diff --git a/lib/pacemaker/pcmk_sched_bundle.c b/lib/pacemaker/pcmk_sched_bundle.c index f3bad36aad..4270f6d0f6 100644 --- a/lib/pacemaker/pcmk_sched_bundle.c +++ b/lib/pacemaker/pcmk_sched_bundle.c @@ -1,890 +1,876 @@ /* * Copyright 2004-2023 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include "libpacemaker_private.h" #define PE__VARIANT_BUNDLE 1 #include static bool is_bundle_node(pe__bundle_variant_data_t *data, pe_node_t *node) { for (GList *gIter = data->replicas; gIter != NULL; gIter = gIter->next) { pe__bundle_replica_t *replica = gIter->data; if (node->details == replica->node->details) { return TRUE; } } return FALSE; } /*! * \internal * \brief Assign a bundle resource to a node * * \param[in,out] rsc Resource to assign to a node * \param[in] prefer Node to prefer, if all else is equal * * \return Node that \p rsc is assigned to, if assigned entirely to one node */ pe_node_t * pcmk__bundle_allocate(pe_resource_t *rsc, const pe_node_t *prefer) { GList *containers = NULL; pe__bundle_variant_data_t *bundle_data = NULL; CRM_CHECK(rsc != NULL, return NULL); get_bundle_variant_data(bundle_data, rsc); pe__set_resource_flags(rsc, pe_rsc_allocating); containers = pe__bundle_containers(rsc); pe__show_node_weights(!pcmk_is_set(rsc->cluster->flags, pe_flag_show_scores), rsc, __func__, rsc->allowed_nodes, rsc->cluster); containers = g_list_sort(containers, pcmk__cmp_instance); pcmk__assign_instances(rsc, containers, bundle_data->nreplicas, bundle_data->nreplicas_per_host); g_list_free(containers); for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pe__bundle_replica_t *replica = gIter->data; pe_node_t *container_host = NULL; CRM_ASSERT(replica); if (replica->ip) { pe_rsc_trace(rsc, "Allocating bundle %s IP %s", rsc->id, replica->ip->id); replica->ip->cmds->assign(replica->ip, prefer); } container_host = replica->container->allocated_to; if (replica->remote && pe__is_guest_or_remote_node(container_host)) { /* We need 'nested' connection resources to be on the same * host because pacemaker-remoted only supports a single * active connection */ pcmk__new_colocation("child-remote-with-docker-remote", NULL, INFINITY, replica->remote, container_host->details->remote_rsc, NULL, NULL, true, rsc->cluster); } if (replica->remote) { pe_rsc_trace(rsc, "Allocating bundle %s connection %s", rsc->id, replica->remote->id); replica->remote->cmds->assign(replica->remote, prefer); } // Explicitly allocate replicas' children before bundle child if (replica->child) { pe_node_t *node = NULL; GHashTableIter iter; g_hash_table_iter_init(&iter, replica->child->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (gpointer *) & node)) { if (node->details != replica->node->details) { node->weight = -INFINITY; } else if (!pcmk__threshold_reached(replica->child, node, NULL)) { node->weight = INFINITY; } } pe__set_resource_flags(replica->child->parent, pe_rsc_allocating); pe_rsc_trace(rsc, "Allocating bundle %s replica child %s", rsc->id, replica->child->id); replica->child->cmds->assign(replica->child, replica->node); pe__clear_resource_flags(replica->child->parent, pe_rsc_allocating); } } if (bundle_data->child) { pe_node_t *node = NULL; GHashTableIter iter; g_hash_table_iter_init(&iter, bundle_data->child->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (gpointer *) & node)) { if (is_bundle_node(bundle_data, node)) { node->weight = 0; } else { node->weight = -INFINITY; } } pe_rsc_trace(rsc, "Allocating bundle %s child %s", rsc->id, bundle_data->child->id); bundle_data->child->cmds->assign(bundle_data->child, prefer); } pe__clear_resource_flags(rsc, pe_rsc_allocating|pe_rsc_provisional); return NULL; } void pcmk__bundle_create_actions(pe_resource_t *rsc) { pe_action_t *action = NULL; GList *containers = NULL; pe__bundle_variant_data_t *bundle_data = NULL; CRM_CHECK(rsc != NULL, return); containers = pe__bundle_containers(rsc); get_bundle_variant_data(bundle_data, rsc); for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pe__bundle_replica_t *replica = gIter->data; CRM_ASSERT(replica); if (replica->ip) { replica->ip->cmds->create_actions(replica->ip); } if (replica->container) { replica->container->cmds->create_actions(replica->container); } if (replica->remote) { replica->remote->cmds->create_actions(replica->remote); } } pcmk__create_instance_actions(rsc, containers, NULL, NULL); if (bundle_data->child) { bundle_data->child->cmds->create_actions(bundle_data->child); if (pcmk_is_set(bundle_data->child->flags, pe_rsc_promotable)) { /* promote */ pe__new_rsc_pseudo_action(rsc, RSC_PROMOTE, true, true); action = pe__new_rsc_pseudo_action(rsc, RSC_PROMOTED, true, true); action->priority = INFINITY; /* demote */ pe__new_rsc_pseudo_action(rsc, RSC_DEMOTE, true, true); action = pe__new_rsc_pseudo_action(rsc, RSC_DEMOTED, true, true); action->priority = INFINITY; } } g_list_free(containers); } void pcmk__bundle_internal_constraints(pe_resource_t *rsc) { pe__bundle_variant_data_t *bundle_data = NULL; CRM_CHECK(rsc != NULL, return); get_bundle_variant_data(bundle_data, rsc); if (bundle_data->child) { pcmk__order_resource_actions(rsc, RSC_START, bundle_data->child, RSC_START, pe_order_implies_first_printed); pcmk__order_resource_actions(rsc, RSC_STOP, bundle_data->child, RSC_STOP, pe_order_implies_first_printed); if (bundle_data->child->children) { pcmk__order_resource_actions(bundle_data->child, RSC_STARTED, rsc, RSC_STARTED, pe_order_implies_then_printed); pcmk__order_resource_actions(bundle_data->child, RSC_STOPPED, rsc, RSC_STOPPED, pe_order_implies_then_printed); } else { pcmk__order_resource_actions(bundle_data->child, RSC_START, rsc, RSC_STARTED, pe_order_implies_then_printed); pcmk__order_resource_actions(bundle_data->child, RSC_STOP, rsc, RSC_STOPPED, pe_order_implies_then_printed); } } for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pe__bundle_replica_t *replica = gIter->data; CRM_ASSERT(replica); CRM_ASSERT(replica->container); replica->container->cmds->internal_constraints(replica->container); pcmk__order_starts(rsc, replica->container, pe_order_runnable_left|pe_order_implies_first_printed); if (replica->child) { pcmk__order_stops(rsc, replica->child, pe_order_implies_first_printed); } pcmk__order_stops(rsc, replica->container, pe_order_implies_first_printed); pcmk__order_resource_actions(replica->container, RSC_START, rsc, RSC_STARTED, pe_order_implies_then_printed); pcmk__order_resource_actions(replica->container, RSC_STOP, rsc, RSC_STOPPED, pe_order_implies_then_printed); if (replica->ip) { replica->ip->cmds->internal_constraints(replica->ip); // Start IP then container pcmk__order_starts(replica->ip, replica->container, pe_order_runnable_left|pe_order_preserve); pcmk__order_stops(replica->container, replica->ip, pe_order_implies_first|pe_order_preserve); pcmk__new_colocation("ip-with-docker", NULL, INFINITY, replica->ip, replica->container, NULL, NULL, true, rsc->cluster); } if (replica->remote) { /* This handles ordering and colocating remote relative to container * (via "resource-with-container"). Since IP is also ordered and * colocated relative to the container, we don't need to do anything * explicit here with IP. */ replica->remote->cmds->internal_constraints(replica->remote); } if (replica->child) { CRM_ASSERT(replica->remote); // "Start remote then child" is implicit in scheduler's remote logic } } if (bundle_data->child) { bundle_data->child->cmds->internal_constraints(bundle_data->child); if (pcmk_is_set(bundle_data->child->flags, pe_rsc_promotable)) { pcmk__promotable_restart_ordering(rsc); /* child demoted before global demoted */ pcmk__order_resource_actions(bundle_data->child, RSC_DEMOTED, rsc, RSC_DEMOTED, pe_order_implies_then_printed); /* global demote before child demote */ pcmk__order_resource_actions(rsc, RSC_DEMOTE, bundle_data->child, RSC_DEMOTE, pe_order_implies_first_printed); /* child promoted before global promoted */ pcmk__order_resource_actions(bundle_data->child, RSC_PROMOTED, rsc, RSC_PROMOTED, pe_order_implies_then_printed); /* global promote before child promote */ pcmk__order_resource_actions(rsc, RSC_PROMOTE, bundle_data->child, RSC_PROMOTE, pe_order_implies_first_printed); } } } static pe_resource_t * compatible_replica_for_node(const pe_resource_t *rsc_lh, const pe_node_t *candidate, const pe_resource_t *rsc, enum rsc_role_e filter, gboolean current) { pe__bundle_variant_data_t *bundle_data = NULL; CRM_CHECK(candidate != NULL, return NULL); get_bundle_variant_data(bundle_data, rsc); crm_trace("Looking for compatible child from %s for %s on %s", rsc_lh->id, rsc->id, pe__node_name(candidate)); for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pe__bundle_replica_t *replica = gIter->data; if (pcmk__instance_matches(replica->container, candidate, filter, current)) { crm_trace("Pairing %s with %s on %s", rsc_lh->id, replica->container->id, pe__node_name(candidate)); return replica->container; } } crm_trace("Can't pair %s with %s", rsc_lh->id, rsc->id); return NULL; } static pe_resource_t * compatible_replica(const pe_resource_t *rsc_lh, const pe_resource_t *rsc, enum rsc_role_e filter, gboolean current, pe_working_set_t *data_set) { GList *scratch = NULL; pe_resource_t *pair = NULL; pe_node_t *active_node_lh = NULL; active_node_lh = rsc_lh->fns->location(rsc_lh, NULL, current); if (active_node_lh) { return compatible_replica_for_node(rsc_lh, active_node_lh, rsc, filter, current); } scratch = g_hash_table_get_values(rsc_lh->allowed_nodes); scratch = pcmk__sort_nodes(scratch, NULL); for (GList *gIter = scratch; gIter != NULL; gIter = gIter->next) { pe_node_t *node = (pe_node_t *) gIter->data; pair = compatible_replica_for_node(rsc_lh, node, rsc, filter, current); if (pair) { goto done; } } pe_rsc_debug(rsc, "Can't pair %s with %s", rsc_lh->id, (rsc? rsc->id : "none")); done: g_list_free(scratch); return pair; } int copies_per_node(pe_resource_t * rsc) { /* Strictly speaking, there should be a 'copies_per_node' addition * to the resource function table and each case would be a * function. However that would be serious overkill to return an * int. In fact, it seems to me that both function tables * could/should be replaced by resources.{c,h} full of * rsc_{some_operation} functions containing a switch as below * which calls out to functions named {variant}_{some_operation} * as needed. */ switch(rsc->variant) { case pe_unknown: return 0; case pe_native: case pe_group: return 1; case pe_clone: { const char *max_clones_node = g_hash_table_lookup(rsc->meta, XML_RSC_ATTR_INCARNATION_NODEMAX); if (max_clones_node == NULL) { return 1; } else { int max_i; pcmk__scan_min_int(max_clones_node, &max_i, 0); return max_i; } } case pe_container: { pe__bundle_variant_data_t *data = NULL; get_bundle_variant_data(data, rsc); return data->nreplicas_per_host; } } return 0; } /*! * \internal * \brief Apply a colocation's score to node weights or resource priority * * Given a colocation constraint, apply its score to the dependent's * allowed node weights (if we are still placing resources) or priority (if * we are choosing promotable clone instance roles). * * \param[in,out] dependent Dependent resource in colocation * \param[in] primary Primary resource in colocation * \param[in] colocation Colocation constraint to apply * \param[in] for_dependent true if called on behalf of dependent */ void pcmk__bundle_apply_coloc_score(pe_resource_t *dependent, const pe_resource_t *primary, const pcmk__colocation_t *colocation, bool for_dependent) { GList *allocated_primaries = NULL; pe__bundle_variant_data_t *bundle_data = NULL; /* This should never be called for the bundle itself as a dependent. * Instead, we add its colocation constraints to its replicas and call the * apply_coloc_score() for the replicas as dependents. */ CRM_ASSERT(!for_dependent); CRM_CHECK((colocation != NULL) && (dependent != NULL) && (primary != NULL), return); CRM_ASSERT(dependent->variant == pe_native); if (pcmk_is_set(primary->flags, pe_rsc_provisional)) { pe_rsc_trace(primary, "%s is still provisional", primary->id); return; } else if (colocation->dependent->variant > pe_group) { pe_resource_t *primary_replica = compatible_replica(dependent, primary, RSC_ROLE_UNKNOWN, FALSE, dependent->cluster); if (primary_replica) { pe_rsc_debug(primary, "Pairing %s with %s", dependent->id, primary_replica->id); dependent->cmds->apply_coloc_score(dependent, primary_replica, colocation, true); } else if (colocation->score >= INFINITY) { crm_notice("Cannot pair %s with instance of %s", dependent->id, primary->id); pcmk__assign_resource(dependent, NULL, true); } else { pe_rsc_debug(primary, "Cannot pair %s with instance of %s", dependent->id, primary->id); } return; } get_bundle_variant_data(bundle_data, primary); pe_rsc_trace(primary, "Processing constraint %s: %s -> %s %d", colocation->id, dependent->id, primary->id, colocation->score); for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pe__bundle_replica_t *replica = gIter->data; if (colocation->score < INFINITY) { replica->container->cmds->apply_coloc_score(dependent, replica->container, colocation, false); } else { pe_node_t *chosen = replica->container->fns->location(replica->container, NULL, FALSE); if ((chosen == NULL) || is_set_recursive(replica->container, pe_rsc_block, TRUE)) { continue; } if ((colocation->primary_role >= RSC_ROLE_PROMOTED) && (replica->child == NULL)) { continue; } if ((colocation->primary_role >= RSC_ROLE_PROMOTED) && (replica->child->next_role < RSC_ROLE_PROMOTED)) { continue; } pe_rsc_trace(primary, "Allowing %s: %s %d", colocation->id, pe__node_name(chosen), chosen->weight); allocated_primaries = g_list_prepend(allocated_primaries, chosen); } } if (colocation->score >= INFINITY) { node_list_exclude(dependent->allowed_nodes, allocated_primaries, FALSE); } g_list_free(allocated_primaries); } // Bundle implementation of resource_alloc_functions_t:with_this_colocations() void pcmk__with_bundle_colocations(const pe_resource_t *rsc, const pe_resource_t *orig_rsc, GList **list) { CRM_CHECK((rsc != NULL) && (rsc->variant == pe_container) && (orig_rsc != NULL) && (list != NULL), return); - // @COMPAT with previous (incorrect) behavior - if ((rsc != orig_rsc) - && pcmk_is_set(rsc->flags, pe_rsc_provisional) - && !pcmk_is_set(rsc->flags, pe_rsc_allocating)) { - return; // Containers have not yet received bundle colocations - } - if (rsc == orig_rsc) { // Colocations are wanted for bundle itself pcmk__add_with_this_list(list, rsc->rsc_cons_lhs); // Only the bundle replicas' containers get the bundle's constraints } else if (pcmk_is_set(orig_rsc->flags, pe_rsc_replica_container)) { pcmk__add_collective_constraints(list, orig_rsc, rsc, true); } } // Bundle implementation of resource_alloc_functions_t:this_with_colocations() void pcmk__bundle_with_colocations(const pe_resource_t *rsc, const pe_resource_t *orig_rsc, GList **list) { CRM_CHECK((rsc != NULL) && (rsc->variant == pe_container) && (orig_rsc != NULL) && (list != NULL), return); - // @COMPAT with previous (incorrect) behavior - if ((rsc != orig_rsc) - && pcmk_is_set(rsc->flags, pe_rsc_provisional) - && !pcmk_is_set(rsc->flags, pe_rsc_allocating)) { - return; // Containers have not yet received bundle colocations - } - if (rsc == orig_rsc) { // Colocations are wanted for bundle itself pcmk__add_this_with_list(list, rsc->rsc_cons); // Only the bundle replicas' containers get the bundle's constraints } else if (pcmk_is_set(orig_rsc->flags, pe_rsc_replica_container)) { pcmk__add_collective_constraints(list, orig_rsc, rsc, false); } } enum pe_action_flags pcmk__bundle_action_flags(pe_action_t *action, const pe_node_t *node) { GList *containers = NULL; enum pe_action_flags flags = 0; pe__bundle_variant_data_t *data = NULL; get_bundle_variant_data(data, action->rsc); if(data->child) { enum action_tasks task = get_complex_task(data->child, action->task); switch(task) { case no_action: case action_notify: case action_notified: case action_promote: case action_promoted: case action_demote: case action_demoted: return pcmk__collective_action_flags(action, data->child->children, node); default: break; } } containers = pe__bundle_containers(action->rsc); flags = pcmk__collective_action_flags(action, containers, node); g_list_free(containers); return flags; } /*! * \internal * \brief Get containerized resource corresponding to a given bundle container * * \param[in] instance Collective instance that might be a bundle container * * \return Bundled resource instance inside \p instance if it is a bundle * container instance, otherwise NULL */ const pe_resource_t * pcmk__get_rsc_in_container(const pe_resource_t *instance) { const pe__bundle_variant_data_t *data = NULL; const pe_resource_t *top = pe__const_top_resource(instance, true); if ((top == NULL) || (top->variant != pe_container)) { return NULL; } get_bundle_variant_data(data, top); for (const GList *iter = data->replicas; iter != NULL; iter = iter->next) { const pe__bundle_replica_t *replica = iter->data; if (instance == replica->container) { return replica->child; } } return NULL; } void pcmk__bundle_rsc_location(pe_resource_t *rsc, pe__location_t *constraint) { pe__bundle_variant_data_t *bundle_data = NULL; get_bundle_variant_data(bundle_data, rsc); pcmk__apply_location(rsc, constraint); for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pe__bundle_replica_t *replica = gIter->data; if (replica->container) { replica->container->cmds->apply_location(replica->container, constraint); } if (replica->ip) { replica->ip->cmds->apply_location(replica->ip, constraint); } } if (bundle_data->child && ((constraint->role_filter == RSC_ROLE_UNPROMOTED) || (constraint->role_filter == RSC_ROLE_PROMOTED))) { bundle_data->child->cmds->apply_location(bundle_data->child, constraint); bundle_data->child->rsc_location = g_list_prepend(bundle_data->child->rsc_location, constraint); } } /*! * \internal * \brief Add a resource's actions to the transition graph * * \param[in,out] rsc Resource whose actions should be added */ void pcmk__bundle_expand(pe_resource_t *rsc) { pe__bundle_variant_data_t *bundle_data = NULL; CRM_CHECK(rsc != NULL, return); get_bundle_variant_data(bundle_data, rsc); if (bundle_data->child) { bundle_data->child->cmds->add_actions_to_graph(bundle_data->child); } for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pe__bundle_replica_t *replica = gIter->data; CRM_ASSERT(replica); if (replica->remote && replica->container && pe__bundle_needs_remote_name(replica->remote)) { /* REMOTE_CONTAINER_HACK: Allow remote nodes to run containers that * run pacemaker-remoted inside, without needing a separate IP for * the container. This is done by configuring the inner remote's * connection host as the magic string "#uname", then * replacing it with the underlying host when needed. */ xmlNode *nvpair = get_xpath_object("//nvpair[@name='" XML_RSC_ATTR_REMOTE_RA_ADDR "']", replica->remote->xml, LOG_ERR); const char *calculated_addr = NULL; // Replace the value in replica->remote->xml (if appropriate) calculated_addr = pe__add_bundle_remote_name(replica->remote, rsc->cluster, nvpair, "value"); if (calculated_addr) { /* Since this is for the bundle as a resource, and not any * particular action, replace the value in the default * parameters (not evaluated for node). create_graph_action() * will grab it from there to replace it in node-evaluated * parameters. */ GHashTable *params = pe_rsc_params(replica->remote, NULL, rsc->cluster); g_hash_table_replace(params, strdup(XML_RSC_ATTR_REMOTE_RA_ADDR), strdup(calculated_addr)); } else { /* The only way to get here is if the remote connection is * neither currently running nor scheduled to run. That means we * won't be doing any operations that require addr (only start * requires it; we additionally use it to compare digests when * unpacking status, promote, and migrate_from history, but * that's already happened by this point). */ crm_info("Unable to determine address for bundle %s remote connection", rsc->id); } } if (replica->ip) { replica->ip->cmds->add_actions_to_graph(replica->ip); } if (replica->container) { replica->container->cmds->add_actions_to_graph(replica->container); } if (replica->remote) { replica->remote->cmds->add_actions_to_graph(replica->remote); } } } /*! * \internal * * \brief Schedule any probes needed for a resource on a node * * \param[in,out] rsc Resource to create probe for * \param[in,out] node Node to create probe on * * \return true if any probe was created, otherwise false */ bool pcmk__bundle_create_probe(pe_resource_t *rsc, pe_node_t *node) { bool any_created = false; pe__bundle_variant_data_t *bundle_data = NULL; CRM_CHECK(rsc != NULL, return false); get_bundle_variant_data(bundle_data, rsc); for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pe__bundle_replica_t *replica = gIter->data; CRM_ASSERT(replica); if ((replica->ip != NULL) && replica->ip->cmds->create_probe(replica->ip, node)) { any_created = true; } if ((replica->child != NULL) && (node->details == replica->node->details) && replica->child->cmds->create_probe(replica->child, node)) { any_created = true; } if ((replica->container != NULL) && replica->container->cmds->create_probe(replica->container, node)) { any_created = true; /* If we're limited to one replica per host (due to * the lack of an IP range probably), then we don't * want any of our peer containers starting until * we've established that no other copies are already * running. * * Partly this is to ensure that nreplicas_per_host is * observed, but also to ensure that the containers * don't fail to start because the necessary port * mappings (which won't include an IP for uniqueness) * are already taken */ for (GList *tIter = bundle_data->replicas; tIter && (bundle_data->nreplicas_per_host == 1); tIter = tIter->next) { pe__bundle_replica_t *other = tIter->data; if ((other != replica) && (other != NULL) && (other->container != NULL)) { pcmk__new_ordering(replica->container, pcmk__op_key(replica->container->id, RSC_STATUS, 0), NULL, other->container, pcmk__op_key(other->container->id, RSC_START, 0), NULL, pe_order_optional|pe_order_same_node, rsc->cluster); } } } if ((replica->container != NULL) && (replica->remote != NULL) && replica->remote->cmds->create_probe(replica->remote, node)) { /* Do not probe the remote resource until we know where the * container is running. This is required for REMOTE_CONTAINER_HACK * to correctly probe remote resources. */ char *probe_uuid = pcmk__op_key(replica->remote->id, RSC_STATUS, 0); pe_action_t *probe = find_first_action(replica->remote->actions, probe_uuid, NULL, node); free(probe_uuid); if (probe != NULL) { any_created = true; crm_trace("Ordering %s probe on %s", replica->remote->id, pe__node_name(node)); pcmk__new_ordering(replica->container, pcmk__op_key(replica->container->id, RSC_START, 0), NULL, replica->remote, NULL, probe, pe_order_probe, rsc->cluster); } } } return any_created; } void pcmk__output_bundle_actions(pe_resource_t *rsc) { pe__bundle_variant_data_t *bundle_data = NULL; CRM_CHECK(rsc != NULL, return); get_bundle_variant_data(bundle_data, rsc); for (GList *gIter = bundle_data->replicas; gIter != NULL; gIter = gIter->next) { pe__bundle_replica_t *replica = gIter->data; CRM_ASSERT(replica); if (replica->ip != NULL) { replica->ip->cmds->output_actions(replica->ip); } if (replica->container != NULL) { replica->container->cmds->output_actions(replica->container); } if (replica->remote != NULL) { replica->remote->cmds->output_actions(replica->remote); } if (replica->child != NULL) { replica->child->cmds->output_actions(replica->child); } } } // Bundle implementation of resource_alloc_functions_t:add_utilization() void pcmk__bundle_add_utilization(const pe_resource_t *rsc, const pe_resource_t *orig_rsc, GList *all_rscs, GHashTable *utilization) { pe__bundle_variant_data_t *bundle_data = NULL; pe__bundle_replica_t *replica = NULL; if (!pcmk_is_set(rsc->flags, pe_rsc_provisional)) { return; } get_bundle_variant_data(bundle_data, rsc); if (bundle_data->replicas == NULL) { return; } /* All bundle replicas are identical, so using the utilization of the first * is sufficient for any. Only the implicit container resource can have * utilization values. */ replica = (pe__bundle_replica_t *) bundle_data->replicas->data; if (replica->container != NULL) { replica->container->cmds->add_utilization(replica->container, orig_rsc, all_rscs, utilization); } } // Bundle implementation of resource_alloc_functions_t:shutdown_lock() void pcmk__bundle_shutdown_lock(pe_resource_t *rsc) { return; // Bundles currently don't support shutdown locks } diff --git a/lib/pacemaker/pcmk_sched_clone.c b/lib/pacemaker/pcmk_sched_clone.c index 6008f74437..3c883315f4 100644 --- a/lib/pacemaker/pcmk_sched_clone.c +++ b/lib/pacemaker/pcmk_sched_clone.c @@ -1,693 +1,679 @@ /* * Copyright 2004-2023 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include "libpacemaker_private.h" #define VARIANT_CLONE 1 #include /*! * \internal * \brief Assign a clone resource to a node * * \param[in,out] rsc Resource to assign to a node * \param[in] prefer Node to prefer, if all else is equal * * \return Node that \p rsc is assigned to, if assigned entirely to one node */ pe_node_t * pcmk__clone_allocate(pe_resource_t *rsc, const pe_node_t *prefer) { clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, rsc); if (!pcmk_is_set(rsc->flags, pe_rsc_provisional)) { return NULL; } else if (pcmk_is_set(rsc->flags, pe_rsc_allocating)) { pe_rsc_debug(rsc, "Dependency loop detected involving %s", rsc->id); return NULL; } if (pcmk_is_set(rsc->flags, pe_rsc_promotable)) { pcmk__add_promotion_scores(rsc); } pe__set_resource_flags(rsc, pe_rsc_allocating); /* If this clone is colocated with any other resources, assign those first. * Since the this_with_colocations() method boils down to a copy of rsc_cons * for clones, we can use that here directly for efficiency. */ for (GList *gIter = rsc->rsc_cons; gIter != NULL; gIter = gIter->next) { pcmk__colocation_t *constraint = (pcmk__colocation_t *) gIter->data; pe_rsc_trace(rsc, "%s: Allocating %s first", rsc->id, constraint->primary->id); constraint->primary->cmds->assign(constraint->primary, prefer); } /* If any resources are colocated with this one, consider their preferences. * Because the with_this_colocations() method boils down to a copy of * rsc_cons_lhs for clones, we can use that here directly for efficiency. */ for (GList *gIter = rsc->rsc_cons_lhs; gIter != NULL; gIter = gIter->next) { pcmk__colocation_t *constraint = (pcmk__colocation_t *) gIter->data; if (pcmk__colocation_has_influence(constraint, NULL)) { pe_resource_t *dependent = constraint->dependent; const char *attr = constraint->node_attribute; const float factor = constraint->score / (float) INFINITY; const uint32_t flags = pcmk__coloc_select_active |pcmk__coloc_select_nonnegative; pcmk__add_colocated_node_scores(dependent, rsc->id, &rsc->allowed_nodes, attr, factor, flags); } } pe__show_node_weights(!pcmk_is_set(rsc->cluster->flags, pe_flag_show_scores), rsc, __func__, rsc->allowed_nodes, rsc->cluster); rsc->children = g_list_sort(rsc->children, pcmk__cmp_instance); pcmk__assign_instances(rsc, rsc->children, clone_data->clone_max, clone_data->clone_node_max); if (pcmk_is_set(rsc->flags, pe_rsc_promotable)) { pcmk__set_instance_roles(rsc); } pe__clear_resource_flags(rsc, pe_rsc_provisional|pe_rsc_allocating); pe_rsc_trace(rsc, "Done allocating %s", rsc->id); return NULL; } static pe_action_t * find_rsc_action(pe_resource_t *rsc, const char *task) { pe_action_t *match = NULL; GList *actions = pe__resource_actions(rsc, NULL, task, FALSE); for (GList *item = actions; item != NULL; item = item->next) { pe_action_t *op = (pe_action_t *) item->data; if (!pcmk_is_set(op->flags, pe_action_optional)) { if (match != NULL) { // More than one match, don't return any match = NULL; break; } match = op; } } g_list_free(actions); return match; } static void child_ordering_constraints(pe_resource_t * rsc, pe_working_set_t * data_set) { pe_action_t *stop = NULL; pe_action_t *start = NULL; pe_action_t *last_stop = NULL; pe_action_t *last_start = NULL; GList *gIter = NULL; if (!pe__clone_is_ordered(rsc)) { return; } /* we have to maintain a consistent sorted child list when building order constraints */ rsc->children = g_list_sort(rsc->children, pcmk__cmp_instance_number); for (gIter = rsc->children; gIter != NULL; gIter = gIter->next) { pe_resource_t *child = (pe_resource_t *) gIter->data; stop = find_rsc_action(child, RSC_STOP); if (stop) { if (last_stop) { /* child/child relative stop */ order_actions(stop, last_stop, pe_order_optional); } last_stop = stop; } start = find_rsc_action(child, RSC_START); if (start) { if (last_start) { /* child/child relative start */ order_actions(last_start, start, pe_order_optional); } last_start = start; } } } void clone_create_actions(pe_resource_t *rsc) { clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, rsc); pe_rsc_debug(rsc, "Creating actions for clone %s", rsc->id); pcmk__create_instance_actions(rsc, rsc->children, &clone_data->start_notify, &clone_data->stop_notify); child_ordering_constraints(rsc, rsc->cluster); if (pcmk_is_set(rsc->flags, pe_rsc_promotable)) { pcmk__create_promotable_actions(rsc); } } void clone_internal_constraints(pe_resource_t *rsc) { pe_resource_t *last_rsc = NULL; GList *gIter; bool ordered = pe__clone_is_ordered(rsc); pe_rsc_trace(rsc, "Internal constraints for %s", rsc->id); pcmk__order_resource_actions(rsc, RSC_STOPPED, rsc, RSC_START, pe_order_optional); pcmk__order_resource_actions(rsc, RSC_START, rsc, RSC_STARTED, pe_order_runnable_left); pcmk__order_resource_actions(rsc, RSC_STOP, rsc, RSC_STOPPED, pe_order_runnable_left); if (pcmk_is_set(rsc->flags, pe_rsc_promotable)) { pcmk__order_resource_actions(rsc, RSC_DEMOTED, rsc, RSC_STOP, pe_order_optional); pcmk__order_resource_actions(rsc, RSC_STARTED, rsc, RSC_PROMOTE, pe_order_runnable_left); } if (ordered) { /* we have to maintain a consistent sorted child list when building order constraints */ rsc->children = g_list_sort(rsc->children, pcmk__cmp_instance_number); } for (gIter = rsc->children; gIter != NULL; gIter = gIter->next) { pe_resource_t *child_rsc = (pe_resource_t *) gIter->data; child_rsc->cmds->internal_constraints(child_rsc); pcmk__order_starts(rsc, child_rsc, pe_order_runnable_left|pe_order_implies_first_printed); pcmk__order_resource_actions(child_rsc, RSC_START, rsc, RSC_STARTED, pe_order_implies_then_printed); if (ordered && (last_rsc != NULL)) { pcmk__order_starts(last_rsc, child_rsc, pe_order_optional); } pcmk__order_stops(rsc, child_rsc, pe_order_implies_first_printed); pcmk__order_resource_actions(child_rsc, RSC_STOP, rsc, RSC_STOPPED, pe_order_implies_then_printed); if (ordered && (last_rsc != NULL)) { pcmk__order_stops(child_rsc, last_rsc, pe_order_optional); } last_rsc = child_rsc; } if (pcmk_is_set(rsc->flags, pe_rsc_promotable)) { pcmk__order_promotable_instances(rsc); } } /*! * \internal * \brief Apply a colocation's score to node weights or resource priority * * Given a colocation constraint, apply its score to the dependent's * allowed node weights (if we are still placing resources) or priority (if * we are choosing promotable clone instance roles). * * \param[in,out] dependent Dependent resource in colocation * \param[in] primary Primary resource in colocation * \param[in] colocation Colocation constraint to apply * \param[in] for_dependent true if called on behalf of dependent */ void pcmk__clone_apply_coloc_score(pe_resource_t *dependent, const pe_resource_t *primary, const pcmk__colocation_t *colocation, bool for_dependent) { GList *gIter = NULL; gboolean do_interleave = FALSE; const char *interleave_s = NULL; /* This should never be called for the clone itself as a dependent. Instead, * we add its colocation constraints to its instances and call the * apply_coloc_score() for the instances as dependents. */ CRM_ASSERT(!for_dependent); CRM_CHECK((colocation != NULL) && (dependent != NULL) && (primary != NULL), return); CRM_CHECK(dependent->variant == pe_native, return); pe_rsc_trace(primary, "Processing constraint %s: %s -> %s %d", colocation->id, dependent->id, primary->id, colocation->score); if (pcmk_is_set(primary->flags, pe_rsc_promotable)) { if (pcmk_is_set(primary->flags, pe_rsc_provisional)) { // We haven't placed the primary yet, so we can't apply colocation pe_rsc_trace(primary, "%s is still provisional", primary->id); return; } else if (colocation->primary_role == RSC_ROLE_UNKNOWN) { // This isn't a role-specfic colocation, so handle normally pe_rsc_trace(primary, "Handling %s as a clone colocation", colocation->id); } else if (pcmk_is_set(dependent->flags, pe_rsc_provisional)) { // We're placing the dependent pcmk__update_dependent_with_promotable(primary, dependent, colocation); return; } else if (colocation->dependent_role == RSC_ROLE_PROMOTED) { // We're choosing roles for the dependent pcmk__update_promotable_dependent_priority(primary, dependent, colocation); return; } } // Only the dependent needs to be marked for interleave interleave_s = g_hash_table_lookup(colocation->dependent->meta, XML_RSC_ATTR_INTERLEAVE); if (crm_is_true(interleave_s) && (colocation->dependent->variant > pe_group)) { /* @TODO Do we actually care about multiple primary copies sharing a * dependent copy anymore? */ if (copies_per_node(colocation->dependent) != copies_per_node(colocation->primary)) { pcmk__config_err("Cannot interleave %s and %s because they do not " "support the same number of instances per node", colocation->dependent->id, colocation->primary->id); } else { do_interleave = TRUE; } } if (pcmk_is_set(primary->flags, pe_rsc_provisional)) { pe_rsc_trace(primary, "%s is still provisional", primary->id); return; } else if (do_interleave) { pe_resource_t *primary_instance = NULL; primary_instance = pcmk__find_compatible_instance(dependent, primary, RSC_ROLE_UNKNOWN, false); if (primary_instance != NULL) { pe_rsc_debug(primary, "Pairing %s with %s", dependent->id, primary_instance->id); dependent->cmds->apply_coloc_score(dependent, primary_instance, colocation, true); } else if (colocation->score >= INFINITY) { crm_notice("Cannot pair %s with instance of %s", dependent->id, primary->id); pcmk__assign_resource(dependent, NULL, true); } else { pe_rsc_debug(primary, "Cannot pair %s with instance of %s", dependent->id, primary->id); } return; } else if (colocation->score >= INFINITY) { GList *affected_nodes = NULL; gIter = primary->children; for (; gIter != NULL; gIter = gIter->next) { pe_resource_t *child_rsc = (pe_resource_t *) gIter->data; pe_node_t *chosen = child_rsc->fns->location(child_rsc, NULL, FALSE); if (chosen != NULL && is_set_recursive(child_rsc, pe_rsc_block, TRUE) == FALSE) { pe_rsc_trace(primary, "Allowing %s: %s %d", colocation->id, pe__node_name(chosen), chosen->weight); affected_nodes = g_list_prepend(affected_nodes, chosen); } } node_list_exclude(dependent->allowed_nodes, affected_nodes, FALSE); g_list_free(affected_nodes); return; } gIter = primary->children; for (; gIter != NULL; gIter = gIter->next) { pe_resource_t *child_rsc = (pe_resource_t *) gIter->data; child_rsc->cmds->apply_coloc_score(dependent, child_rsc, colocation, false); } } // Clone implementation of resource_alloc_functions_t:with_this_colocations() void pcmk__with_clone_colocations(const pe_resource_t *rsc, const pe_resource_t *orig_rsc, GList **list) { CRM_CHECK((rsc != NULL) && (orig_rsc != NULL) && (list != NULL), return); - // @COMPAT with previous (incorrect) behavior - if ((rsc != orig_rsc) - && pcmk_is_set(rsc->flags, pe_rsc_provisional) - && !pcmk_is_set(rsc->flags, pe_rsc_allocating)) { - return; // Instances have not yet received clone colocations - } - if (rsc == orig_rsc) { // Colocations are wanted for clone itself pcmk__add_with_this_list(list, rsc->rsc_cons_lhs); } else { pcmk__add_collective_constraints(list, orig_rsc, rsc, true); } } // Clone implementation of resource_alloc_functions_t:this_with_colocations() void pcmk__clone_with_colocations(const pe_resource_t *rsc, const pe_resource_t *orig_rsc, GList **list) { CRM_CHECK((rsc != NULL) && (orig_rsc != NULL) && (list != NULL), return); - // @COMPAT with previous (incorrect) behavior - if ((rsc != orig_rsc) - && pcmk_is_set(rsc->flags, pe_rsc_provisional) - && !pcmk_is_set(rsc->flags, pe_rsc_allocating)) { - return; // Instances have not yet received clone colocations - } - if (rsc == orig_rsc) { // Colocations are wanted for clone itself pcmk__add_this_with_list(list, rsc->rsc_cons); } else { pcmk__add_collective_constraints(list, orig_rsc, rsc, false); } } enum pe_action_flags clone_action_flags(pe_action_t *action, const pe_node_t *node) { return pcmk__collective_action_flags(action, action->rsc->children, node); } void clone_rsc_location(pe_resource_t *rsc, pe__location_t *constraint) { GList *gIter = rsc->children; pe_rsc_trace(rsc, "Processing location constraint %s for %s", constraint->id, rsc->id); pcmk__apply_location(rsc, constraint); for (; gIter != NULL; gIter = gIter->next) { pe_resource_t *child_rsc = (pe_resource_t *) gIter->data; child_rsc->cmds->apply_location(child_rsc, constraint); } } /*! * \internal * \brief Add a resource's actions to the transition graph * * \param[in,out] rsc Resource whose actions should be added */ void clone_expand(pe_resource_t *rsc) { GList *gIter = NULL; clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, rsc); g_list_foreach(rsc->actions, (GFunc) rsc->cmds->action_flags, NULL); pe__create_notifications(rsc, clone_data->start_notify); pe__create_notifications(rsc, clone_data->stop_notify); pe__create_notifications(rsc, clone_data->promote_notify); pe__create_notifications(rsc, clone_data->demote_notify); /* Now that the notifcations have been created we can expand the children */ gIter = rsc->children; for (; gIter != NULL; gIter = gIter->next) { pe_resource_t *child_rsc = (pe_resource_t *) gIter->data; child_rsc->cmds->add_actions_to_graph(child_rsc); } pcmk__add_rsc_actions_to_graph(rsc); /* The notifications are in the graph now, we can destroy the notify_data */ pe__free_notification_data(clone_data->demote_notify); clone_data->demote_notify = NULL; pe__free_notification_data(clone_data->stop_notify); clone_data->stop_notify = NULL; pe__free_notification_data(clone_data->start_notify); clone_data->start_notify = NULL; pe__free_notification_data(clone_data->promote_notify); clone_data->promote_notify = NULL; } // Check whether a resource or any of its children is known on node static bool rsc_known_on(const pe_resource_t *rsc, const pe_node_t *node) { if (rsc->children) { for (GList *child_iter = rsc->children; child_iter != NULL; child_iter = child_iter->next) { pe_resource_t *child = (pe_resource_t *) child_iter->data; if (rsc_known_on(child, node)) { return TRUE; } } } else if (rsc->known_on) { GHashTableIter iter; pe_node_t *known_node = NULL; g_hash_table_iter_init(&iter, rsc->known_on); while (g_hash_table_iter_next(&iter, NULL, (gpointer *) &known_node)) { if (node->details == known_node->details) { return TRUE; } } } return FALSE; } // Look for an instance of clone that is known on node static pe_resource_t * find_instance_on(const pe_resource_t *clone, const pe_node_t *node) { for (GList *gIter = clone->children; gIter != NULL; gIter = gIter->next) { pe_resource_t *child = (pe_resource_t *) gIter->data; if (rsc_known_on(child, node)) { return child; } } return NULL; } // For anonymous clones, only a single instance needs to be probed static bool probe_anonymous_clone(pe_resource_t *rsc, pe_node_t *node, pe_working_set_t *data_set) { // First, check if we probed an instance on this node last time pe_resource_t *child = find_instance_on(rsc, node); // Otherwise, check if we plan to start an instance on this node if (child == NULL) { for (GList *child_iter = rsc->children; child_iter && !child; child_iter = child_iter->next) { pe_node_t *local_node = NULL; pe_resource_t *child_rsc = (pe_resource_t *) child_iter->data; if (child_rsc) { /* make clang analyzer happy */ local_node = child_rsc->fns->location(child_rsc, NULL, FALSE); if (local_node && (local_node->details == node->details)) { child = child_rsc; } } } } // Otherwise, use the first clone instance if (child == NULL) { child = rsc->children->data; } CRM_ASSERT(child); return child->cmds->create_probe(child, node); } /*! * \internal * * \brief Schedule any probes needed for a resource on a node * * \param[in,out] rsc Resource to create probe for * \param[in,out] node Node to create probe on * * \return true if any probe was created, otherwise false */ bool clone_create_probe(pe_resource_t *rsc, pe_node_t *node) { CRM_ASSERT(rsc); rsc->children = g_list_sort(rsc->children, pcmk__cmp_instance_number); if (rsc->children == NULL) { pe_warn("Clone %s has no children", rsc->id); return false; } if (rsc->exclusive_discover) { pe_node_t *allowed = g_hash_table_lookup(rsc->allowed_nodes, node->details->id); if (allowed && allowed->rsc_discover_mode != pe_discover_exclusive) { /* exclusive discover is enabled and this node is not marked * as a node this resource should be discovered on * * remove the node from allowed_nodes so that the * notification contains only nodes that we might ever run * on */ g_hash_table_remove(rsc->allowed_nodes, node->details->id); /* Bit of a shortcut - might as well take it */ return false; } } if (pcmk_is_set(rsc->flags, pe_rsc_unique)) { return pcmk__probe_resource_list(rsc->children, node); } else { return probe_anonymous_clone(rsc, node, rsc->cluster); } } void clone_append_meta(const pe_resource_t *rsc, xmlNode *xml) { char *name = NULL; clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, rsc); name = crm_meta_name(XML_RSC_ATTR_UNIQUE); crm_xml_add(xml, name, pe__rsc_bool_str(rsc, pe_rsc_unique)); free(name); name = crm_meta_name(XML_RSC_ATTR_NOTIFY); crm_xml_add(xml, name, pe__rsc_bool_str(rsc, pe_rsc_notify)); free(name); name = crm_meta_name(XML_RSC_ATTR_INCARNATION_MAX); crm_xml_add_int(xml, name, clone_data->clone_max); free(name); name = crm_meta_name(XML_RSC_ATTR_INCARNATION_NODEMAX); crm_xml_add_int(xml, name, clone_data->clone_node_max); free(name); if (pcmk_is_set(rsc->flags, pe_rsc_promotable)) { int promoted_max = pe__clone_promoted_max(rsc); int promoted_node_max = pe__clone_promoted_node_max(rsc); name = crm_meta_name(XML_RSC_ATTR_PROMOTED_MAX); crm_xml_add_int(xml, name, promoted_max); free(name); name = crm_meta_name(XML_RSC_ATTR_PROMOTED_NODEMAX); crm_xml_add_int(xml, name, promoted_node_max); free(name); /* @COMPAT Maintain backward compatibility with resource agents that * expect the old names (deprecated since 2.0.0). */ name = crm_meta_name(PCMK_XA_PROMOTED_MAX_LEGACY); crm_xml_add_int(xml, name, promoted_max); free(name); name = crm_meta_name(PCMK_XA_PROMOTED_NODE_MAX_LEGACY); crm_xml_add_int(xml, name, promoted_node_max); free(name); } } // Clone implementation of resource_alloc_functions_t:add_utilization() void pcmk__clone_add_utilization(const pe_resource_t *rsc, const pe_resource_t *orig_rsc, GList *all_rscs, GHashTable *utilization) { bool existing = false; pe_resource_t *child = NULL; if (!pcmk_is_set(rsc->flags, pe_rsc_provisional)) { return; } // Look for any child already existing in the list for (GList *iter = rsc->children; iter != NULL; iter = iter->next) { child = (pe_resource_t *) iter->data; if (g_list_find(all_rscs, child)) { existing = true; // Keep checking remaining children } else { // If this is a clone of a group, look for group's members for (GList *member_iter = child->children; member_iter != NULL; member_iter = member_iter->next) { pe_resource_t *member = (pe_resource_t *) member_iter->data; if (g_list_find(all_rscs, member) != NULL) { // Add *child's* utilization, not group member's child->cmds->add_utilization(child, orig_rsc, all_rscs, utilization); existing = true; break; } } } } if (!existing && (rsc->children != NULL)) { // If nothing was found, still add first child's utilization child = (pe_resource_t *) rsc->children->data; child->cmds->add_utilization(child, orig_rsc, all_rscs, utilization); } } // Clone implementation of resource_alloc_functions_t:shutdown_lock() void pcmk__clone_shutdown_lock(pe_resource_t *rsc) { return; // Clones currently don't support shutdown locks }