cluster_summary_header(stream); header_printed = TRUE; } print_cluster_dc(stream, data_set); } if (show & mon_show_times) { if (header_printed == FALSE) { print_cluster_summary_header(stream); header_printed = TRUE; } print_cluster_times(stream, data_set); } if (is_set(data_set->flags, pe_flag_maintenance_mode) || data_set->disabled_resources || data_set->blocked_resources || is_set(show, mon_show_count)) { if (header_printed == FALSE) { print_cluster_summary_header(stream); header_printed = TRUE; } print_cluster_counts(stream, data_set, stack_s); } /* There is not a separate option for showing cluster options, so show with * stack for now; a separate option could be added if there is demand */ if (show & mon_show_stack) { print_cluster_options(stream, data_set); } if (header_printed) { print_cluster_summary_footer(stream); } } /*! * \internal * \brief Print a failed action * * \param[in] stream File stream to display output to * \param[in] xml_op Root of XML tree describing failed action */ static void print_failed_action(FILE *stream, xmlNode *xml_op) { const char *op_key = crm_element_value(xml_op, XML_LRM_ATTR_TASK_KEY); const char *op_key_attr = "op_key"; const char *last = crm_element_value(xml_op, XML_RSC_OP_LAST_CHANGE); const char *node = crm_element_value(xml_op, XML_ATTR_UNAME); const char *call = crm_element_value(xml_op, XML_LRM_ATTR_CALLID); const char *exit_reason = crm_element_value(xml_op, XML_LRM_ATTR_EXIT_REASON); int rc = crm_parse_int(crm_element_value(xml_op, XML_LRM_ATTR_RC), "0"); int status = crm_parse_int(crm_element_value(xml_op, XML_LRM_ATTR_OPSTATUS), "0"); char *exit_reason_cleaned; /* If no op_key was given, use id instead */ if (op_key == NULL) { op_key = ID(xml_op); op_key_attr = "id"; } /* If no exit reason was given, use "none" */ if (exit_reason == NULL) { exit_reason = "none"; } /* Print common action information */ switch (output_format) { case mon_output_plain: case mon_output_console: print_as("* %s on %s '%s' (%d): call=%s, status=%s, exitreason='%s'", op_key, node, services_ocf_exitcode_str(rc), rc, call, services_lrm_status_str(status), exit_reason); break; case mon_output_html: case mon_output_cgi: fprintf(stream, "

  • %s on %s '%s' (%d): call=%s, status=%s, exitreason='%s'", op_key, node, services_ocf_exitcode_str(rc), rc, call, services_lrm_status_str(status), exit_reason); break; case mon_output_xml: exit_reason_cleaned = crm_xml_escape(exit_reason); fprintf(stream, " \n"); break; case mon_output_xml: fprintf(stream, " />\n"); break; default: break; } } /*! * \internal * \brief Print a section for failed actions * * \param[in] stream File stream to display output to * \param[in] data_set Working set of CIB state */ static void print_failed_actions(FILE *stream, pe_working_set_t *data_set) { xmlNode *xml_op = NULL; /* Print section heading */ switch (output_format) { case mon_output_plain: case mon_output_console: print_as("\nFailed Resource Actions:\n"); break; case mon_output_html: case mon_output_cgi: fprintf(stream, "
    \n

    Failed Resource Actions

    \n
      \n"); break; case mon_output_xml: fprintf(stream, " \n"); break; default: break; } /* Print each failed action */ for (xml_op = __xml_first_child(data_set->failed); xml_op != NULL; xml_op = __xml_next(xml_op)) { print_failed_action(stream, xml_op); } /* End section */ switch (output_format) { case mon_output_html: case mon_output_cgi: fprintf(stream, "
    \n"); break; case mon_output_xml: fprintf(stream, " \n"); break; default: break; } } /*! * \internal * \brief Reduce the stonith-history * for successful actions we keep the last of every action-type & target * for failed actions we record as well who had failed * for actions in progress we keep full track * * \param[in] history List of stonith actions * */ static stonith_history_t * reduce_stonith_history(stonith_history_t *history) { stonith_history_t *new = NULL, *hp, *np, *tmp; for (hp = history; hp; ) { for (np = new; np; np = np->next) { if ((hp->state == st_done) || (hp->state == st_failed)) { /* action not in progress */ if (safe_str_eq(hp->target, np->target) && safe_str_eq(hp->action, np->action) && (hp->state == np->state)) { if ((hp->state == st_done) || safe_str_eq(hp->delegate, np->delegate)) { /* replace or purge */ if (hp->completed < np->completed) { /* purge older hp */ tmp = hp->next; hp->next = NULL; stonith_history_free(hp); hp = tmp; break; } /* damn single linked list */ free(hp->target); free(hp->action); free(np->origin); np->origin = hp->origin; free(np->delegate); np->delegate = hp->delegate; free(np->client); np->client = hp->client; np->completed = hp->completed; tmp = hp; hp = hp->next; free(tmp); break; } } if (np->next) { continue; } } np = 0; /* let outer loop progress hp */ break; } /* simply move hp from history to new */ if (np == NULL) { tmp = hp->next; hp->next = new; new = hp; hp = tmp; } } return new; } /*! * \internal * \brief Sort the stonith-history * sort by competed most current on the top * pending actions lacking a completed-stamp are gathered at the top * * \param[in] history List of stonith actions * */ static stonith_history_t * sort_stonith_history(stonith_history_t *history) { stonith_history_t *new = NULL, *pending = NULL, *hp, *np, *tmp; for (hp = history; hp; ) { tmp = hp->next; if ((hp->state == st_done) || (hp->state == st_failed)) { /* sort into new */ if ((!new) || (hp->completed > new->completed)) { hp->next = new; new = hp; } else { np = new; do { if ((!np->next) || (hp->completed > np->next->completed)) { hp->next = np->next; np->next = hp; break; } np = np->next; } while (1); } } else { /* put into pending */ hp->next = pending; pending = hp; } hp = tmp; } /* pending actions don't have a completed-stamp so make them go front */ if (pending) { stonith_history_t *last_pending = pending; while (last_pending->next) { last_pending = last_pending->next; } last_pending->next = new; new = pending; } return new; } /*! * \internal * \brief Turn stonith action into a better readable string * * \param[in] action Stonith action */ static char * fence_action_str(const char *action) { char *str = NULL; if (action == NULL) { str = strdup("fencing"); } else if (!strcmp(action, "on")) { str = strdup("unfencing"); } else if (!strcmp(action, "off")) { str = strdup("turning off"); } else { str = strdup(action); } return str; } /*! * \internal * \brief Print a stonith action * * \param[in] stream File stream to display output to * \param[in] event stonith event */ static void print_stonith_action(FILE *stream, stonith_history_t *event) { char *action_s = fence_action_str(event->action); char *run_at_s = ctime(&event->completed); if ((run_at_s) && (run_at_s[0] != 0)) { run_at_s[strlen(run_at_s)-1] = 0; /* Overwrite the newline */ } switch(output_format) { case mon_output_xml: fprintf(stream, " target, event->action); switch(event->state) { case st_done: fprintf(stream, " state=\"success\""); break; case st_failed: fprintf(stream, " state=\"failed\""); break; default: fprintf(stream, " state=\"pending\""); } fprintf(stream, " origin=\"%s\" client=\"%s\"", event->origin, event->client); if (event->delegate) { fprintf(stream, " delegate=\"%s\"", event->delegate); } switch(event->state) { case st_done: case st_failed: fprintf(stream, " completed=\"%s\"", run_at_s?run_at_s:""); break; default: fprintf(stream, " state=\"pending\""); } fprintf(stream, " />\n"); break; case mon_output_plain: case mon_output_console: switch(event->state) { case st_done: print_as("* %s of %s successful: delegate=%s, client=%s, origin=%s,\n" " %s='%s'\n", action_s, event->target, event->delegate ? event->delegate : "", event->client, event->origin, ((!fence_full_history) && (output_format != mon_output_xml))? "last-successful":"completed", run_at_s?run_at_s:""); break; case st_failed: print_as("* %s of %s failed: delegate=%s, client=%s, origin=%s,\n" " %s='%s'\n", action_s, event->target, event->delegate ? event->delegate : "", event->client, event->origin, ((!fence_full_history) && (output_format != mon_output_xml))? "last-failed":"completed", run_at_s?run_at_s:""); break; default: print_as("* %s of %s pending: client=%s, origin=%s\n", action_s, event->target, event->client, event->origin); } break; case mon_output_html: case mon_output_cgi: switch(event->state) { case st_done: fprintf(stream, "
  • %s of %s successful: delegate=%s, " "client=%s, origin=%s, %s='%s'
    • \n", action_s, event->target, event->delegate ? event->delegate : "", event->client, event->origin, ((!fence_full_history) && (output_format != mon_output_xml))? "last-successful":"completed", run_at_s?run_at_s:""); break; case st_failed: fprintf(stream, "
  • %s of %s failed: delegate=%s, " "client=%s, origin=%s, %s='%s'
    • \n", action_s, event->target, event->delegate ? event->delegate : "", event->client, event->origin, ((!fence_full_history) && (output_format != mon_output_xml))? "last-failed":"completed", run_at_s?run_at_s:""); break; default: fprintf(stream, "
  • %s of %s pending: client=%s, " "origin=%s
    • \n", action_s, event->target, event->client, event->origin); } break; default: /* no support for fence history for other formats so far */ break; } free(action_s); } /*! * \internal * \brief Print a section for failed stonith actions * * \param[in] stream File stream to display output to * \param[in] history List of stonith actions * */ static void print_failed_stonith_actions(FILE *stream, stonith_history_t *history) { stonith_history_t *hp; for (hp = history; hp; hp = hp->next) { if (hp->state == st_failed) { break; } } if (!hp) { return; } /* Print section heading */ switch (output_format) { /* no need to take care of xml in here as xml gets full * history anyway */ case mon_output_plain: case mon_output_console: print_as("\nFailed Fencing Actions:\n"); break; case mon_output_html: case mon_output_cgi: fprintf(stream, "
    \n

    Failed Fencing Actions

    \n
      \n"); break; default: break; } /* Print each failed stonith action */ for (hp = history; hp; hp = hp->next) { if (hp->state == st_failed) { print_stonith_action(stream, hp); } } /* End section */ switch (output_format) { case mon_output_html: case mon_output_cgi: fprintf(stream, "
    \n"); break; default: break; } } /*! * \internal * \brief Print a section for stonith-history * * \param[in] stream File stream to display output to * \param[in] history List of stonith actions * */ static void print_stonith_history(FILE *stream, stonith_history_t *history) { stonith_history_t *hp; /* Print section heading */ switch (output_format) { case mon_output_plain: case mon_output_console: print_as("\nFencing History:\n"); break; case mon_output_html: case mon_output_cgi: fprintf(stream, "
    \n

    Fencing History

    \n
      \n"); break; case mon_output_xml: fprintf(stream, " \n"); break; default: break; } for (hp = history; hp; hp = hp->next) { if ((hp->state != st_failed) || (output_format == mon_output_xml)) { print_stonith_action(stream, hp); } } /* End section */ switch (output_format) { case mon_output_html: case mon_output_cgi: fprintf(stream, "
    \n"); break; case mon_output_xml: fprintf(stream, " \n"); break; default: break; } } /*! * \internal * \brief Print cluster status to screen * * This uses the global display preferences set by command-line options * to display cluster status in a human-friendly way. * * \param[in] data_set Working set of CIB state * \param[in] stonith_history List of stonith actions */ static void print_status(pe_working_set_t * data_set, stonith_history_t *stonith_history) { GListPtr gIter = NULL; int print_opts = get_resource_display_options(); /* space-separated lists of node names */ char *online_nodes = NULL; char *online_remote_nodes = NULL; char *online_guest_nodes = NULL; char *offline_nodes = NULL; char *offline_remote_nodes = NULL; if (output_format == mon_output_console) { blank_screen(); } print_cluster_summary(stdout, data_set); print_as("\n"); /* Gather node information (and print if in bad state or grouping by node) */ for (gIter = data_set->nodes; gIter != NULL; gIter = gIter->next) { node_t *node = (node_t *) gIter->data; const char *node_mode = NULL; char *node_name = get_node_display_name(node); /* Get node mode */ if (node->details->unclean) { if (node->details->online) { node_mode = "UNCLEAN (online)"; } else if (node->details->pending) { node_mode = "UNCLEAN (pending)"; } else { node_mode = "UNCLEAN (offline)"; } } else if (node->details->pending) { node_mode = "pending"; } else if (node->details->standby_onfail && node->details->online) { node_mode = "standby (on-fail)"; } else if (node->details->standby) { if (node->details->online) { node_mode = "standby"; } else { node_mode = "OFFLINE (standby)"; } } else if (node->details->maintenance) { if (node->details->online) { node_mode = "maintenance"; } else { node_mode = "OFFLINE (maintenance)"; } } else if (node->details->online) { node_mode = "online"; if (group_by_node == FALSE) { if (is_container_remote_node(node)) { online_guest_nodes = add_list_element(online_guest_nodes, node_name); } else if (is_baremetal_remote_node(node)) { online_remote_nodes = add_list_element(online_remote_nodes, node_name); } else { online_nodes = add_list_element(online_nodes, node_name); } free(node_name); continue; } } else { node_mode = "OFFLINE"; if (group_by_node == FALSE) { if (is_baremetal_remote_node(node)) { offline_remote_nodes = add_list_element(offline_remote_nodes, node_name); } else if (is_container_remote_node(node)) { /* ignore offline guest nodes */ } else { offline_nodes = add_list_element(offline_nodes, node_name); } free(node_name); continue; } } /* If we get here, node is in bad state, or we're grouping by node */ /* Print the node name and status */ if (is_container_remote_node(node)) { print_as("Guest"); } else if (is_baremetal_remote_node(node)) { print_as("Remote"); } print_as("Node %s: %s\n", node_name, node_mode); /* If we're grouping by node, print its resources */ if (group_by_node) { if (print_brief) { print_rscs_brief(node->details->running_rsc, "\t", print_opts | pe_print_rsconly, stdout, FALSE); } else { GListPtr gIter2 = NULL; for (gIter2 = node->details->running_rsc; gIter2 != NULL; gIter2 = gIter2->next) { resource_t *rsc = (resource_t *) gIter2->data; rsc->fns->print(rsc, "\t", print_opts | pe_print_rsconly, stdout); } } } free(node_name); } /* If we're not grouping by node, summarize nodes by status */ if (online_nodes) { print_as("Online: [%s ]\n", online_nodes); free(online_nodes); } if (offline_nodes) { print_as("OFFLINE: [%s ]\n", offline_nodes); free(offline_nodes); } if (online_remote_nodes) { print_as("RemoteOnline: [%s ]\n", online_remote_nodes); free(online_remote_nodes); } if (offline_remote_nodes) { print_as("RemoteOFFLINE: [%s ]\n", offline_remote_nodes); free(offline_remote_nodes); } if (online_guest_nodes) { print_as("GuestOnline: [%s ]\n", online_guest_nodes); free(online_guest_nodes); } /* Print resources section, if needed */ print_resources(stdout, data_set, print_opts); /* print Node Attributes section if requested */ if (show & mon_show_attributes) { print_node_attributes(stdout, data_set); } /* If requested, print resource operations (which includes failcounts) * or just failcounts */ if (show & (mon_show_operations | mon_show_failcounts)) { print_node_summary(stdout, data_set, ((show & mon_show_operations)? TRUE : FALSE)); } /* If there were any failed actions, print them */ if (xml_has_children(data_set->failed)) { print_failed_actions(stdout, data_set); } /* Print failed stonith actions */ if (fence_history) { print_failed_stonith_actions(stdout, stonith_history); } /* Print tickets if requested */ if (show & mon_show_tickets) { print_cluster_tickets(stdout, data_set); } /* Print negative location constraints if requested */ if (show & mon_show_bans) { print_neg_locations(stdout, data_set); } /* Print stonith history */ if (fence_history && (show & mon_show_fence_history)) { print_stonith_history(stdout, stonith_history); } #if CURSES_ENABLED if (output_format == mon_output_console) { refresh(); } #endif } /*! * \internal * \brief Print cluster status in XML format * * \param[in] data_set Working set of CIB state */ static void print_xml_status(pe_working_set_t * data_set, stonith_history_t *stonith_history) { FILE *stream = stdout; GListPtr gIter = NULL; int print_opts = get_resource_display_options(); fprintf(stream, "\n"); fprintf(stream, "\n", VERSION); print_cluster_summary(stream, data_set); /*** NODES ***/ fprintf(stream, " \n"); for (gIter = data_set->nodes; gIter != NULL; gIter = gIter->next) { node_t *node = (node_t *) gIter->data; const char *node_type = "unknown"; switch (node->details->type) { case node_member: node_type = "member"; break; case node_remote: node_type = "remote"; break; case node_ping: node_type = "ping"; break; } fprintf(stream, " details->uname); fprintf(stream, "id=\"%s\" ", node->details->id); fprintf(stream, "online=\"%s\" ", node->details->online ? "true" : "false"); fprintf(stream, "standby=\"%s\" ", node->details->standby ? "true" : "false"); fprintf(stream, "standby_onfail=\"%s\" ", node->details->standby_onfail ? "true" : "false"); fprintf(stream, "maintenance=\"%s\" ", node->details->maintenance ? "true" : "false"); fprintf(stream, "pending=\"%s\" ", node->details->pending ? "true" : "false"); fprintf(stream, "unclean=\"%s\" ", node->details->unclean ? "true" : "false"); fprintf(stream, "shutdown=\"%s\" ", node->details->shutdown ? "true" : "false"); fprintf(stream, "expected_up=\"%s\" ", node->details->expected_up ? "true" : "false"); fprintf(stream, "is_dc=\"%s\" ", node->details->is_dc ? "true" : "false"); fprintf(stream, "resources_running=\"%d\" ", g_list_length(node->details->running_rsc)); fprintf(stream, "type=\"%s\" ", node_type); if (is_container_remote_node(node)) { fprintf(stream, "id_as_resource=\"%s\" ", node->details->remote_rsc->container->id); } if (group_by_node) { GListPtr lpc2 = NULL; fprintf(stream, ">\n"); for (lpc2 = node->details->running_rsc; lpc2 != NULL; lpc2 = lpc2->next) { resource_t *rsc = (resource_t *) lpc2->data; rsc->fns->print(rsc, " ", print_opts | pe_print_rsconly, stream); } fprintf(stream, " \n"); } else { fprintf(stream, "/>\n"); } } fprintf(stream, " \n"); /* Print resources section, if needed */ print_resources(stream, data_set, print_opts); /* print Node Attributes section if requested */ if (show & mon_show_attributes) { print_node_attributes(stream, data_set); } /* If requested, print resource operations (which includes failcounts) * or just failcounts */ if (show & (mon_show_operations | mon_show_failcounts)) { print_node_summary(stream, data_set, ((show & mon_show_operations)? TRUE : FALSE)); } /* If there were any failed actions, print them */ if (xml_has_children(data_set->failed)) { print_failed_actions(stream, data_set); } /* Print stonith history */ if (fence_history) { print_stonith_history(stdout, stonith_history); } /* Print tickets if requested */ if (show & mon_show_tickets) { print_cluster_tickets(stream, data_set); } /* Print negative location constraints if requested */ if (show & mon_show_bans) { print_neg_locations(stream, data_set); } fprintf(stream, "\n"); fflush(stream); fclose(stream); } /*! * \internal * \brief Print cluster status in HTML format (with HTTP headers if CGI) * * \param[in] data_set Working set of CIB state * \param[in] filename Name of file to write HTML to (ignored if CGI) * * \return 0 on success, -1 on error */ static int print_html_status(pe_working_set_t * data_set, const char *filename, stonith_history_t *stonith_history) { FILE *stream; GListPtr gIter = NULL; char *filename_tmp = NULL; int print_opts = get_resource_display_options(); if (output_format == mon_output_cgi) { stream = stdout; fprintf(stream, "Content-Type: text/html\n\n"); } else { filename_tmp = crm_concat(filename, "tmp", '.'); stream = fopen(filename_tmp, "w"); if (stream == NULL) { crm_perror(LOG_ERR, "Cannot open %s for writing", filename_tmp); free(filename_tmp); return -1; } } fprintf(stream, "\n"); fprintf(stream, " \n"); fprintf(stream, " Cluster status\n"); fprintf(stream, " \n", reconnect_msec / 1000); fprintf(stream, " \n"); fprintf(stream, "\n"); print_cluster_summary(stream, data_set); /*** NODE LIST ***/ fprintf(stream, "
    \n

    Node List

    \n"); fprintf(stream, "
      \n"); for (gIter = data_set->nodes; gIter != NULL; gIter = gIter->next) { node_t *node = (node_t *) gIter->data; char *node_name = get_node_display_name(node); fprintf(stream, "
    • Node: %s: ", node_name); if (node->details->standby_onfail && node->details->online) { fprintf(stream, "standby (on-fail)\n"); } else if (node->details->standby && node->details->online) { fprintf(stream, "standby\n"); } else if (node->details->standby) { fprintf(stream, "OFFLINE (standby)\n"); } else if (node->details->maintenance && node->details->online) { fprintf(stream, "maintenance\n"); } else if (node->details->maintenance) { fprintf(stream, "OFFLINE (maintenance)\n"); } else if (node->details->online) { fprintf(stream, "online\n"); } else { fprintf(stream, "OFFLINE\n"); } if (print_brief && group_by_node) { fprintf(stream, "
        \n"); print_rscs_brief(node->details->running_rsc, NULL, print_opts | pe_print_rsconly, stream, FALSE); fprintf(stream, "
      \n"); } else if (group_by_node) { GListPtr lpc2 = NULL; fprintf(stream, "
        \n"); for (lpc2 = node->details->running_rsc; lpc2 != NULL; lpc2 = lpc2->next) { resource_t *rsc = (resource_t *) lpc2->data; fprintf(stream, "
      • "); rsc->fns->print(rsc, NULL, print_opts | pe_print_rsconly, stream); fprintf(stream, "
        • \n"); } fprintf(stream, "
      \n"); } fprintf(stream, "
      • \n"); free(node_name); } fprintf(stream, "
    \n"); /* Print resources section, if needed */ print_resources(stream, data_set, print_opts); /* print Node Attributes section if requested */ if (show & mon_show_attributes) { print_node_attributes(stream, data_set); } /* If requested, print resource operations (which includes failcounts) * or just failcounts */ if (show & (mon_show_operations | mon_show_failcounts)) { print_node_summary(stream, data_set, ((show & mon_show_operations)? TRUE : FALSE)); } /* If there were any failed actions, print them */ if (xml_has_children(data_set->failed)) { print_failed_actions(stream, data_set); } /* Print failed stonith actions */ if (fence_history) { print_failed_stonith_actions(stream, stonith_history); } /* Print stonith history */ if (fence_history && (show & mon_show_fence_history)) { print_stonith_history(stream, stonith_history); } /* Print tickets if requested */ if (show & mon_show_tickets) { print_cluster_tickets(stream, data_set); } /* Print negative location constraints if requested */ if (show & mon_show_bans) { print_neg_locations(stream, data_set); } fprintf(stream, "\n"); fprintf(stream, "\n"); fflush(stream); fclose(stream); if (output_format != mon_output_cgi) { if (rename(filename_tmp, filename) != 0) { crm_perror(LOG_ERR, "Unable to rename %s->%s", filename_tmp, filename); } free(filename_tmp); } return 0; } static int send_custom_trap(const char *node, const char *rsc, const char *task, int target_rc, int rc, int status, const char *desc) { pid_t pid; /*setenv needs chars, these are ints */ char *rc_s = crm_itoa(rc); char *status_s = crm_itoa(status); char *target_rc_s = crm_itoa(target_rc); crm_debug("Sending external notification to '%s' via '%s'", external_recipient, external_agent); if(rsc) { setenv("CRM_notify_rsc", rsc, 1); } if (external_recipient) { setenv("CRM_notify_recipient", external_recipient, 1); } setenv("CRM_notify_node", node, 1); setenv("CRM_notify_task", task, 1); setenv("CRM_notify_desc", desc, 1); setenv("CRM_notify_rc", rc_s, 1); setenv("CRM_notify_target_rc", target_rc_s, 1); setenv("CRM_notify_status", status_s, 1); pid = fork(); if (pid == -1) { crm_perror(LOG_ERR, "notification fork() failed."); } if (pid == 0) { /* crm_debug("notification: I am the child. Executing the nofitication program."); */ execl(external_agent, external_agent, NULL); exit(CRM_EX_ERROR); } crm_trace("Finished running custom notification program '%s'.", external_agent); free(target_rc_s); free(status_s); free(rc_s); return 0; } static void handle_rsc_op(xmlNode * xml, const char *node_id) { int rc = -1; int status = -1; int action = -1; int target_rc = -1; int transition_num = -1; gboolean notify = TRUE; char *rsc = NULL; char *task = NULL; const char *desc = NULL; const char *magic = NULL; const char *id = NULL; char *update_te_uuid = NULL; const char *node = NULL; xmlNode *n = xml; xmlNode * rsc_op = xml; if(strcmp((const char*)xml->name, XML_LRM_TAG_RSC_OP) != 0) { xmlNode *cIter; for(cIter = xml->children; cIter; cIter = cIter->next) { handle_rsc_op(cIter, node_id); } return; } id = crm_element_value(rsc_op, XML_LRM_ATTR_TASK_KEY); if (id == NULL) { /* Compatibility with <= 1.1.5 */ id = ID(rsc_op); } magic = crm_element_value(rsc_op, XML_ATTR_TRANSITION_MAGIC); if (magic == NULL) { /* non-change */ return; } if (FALSE == decode_transition_magic(magic, &update_te_uuid, &transition_num, &action, &status, &rc, &target_rc)) { crm_err("Invalid event %s detected for %s", magic, id); return; } if (parse_op_key(id, &rsc, &task, NULL) == FALSE) { crm_err("Invalid event detected for %s", id); goto bail; } node = crm_element_value(rsc_op, XML_LRM_ATTR_TARGET); while (n != NULL && safe_str_neq(XML_CIB_TAG_STATE, TYPE(n))) { n = n->parent; } if(node == NULL && n) { node = crm_element_value(n, XML_ATTR_UNAME); } if (node == NULL && n) { node = ID(n); } if (node == NULL) { node = node_id; } if (node == NULL) { crm_err("No node detected for event %s (%s)", magic, id); goto bail; } /* look up where we expected it to be? */ desc = pcmk_strerror(pcmk_ok); if (status == PCMK_LRM_OP_DONE && target_rc == rc) { crm_notice("%s of %s on %s completed: %s", task, rsc, node, desc); if (rc == PCMK_OCF_NOT_RUNNING) { notify = FALSE; } } else if (status == PCMK_LRM_OP_DONE) { desc = services_ocf_exitcode_str(rc); crm_warn("%s of %s on %s failed: %s", task, rsc, node, desc); } else { desc = services_lrm_status_str(status); crm_warn("%s of %s on %s failed: %s", task, rsc, node, desc); } if (notify && external_agent) { send_custom_trap(node, rsc, task, target_rc, rc, status, desc); } bail: free(update_te_uuid); free(rsc); free(task); } static gboolean mon_trigger_refresh(gpointer user_data) { mainloop_set_trigger(refresh_trigger); return FALSE; } #define NODE_PATT "/lrm[@id=" static char *get_node_from_xpath(const char *xpath) { char *nodeid = NULL; char *tmp = strstr(xpath, NODE_PATT); if(tmp) { tmp += strlen(NODE_PATT); tmp += 1; nodeid = strdup(tmp); tmp = strstr(nodeid, "\'"); CRM_ASSERT(tmp); tmp[0] = 0; } return nodeid; } static void crm_diff_update_v2(const char *event, xmlNode * msg) { xmlNode *change = NULL; xmlNode *diff = get_message_xml(msg, F_CIB_UPDATE_RESULT); for (change = __xml_first_child(diff); change != NULL; change = __xml_next(change)) { const char *name = NULL; const char *op = crm_element_value(change, XML_DIFF_OP); const char *xpath = crm_element_value(change, XML_DIFF_PATH); xmlNode *match = NULL; const char *node = NULL; if(op == NULL) { continue; } else if(strcmp(op, "create") == 0) { match = change->children; } else if(strcmp(op, "move") == 0) { continue; } else if(strcmp(op, "delete") == 0) { continue; } else if(strcmp(op, "modify") == 0) { match = first_named_child(change, XML_DIFF_RESULT); if(match) { match = match->children; } } if(match) { name = (const char *)match->name; } crm_trace("Handling %s operation for %s %p, %s", op, xpath, match, name); if(xpath == NULL) { /* Version field, ignore */ } else if(name == NULL) { crm_debug("No result for %s operation to %s", op, xpath); CRM_ASSERT(strcmp(op, "delete") == 0 || strcmp(op, "move") == 0); } else if(strcmp(name, XML_TAG_CIB) == 0) { xmlNode *state = NULL; xmlNode *status = first_named_child(match, XML_CIB_TAG_STATUS); for (state = __xml_first_child(status); state != NULL; state = __xml_next(state)) { node = crm_element_value(state, XML_ATTR_UNAME); if (node == NULL) { node = ID(state); } handle_rsc_op(state, node); } } else if(strcmp(name, XML_CIB_TAG_STATUS) == 0) { xmlNode *state = NULL; for (state = __xml_first_child(match); state != NULL; state = __xml_next(state)) { node = crm_element_value(state, XML_ATTR_UNAME); if (node == NULL) { node = ID(state); } handle_rsc_op(state, node); } } else if(strcmp(name, XML_CIB_TAG_STATE) == 0) { node = crm_element_value(match, XML_ATTR_UNAME); if (node == NULL) { node = ID(match); } handle_rsc_op(match, node); } else if(strcmp(name, XML_CIB_TAG_LRM) == 0) { node = ID(match); handle_rsc_op(match, node); } else if(strcmp(name, XML_LRM_TAG_RESOURCES) == 0) { char *local_node = get_node_from_xpath(xpath); handle_rsc_op(match, local_node); free(local_node); } else if(strcmp(name, XML_LRM_TAG_RESOURCE) == 0) { char *local_node = get_node_from_xpath(xpath); handle_rsc_op(match, local_node); free(local_node); } else if(strcmp(name, XML_LRM_TAG_RSC_OP) == 0) { char *local_node = get_node_from_xpath(xpath); handle_rsc_op(match, local_node); free(local_node); } else { crm_trace("Ignoring %s operation for %s %p, %s", op, xpath, match, name); } } } static void crm_diff_update_v1(const char *event, xmlNode * msg) { /* Process operation updates */ xmlXPathObject *xpathObj = xpath_search(msg, "//" F_CIB_UPDATE_RESULT "//" XML_TAG_DIFF_ADDED "//" XML_LRM_TAG_RSC_OP); int lpc = 0, max = numXpathResults(xpathObj); for (lpc = 0; lpc < max; lpc++) { xmlNode *rsc_op = getXpathResult(xpathObj, lpc); handle_rsc_op(rsc_op, NULL); } freeXpathObject(xpathObj); } void crm_diff_update(const char *event, xmlNode * msg) { int rc = -1; static bool stale = FALSE; gboolean cib_updated = FALSE; xmlNode *diff = get_message_xml(msg, F_CIB_UPDATE_RESULT); print_dot(); if (current_cib != NULL) { rc = xml_apply_patchset(current_cib, diff, TRUE); switch (rc) { case -pcmk_err_diff_resync: case -pcmk_err_diff_failed: crm_notice("[%s] Patch aborted: %s (%d)", event, pcmk_strerror(rc), rc); free_xml(current_cib); current_cib = NULL; break; case pcmk_ok: cib_updated = TRUE; break; default: crm_notice("[%s] ABORTED: %s (%d)", event, pcmk_strerror(rc), rc); free_xml(current_cib); current_cib = NULL; } } if (current_cib == NULL) { crm_trace("Re-requesting the full cib"); cib->cmds->query(cib, NULL, ¤t_cib, cib_scope_local | cib_sync_call); } if (external_agent) { int format = 0; crm_element_value_int(diff, "format", &format); switch(format) { case 1: crm_diff_update_v1(event, msg); break; case 2: crm_diff_update_v2(event, msg); break; default: crm_err("Unknown patch format: %d", format); } } if (current_cib == NULL) { if(!stale) { print_as("--- Stale data ---"); } stale = TRUE; return; } stale = FALSE; kick_refresh(cib_updated); } gboolean mon_refresh_display(gpointer user_data) { xmlNode *cib_copy = copy_xml(current_cib); pe_working_set_t data_set; stonith_history_t *stonith_history = NULL; last_refresh = time(NULL); if (cli_config_update(&cib_copy, NULL, FALSE) == FALSE) { if (cib) { cib->cmds->signoff(cib); } print_as("Upgrade failed: %s", pcmk_strerror(-pcmk_err_schema_validation)); if (output_format == mon_output_console) { sleep(2); } clean_up(CRM_EX_CONFIG); return FALSE; } /* get the stonith-history if there is evidence we need it */ while (fence_history) { if (st != NULL) { if (st->cmds->history(st, st_opt_sync_call, NULL, &stonith_history, 120)) { fprintf(stderr, "Critical: Unable to get stonith-history\n"); mon_cib_connection_destroy(NULL); } else { if ((!fence_full_history) && (output_format != mon_output_xml)) { stonith_history = reduce_stonith_history(stonith_history); } stonith_history = sort_stonith_history(stonith_history); break; /* all other cases are errors */ } } else { fprintf(stderr, "Critical: No stonith-API\n"); } free_xml(cib_copy); print_as("Reading stonith-history failed"); if (output_format == mon_output_console) { sleep(2); } return FALSE; } set_working_set_defaults(&data_set); data_set.input = cib_copy; cluster_status(&data_set); /* Unpack constraints if any section will need them * (tickets may be referenced in constraints but not granted yet, * and bans need negative location constraints) */ if (show & (mon_show_bans | mon_show_tickets)) { xmlNode *cib_constraints = get_object_root(XML_CIB_TAG_CONSTRAINTS, data_set.input); unpack_constraints(cib_constraints, &data_set); } switch (output_format) { case mon_output_html: case mon_output_cgi: if (print_html_status(&data_set, output_filename, stonith_history) != 0) { fprintf(stderr, "Critical: Unable to output html file\n"); clean_up(CRM_EX_CANTCREAT); } break; case mon_output_xml: print_xml_status(&data_set, stonith_history); break; case mon_output_monitor: print_simple_status(&data_set, stonith_history); if (has_warnings) { clean_up(MON_STATUS_WARN); } break; case mon_output_plain: case mon_output_console: print_status(&data_set, stonith_history); break; case mon_output_none: break; } stonith_history_free(stonith_history); stonith_history = NULL; cleanup_alloc_calculations(&data_set); - - /* poll for pending fence-actions as we don't get notifications so far */ - if ((!one_shot) && (show & mon_show_fence_history)) { - kick_refresh(FALSE); - } return TRUE; } void mon_st_callback_event(stonith_t * st, stonith_event_t * e) { if (st->state == stonith_disconnected) { /* disconnect cib as well and have everything reconnect */ mon_cib_connection_destroy(NULL); } else if (external_agent) { char *desc = crm_strdup_printf("Operation %s requested by %s for peer %s: %s (ref=%s)", e->operation, e->origin, e->target, pcmk_strerror(e->result), e->id); send_custom_trap(e->target, NULL, e->operation, pcmk_ok, e->result, 0, desc); free(desc); } } void kick_refresh(gboolean data_updated) { static int updates = 0; long now = time(NULL); if (data_updated) { updates++; } if(refresh_timer == NULL) { refresh_timer = mainloop_timer_add("refresh", 2000, FALSE, mon_trigger_refresh, NULL); } /* Refresh * - immediately if the last update was more than 5s ago * - every 10 cib-updates * - at most 2s after the last update */ if ((now - last_refresh) > (reconnect_msec / 1000)) { mainloop_set_trigger(refresh_trigger); mainloop_timer_stop(refresh_timer); updates = 0; } else if(updates >= 10) { mainloop_set_trigger(refresh_trigger); mainloop_timer_stop(refresh_timer); updates = 0; } else { - /* if nothing has been updated (e.g. fence-history polling for changes - * in pending actions since atm that doesn't generate notifications or - * cib update has gone wrong) it is enough to update at reconnect-rate - */ - mainloop_timer_set_period(refresh_timer, data_updated?2000:reconnect_msec); mainloop_timer_start(refresh_timer); } } void mon_st_callback_display(stonith_t * st, stonith_event_t * e) { if (st->state == stonith_disconnected) { /* disconnect cib as well and have everything reconnect */ mon_cib_connection_destroy(NULL); } else { print_dot(); kick_refresh(TRUE); } } /* * De-init ncurses, disconnect from the CIB manager, and deallocate memory. */ static void clean_up(crm_exit_t exit_code) { #if CURSES_ENABLED if (output_format == mon_output_console) { output_format = mon_output_plain; echo(); nocbreak(); endwin(); } #endif if (cib != NULL) { cib->cmds->signoff(cib); cib_delete(cib); cib = NULL; } if (st != NULL) { if (st->state != stonith_disconnected) { st->cmds->remove_notification(st, T_STONITH_NOTIFY_DISCONNECT); + st->cmds->remove_notification(st, T_STONITH_NOTIFY_FENCE); + st->cmds->remove_notification(st, T_STONITH_NOTIFY_HISTORY); st->cmds->disconnect(st); } stonith_api_delete(st); st = NULL; } free(output_filename); free(xml_file); free(pid_file); crm_exit(exit_code); } diff --git a/tools/stonith_admin.c b/tools/stonith_admin.c index 0452bc1ab5..77dddc0cb5 100644 --- a/tools/stonith_admin.c +++ b/tools/stonith_admin.c @@ -1,742 +1,767 @@ /* * Copyright 2009-2018 Andrew Beekhof * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include /* *INDENT-OFF* */ static struct crm_option long_options[] = { { "help", no_argument, NULL, '?', "\tDisplay this text and exit." }, { "version", no_argument, NULL, '$', "\tDisplay version information and exit." }, { "verbose", no_argument, NULL, 'V', "\tIncrease debug output (may be specified multiple times)." }, { "quiet", no_argument, NULL, 'q', "\tBe less descriptive in output." }, - + { "cleanup", no_argument, NULL, 'c', + "\tCleanup wherever appropriate." + }, + { "broadcast", no_argument, NULL, 'b', + "\tBroadcast wherever appropriate." + }, { "-spacer-", no_argument, NULL, '-', "\nDevice definition commands:" }, { "register", required_argument, NULL, 'R', "Register the named stonith device. Requires: --agent.\n" "\t\t\tOptional: --option, --env-option." }, { "deregister", required_argument, NULL, 'D', "De-register the named stonith device." }, { "register-level", required_argument, NULL, 'r', "Register a stonith level for the named target,\n" "\t\t\tspecified as one of NAME, @PATTERN, or ATTR=VALUE.\n" "\t\t\tRequires: --index and one or more --device entries." }, { "deregister-level", required_argument, NULL, 'd', "Unregister a stonith level for the named target,\n" "\t\t\tspecified as for --register-level. Requires: --index." }, { "-spacer-", no_argument, NULL, '-', "\nQueries:" }, { "list", required_argument, NULL, 'l', "List devices that can terminate the specified host.\n" "\t\t\tOptional: --timeout." }, { "list-registered", no_argument, NULL, 'L', "List all registered devices. Optional: --timeout." }, { "list-installed", no_argument, NULL, 'I', "List all installed devices. Optional: --timeout." }, { "list-targets", required_argument, NULL, 's', "List the targets that can be fenced by the\n" "\t\t\tnamed device. Optional: --timeout." }, { "metadata", no_argument, NULL, 'M', "\tShow agent metadata. Requires: --agent.\n" "\t\t\tOptional: --timeout." }, { "query", required_argument, NULL, 'Q', "Check the named device's status. Optional: --timeout." }, { "history", required_argument, NULL, 'H', "Show last successful fencing operation for named node\n" - "\t\t\t(or '*' for all nodes). Optional: --timeout, --quiet\n" - "\t\t\t(show only the operation's epoch timestamp),\n" - "\t\t\t--verbose (show all recorded and pending operations)." + "\t\t\t(or '*' for all nodes). Optional: --timeout, --cleanup,\n" + "\t\t\t--quiet (show only the operation's epoch timestamp),\n" + "\t\t\t--verbose (show all recorded and pending operations),\n" + "\t\t\t--broadcast (update history from all nodes available)." }, { "last", required_argument, NULL, 'h', "Indicate when the named node was last fenced.\n" "\t\t\tOptional: --as-node-id." }, { "validate", no_argument, NULL, 'K', "\tValidate a fence device configuration.\n" "\t\t\tRequires: --agent. Optional: --option, --env-option,\n" "\t\t\t--quiet (print no output, only return status).\n" }, { "-spacer-", no_argument, NULL, '-', "\nFencing Commands:" }, { "fence", required_argument, NULL, 'F', "Fence named host. Optional: --timeout, --tolerance." }, { "unfence", required_argument, NULL, 'U', "Unfence named host. Optional: --timeout, --tolerance." }, { "reboot", required_argument, NULL, 'B', "Reboot named host. Optional: --timeout, --tolerance." }, { "confirm", required_argument, NULL, 'C', "Tell cluster that named host is now safely down." }, { "-spacer-", no_argument, NULL, '-', "\nAdditional Options:" }, { "agent", required_argument, NULL, 'a', "The agent to use (for example, fence_xvm;\n" "\t\t\twith --register, --metadata, --validate)." }, { "option", required_argument, NULL, 'o', "Specify a device configuration parameter as NAME=VALUE\n" "\t\t\t(may be specified multiple times; with --register,\n" "\t\t\t--validate)." }, { "env-option", required_argument, NULL, 'e', "Specify a device configuration parameter with the\n" "\t\t\tspecified name, using the value of the\n" "\t\t\tenvironment variable of the same name prefixed with\n" "\t\t\tOCF_RESKEY_ (may be specified multiple times;\n" "\t\t\twith --register, --validate)." }, { "tag", required_argument, NULL, 'T', "Identify fencing operations in logs with the specified\n" "\t\t\ttag; useful when multiple entities might invoke\n" "\t\t\tstonith_admin (used with most commands)." }, { "device", required_argument, NULL, 'v', "Device ID (with --register-level, device to associate with\n" "\t\t\ta given host and level; may be specified multiple times)" #if SUPPORT_CIBSECRETS "\n\t\t\t(with --validate, name to use to load CIB secrets)" #endif "." }, { "index", required_argument, NULL, 'i', "The stonith level (1-9) (with --register-level,\n" "\t\t\t--deregister-level)." }, { "timeout", required_argument, NULL, 't', "Operation timeout in seconds (default 120;\n" "\t\t\tused with most commands)." }, { "as-node-id", no_argument, NULL, 'n', "(Advanced) The supplied node is the corosync node ID\n" "\t\t\t(with --last)." }, { "tolerance", required_argument, NULL, 0, "(Advanced) Do nothing if an equivalent --fence request\n" "\t\t\tsucceeded less than this many seconds earlier\n" "\t\t\t(with --fence, --unfence, --reboot)." }, { 0, 0, 0, 0 } }; /* *INDENT-ON* */ static int st_opts = st_opt_sync_call | st_opt_allow_suicide; static GMainLoop *mainloop = NULL; struct { stonith_t *st; const char *target; const char *action; char *name; int timeout; int tolerance; int rc; } async_fence_data; static int try_mainloop_connect(void) { stonith_t *st = async_fence_data.st; int tries = 10; int i = 0; int rc = 0; for (i = 0; i < tries; i++) { crm_debug("Connecting as %s", async_fence_data.name); rc = st->cmds->connect(st, async_fence_data.name, NULL); if (!rc) { crm_debug("stonith client connection established"); return 0; } else { crm_debug("stonith client connection failed"); } sleep(1); } crm_err("Could not connect to the fencer"); return -1; } static void notify_callback(stonith_t * st, stonith_event_t * e) { if (e->result != pcmk_ok) { return; } if (safe_str_eq(async_fence_data.target, e->target) && safe_str_eq(async_fence_data.action, e->action)) { async_fence_data.rc = e->result; g_main_loop_quit(mainloop); } } static void fence_callback(stonith_t * stonith, stonith_callback_data_t * data) { async_fence_data.rc = data->rc; g_main_loop_quit(mainloop); } static gboolean async_fence_helper(gpointer user_data) { stonith_t *st = async_fence_data.st; int call_id = 0; if (try_mainloop_connect()) { g_main_loop_quit(mainloop); return TRUE; } st->cmds->register_notification(st, T_STONITH_NOTIFY_FENCE, notify_callback); call_id = st->cmds->fence(st, st_opt_allow_suicide, async_fence_data.target, async_fence_data.action, async_fence_data.timeout, async_fence_data.tolerance); if (call_id < 0) { g_main_loop_quit(mainloop); return TRUE; } st->cmds->register_callback(st, call_id, async_fence_data.timeout, st_opt_timeout_updates, NULL, "callback", fence_callback); return TRUE; } static int mainloop_fencing(stonith_t * st, const char *target, const char *action, int timeout, int tolerance) { crm_trigger_t *trig; async_fence_data.st = st; async_fence_data.target = target; async_fence_data.action = action; async_fence_data.timeout = timeout; async_fence_data.tolerance = tolerance; async_fence_data.rc = -1; trig = mainloop_add_trigger(G_PRIORITY_HIGH, async_fence_helper, NULL); mainloop_set_trigger(trig); mainloop = g_main_loop_new(NULL, FALSE); g_main_loop_run(mainloop); return async_fence_data.rc; } static int handle_level(stonith_t *st, char *target, int fence_level, stonith_key_value_t *devices, bool added) { char *node = NULL; char *pattern = NULL; char *name = NULL; char *value = strchr(target, '='); /* Determine if targeting by attribute, node name pattern or node name */ if (value != NULL) { name = target; *value++ = '\0'; } else if (*target == '@') { pattern = target + 1; } else { node = target; } /* Register or unregister level as appropriate */ if (added) { return st->cmds->register_level_full(st, st_opts, node, pattern, name, value, fence_level, devices); } return st->cmds->remove_level_full(st, st_opts, node, pattern, name, value, fence_level); } static char * fence_action_str(const char *action) { char *str = NULL; if (action == NULL) { str = strdup("unknown"); } else if (action[0] == 'o') { // on, off str = crm_concat("turn", action, ' '); } else { str = strdup(action); } return str; } static void print_fence_event(stonith_history_t *event) { char *action_s = fence_action_str(event->action); time_t complete = event->completed; printf("%s was able to %s node %s on behalf of %s from %s at %s\n", (event->delegate? event->delegate : "This node"), action_s, event->target, event->client, event->origin, ctime(&complete)); free(action_s); } static int -show_history(stonith_t *st, const char *target, int timeout, int quiet, - int verbose) +handle_history(stonith_t *st, const char *target, int timeout, int quiet, + int verbose, int cleanup, int broadcast) { stonith_history_t *history = NULL, *hp, *latest = NULL; int rc = 0; - rc = st->cmds->history(st, st_opts, + if (!quiet) { + if (cleanup) { + printf("cleaning up fencing-history%s%s\n", + target?" for node ":"", target?target:""); + } + if (broadcast) { + printf("gather fencing-history from all nodes\n"); + } + } + rc = st->cmds->history(st, st_opts | (cleanup?st_opt_cleanup:0) | + (broadcast?st_opt_broadcast:0), (safe_str_eq(target, "*")? NULL : target), &history, timeout); for (hp = history; hp; hp = hp->next) { char *action_s = NULL; time_t complete = hp->completed; if (hp->state == st_done) { latest = hp; } if (quiet || !verbose) { continue; } if (hp->state == st_failed) { action_s = fence_action_str(hp->action); printf("%s failed to %s node %s on behalf of %s from %s at %s\n", hp->delegate ? hp->delegate : "We", action_s, hp->target, hp->client, hp->origin, ctime(&complete)); } else if (hp->state == st_done) { print_fence_event(latest); } else { /* ocf:pacemaker:controld depends on "wishes to" being * in this output, when used with older versions of DLM * that don't report stateful_merge_wait */ action_s = fence_action_str(hp->action); printf("%s at %s wishes to %s node %s - %d %lld\n", hp->client, hp->origin, action_s, hp->target, hp->state, (long long) complete); } free(action_s); } if (latest) { if (quiet) { printf("%lld\n", (long long) latest->completed); } else if (!verbose) { // already printed if verbose print_fence_event(latest); } } stonith_history_free(history); return rc; } static int validate(stonith_t *st, const char *agent, const char *id, stonith_key_value_t *params, int timeout, int quiet) { int rc = 1; char *output = NULL; char *error_output = NULL; rc = st->cmds->validate(st, st_opt_sync_call, id, NULL, agent, params, timeout, &output, &error_output); if (!quiet) { printf("Validation of %s %s\n", agent, (rc? "failed" : "succeeded")); if (output && *output) { puts(output); free(output); } if (error_output && *error_output) { puts(error_output); free(error_output); } } return rc; } int main(int argc, char **argv) { int flag; int rc = 0; int quiet = 0; + int cleanup = 0; + int broadcast = 0; int verbose = 0; int argerr = 0; int timeout = 120; int option_index = 0; int fence_level = 0; int no_connect = 0; int tolerance = 0; int as_nodeid = FALSE; char *name = NULL; char *value = NULL; char *target = NULL; char *lists = NULL; const char *agent = NULL; const char *device = NULL; const char *longname = NULL; char action = 0; crm_exit_t exit_code = CRM_EX_OK; stonith_t *st = NULL; stonith_key_value_t *params = NULL; stonith_key_value_t *devices = NULL; stonith_key_value_t *dIter = NULL; crm_log_cli_init("stonith_admin"); crm_set_options(NULL, " []", long_options, "access the Pacemaker fencing API"); async_fence_data.name = strdup(crm_system_name); while (1) { flag = crm_get_option_long(argc, argv, &option_index, &longname); if (flag == -1) break; switch (flag) { case 'V': verbose = 1; crm_bump_log_level(argc, argv); break; case '$': case '?': crm_help(flag, CRM_EX_OK); break; case 'I': case 'K': no_connect = 1; /* fall through */ case 'L': action = flag; break; case 'q': quiet = 1; break; + case 'c': + cleanup = 1; + break; + case 'b': + broadcast = 1; + break; case 'Q': case 'R': case 'D': case 's': action = flag; device = optarg; break; case 'T': free(async_fence_data.name); async_fence_data.name = crm_strdup_printf("%s.%s", crm_system_name, optarg); break; case 'a': agent = optarg; break; case 'l': target = optarg; action = 'L'; break; case 'M': no_connect = 1; action = flag; break; case 't': timeout = crm_atoi(optarg, NULL); break; case 'B': case 'F': case 'U': /* using mainloop here */ no_connect = 1; /* fall through */ case 'C': /* Always log the input arguments */ crm_log_args(argc, argv); target = optarg; action = flag; break; case 'n': as_nodeid = TRUE; break; case 'h': case 'H': case 'r': case 'd': target = optarg; action = flag; break; case 'i': fence_level = crm_atoi(optarg, NULL); break; case 'v': devices = stonith_key_value_add(devices, NULL, optarg); break; case 'o': crm_info("Scanning: -o %s", optarg); rc = sscanf(optarg, "%m[^=]=%m[^=]", &name, &value); if (rc != 2) { crm_err("Invalid option: -o %s", optarg); ++argerr; } else { crm_info("Got: '%s'='%s'", name, value); params = stonith_key_value_add(params, name, value); } free(value); value = NULL; free(name); name = NULL; break; case 'e': { char *key = crm_concat("OCF_RESKEY", optarg, '_'); const char *env = getenv(key); if (env == NULL) { crm_err("Invalid option: -e %s", optarg); ++argerr; } else { crm_info("Got: '%s'='%s'", optarg, env); params = stonith_key_value_add(params, optarg, env); } free(key); } break; case 0: if (safe_str_eq("tolerance", longname)) { tolerance = crm_get_msec(optarg) / 1000; /* Send in seconds */ } break; default: ++argerr; break; } } if (optind > argc) { ++argerr; } if (argerr) { crm_help('?', CRM_EX_USAGE); } st = stonith_api_new(); if (!no_connect) { rc = st->cmds->connect(st, async_fence_data.name, NULL); if (rc < 0) { fprintf(stderr, "Could not connect to fencer: %s\n", pcmk_strerror(rc)); exit_code = CRM_EX_DISCONNECT; goto done; } } switch (action) { case 'I': rc = st->cmds->list_agents(st, st_opt_sync_call, NULL, &devices, timeout); for (dIter = devices; dIter; dIter = dIter->next) { fprintf(stdout, " %s\n", dIter->value); } if (rc == 0) { fprintf(stderr, "No devices found\n"); } else if (rc > 0) { fprintf(stderr, "%d devices found\n", rc); rc = 0; } stonith_key_value_freeall(devices, 1, 1); break; case 'L': rc = st->cmds->query(st, st_opts, target, &devices, timeout); for (dIter = devices; dIter; dIter = dIter->next) { fprintf(stdout, " %s\n", dIter->value); } if (rc == 0) { fprintf(stderr, "No devices found\n"); } else if (rc > 0) { fprintf(stderr, "%d devices found\n", rc); rc = 0; } stonith_key_value_freeall(devices, 1, 1); break; case 'Q': rc = st->cmds->monitor(st, st_opts, device, timeout); if (rc < 0) { rc = st->cmds->list(st, st_opts, device, NULL, timeout); } break; case 's': rc = st->cmds->list(st, st_opts, device, &lists, timeout); if (rc == 0) { if (lists) { char *source = lists, *dest = lists; while (*dest) { if ((*dest == '\\') && (*(dest+1) == 'n')) { *source = '\n'; dest++; dest++; source++; } else if ((*dest == ',') || (*dest == ';')) { dest++; } else { *source = *dest; dest++; source++; } if (!(*dest)) { *source = 0; } } fprintf(stdout, "%s", lists); free(lists); } } else { fprintf(stderr, "List command returned error. rc : %d\n", rc); } break; case 'R': rc = st->cmds->register_device(st, st_opts, device, NULL, agent, params); break; case 'D': rc = st->cmds->remove_device(st, st_opts, device); break; case 'd': case 'r': rc = handle_level(st, target, fence_level, devices, action == 'r'); break; case 'M': if (agent == NULL) { printf("Please specify an agent to query using -a,--agent [value]\n"); exit_code = CRM_EX_USAGE; goto done; } else { char *buffer = NULL; rc = st->cmds->metadata(st, st_opt_sync_call, agent, NULL, &buffer, timeout); if (rc == pcmk_ok) { printf("%s\n", buffer); } free(buffer); } break; case 'C': rc = st->cmds->confirm(st, st_opts, target); break; case 'B': rc = mainloop_fencing(st, target, "reboot", timeout, tolerance); break; case 'F': rc = mainloop_fencing(st, target, "off", timeout, tolerance); break; case 'U': rc = mainloop_fencing(st, target, "on", timeout, tolerance); break; case 'h': { time_t when = 0; if(as_nodeid) { uint32_t nodeid = atol(target); when = stonith_api_time(nodeid, NULL, FALSE); } else { when = stonith_api_time(0, target, FALSE); } if(when) { printf("Node %s last kicked at: %s\n", target, ctime(&when)); } else { printf("Node %s has never been kicked\n", target); } } break; case 'H': - rc = show_history(st, target, timeout, quiet, verbose); + rc = handle_history(st, target, timeout, quiet, + verbose, cleanup, broadcast); break; case 'K': if (agent == NULL) { printf("Please specify an agent to validate with --agent\n"); exit_code = CRM_EX_USAGE; goto done; } device = (devices? devices->key : NULL); rc = validate(st, agent, device, params, timeout, quiet); break; } crm_info("Command returned: %s (%d)", pcmk_strerror(rc), rc); exit_code = crm_errno2exit(rc); done: free(async_fence_data.name); stonith_key_value_freeall(params, 1, 1); st->cmds->disconnect(st); stonith_api_delete(st); return exit_code; }