/* * Copyright Elasticsearch B.V. and/or licensed to Elasticsearch B.V. under one * or more contributor license agreements. Licensed under the "Elastic License * 2.0", the "GNU Affero General Public License v3.0 only", and the "Server Side * Public License v 1"; you may not use this file except in compliance with, at * your election, the "Elastic License 2.0", the "GNU Affero General Public * License v3.0 only", or the "Server Side Public License, v 1". */ package org.elasticsearch.action.search; import org.apache.logging.log4j.LogManager; import org.apache.logging.log4j.Logger; import org.elasticsearch.ExceptionsHelper; import org.elasticsearch.TransportVersions; import org.elasticsearch.action.ActionListener; import org.elasticsearch.action.ActionListenerResponseHandler; import org.elasticsearch.action.ActionType; import org.elasticsearch.action.DelegatingActionListener; import org.elasticsearch.action.IndicesRequest; import org.elasticsearch.action.OriginalIndices; import org.elasticsearch.action.RemoteClusterActionType; import org.elasticsearch.action.ResolvedIndices; import org.elasticsearch.action.ShardOperationFailedException; import org.elasticsearch.action.admin.cluster.shards.ClusterSearchShardsRequest; import org.elasticsearch.action.admin.cluster.shards.ClusterSearchShardsResponse; import org.elasticsearch.action.admin.cluster.shards.TransportClusterSearchShardsAction; import org.elasticsearch.action.admin.cluster.stats.CCSUsage; import org.elasticsearch.action.admin.cluster.stats.CCSUsageTelemetry; import org.elasticsearch.action.support.ActionFilters; import org.elasticsearch.action.support.HandledTransportAction; import org.elasticsearch.action.support.IndicesOptions; import org.elasticsearch.action.support.master.MasterNodeRequest; import org.elasticsearch.client.internal.Client; import org.elasticsearch.cluster.ClusterState; import org.elasticsearch.cluster.ProjectState; import org.elasticsearch.cluster.block.ClusterBlockException; import org.elasticsearch.cluster.block.ClusterBlockLevel; import org.elasticsearch.cluster.metadata.IndexAbstraction; import org.elasticsearch.cluster.metadata.IndexMetadata; import org.elasticsearch.cluster.metadata.IndexNameExpressionResolver; import org.elasticsearch.cluster.metadata.IndexNameExpressionResolver.ResolvedExpression; import org.elasticsearch.cluster.metadata.ProjectMetadata; import org.elasticsearch.cluster.node.DiscoveryNode; import org.elasticsearch.cluster.node.DiscoveryNodes; import org.elasticsearch.cluster.project.ProjectResolver; import org.elasticsearch.cluster.routing.OperationRouting; import org.elasticsearch.cluster.routing.ShardIterator; import org.elasticsearch.cluster.routing.ShardRouting; import org.elasticsearch.cluster.service.ClusterService; import org.elasticsearch.common.Strings; import org.elasticsearch.common.breaker.CircuitBreaker; import org.elasticsearch.common.io.stream.NamedWriteableRegistry; import org.elasticsearch.common.logging.DeprecationCategory; import org.elasticsearch.common.logging.DeprecationLogger; import org.elasticsearch.common.regex.Regex; import org.elasticsearch.common.settings.Setting; import org.elasticsearch.common.settings.Setting.Property; import org.elasticsearch.common.util.ArrayUtils; import org.elasticsearch.common.util.CollectionUtils; import org.elasticsearch.common.util.Maps; import org.elasticsearch.common.util.concurrent.CountDown; import org.elasticsearch.common.util.concurrent.EsExecutors; import org.elasticsearch.core.TimeValue; import org.elasticsearch.index.Index; import org.elasticsearch.index.IndexNotFoundException; import org.elasticsearch.index.query.QueryBuilder; import org.elasticsearch.index.query.Rewriteable; import org.elasticsearch.index.shard.ShardId; import org.elasticsearch.index.shard.ShardNotFoundException; import org.elasticsearch.indices.ExecutorSelector; import org.elasticsearch.indices.breaker.CircuitBreakerService; import org.elasticsearch.injection.guice.Inject; import org.elasticsearch.node.ResponseCollectorService; import org.elasticsearch.rest.action.search.SearchResponseMetrics; import org.elasticsearch.search.SearchPhaseResult; import org.elasticsearch.search.SearchService; import org.elasticsearch.search.aggregations.AggregationReduceContext; import org.elasticsearch.search.builder.PointInTimeBuilder; import org.elasticsearch.search.builder.SearchSourceBuilder; import org.elasticsearch.search.internal.AliasFilter; import org.elasticsearch.search.internal.SearchContext; import org.elasticsearch.search.profile.SearchProfileResults; import org.elasticsearch.search.profile.SearchProfileShardResult; import org.elasticsearch.tasks.Task; import org.elasticsearch.tasks.TaskId; import org.elasticsearch.threadpool.ThreadPool; import org.elasticsearch.transport.RemoteClusterAware; import org.elasticsearch.transport.RemoteClusterService; import org.elasticsearch.transport.RemoteTransportException; import org.elasticsearch.transport.Transport; import org.elasticsearch.transport.TransportRequestOptions; import org.elasticsearch.transport.TransportService; import org.elasticsearch.usage.UsageService; import org.elasticsearch.xcontent.ToXContent; import org.elasticsearch.xcontent.XContentFactory; import java.io.IOException; import java.util.ArrayList; import java.util.Arrays; import java.util.Collections; import java.util.HashMap; import java.util.HashSet; import java.util.List; import java.util.Map; import java.util.Objects; import java.util.Set; import java.util.concurrent.ConcurrentHashMap; import java.util.concurrent.Executor; import java.util.concurrent.TimeUnit; import java.util.concurrent.atomic.AtomicReference; import java.util.function.BiConsumer; import java.util.function.BiFunction; import java.util.function.Function; import java.util.function.LongSupplier; import java.util.stream.Collectors; import static org.elasticsearch.action.search.SearchType.DFS_QUERY_THEN_FETCH; import static org.elasticsearch.action.search.SearchType.QUERY_THEN_FETCH; import static org.elasticsearch.action.search.TransportSearchHelper.checkCCSVersionCompatibility; import static org.elasticsearch.search.sort.FieldSortBuilder.hasPrimaryFieldSort; import static org.elasticsearch.threadpool.ThreadPool.Names.SYSTEM_CRITICAL_READ; import static org.elasticsearch.threadpool.ThreadPool.Names.SYSTEM_READ; public class TransportSearchAction extends HandledTransportAction { public static final String NAME = "indices:data/read/search"; public static final ActionType TYPE = new ActionType<>(NAME); public static final RemoteClusterActionType REMOTE_TYPE = new RemoteClusterActionType<>(NAME, SearchResponse::new); private static final Logger logger = LogManager.getLogger(TransportSearchAction.class); private static final DeprecationLogger DEPRECATION_LOGGER = DeprecationLogger.getLogger(TransportSearchAction.class); public static final String FROZEN_INDICES_DEPRECATION_MESSAGE = "Searching frozen indices [{}] is deprecated." + " Consider cold or frozen tiers in place of frozen indices. The frozen feature will be removed in a feature release."; /** The maximum number of shards for a single search request. */ public static final Setting SHARD_COUNT_LIMIT_SETTING = Setting.longSetting( "action.search.shard_count.limit", Long.MAX_VALUE, 1L, Property.Dynamic, Property.NodeScope ); public static final Setting DEFAULT_PRE_FILTER_SHARD_SIZE = Setting.intSetting( "action.search.pre_filter_shard_size.default", SearchRequest.DEFAULT_PRE_FILTER_SHARD_SIZE, 1, Property.NodeScope ); private final ThreadPool threadPool; private final ClusterService clusterService; private final TransportService transportService; private final SearchTransportService searchTransportService; private final RemoteClusterService remoteClusterService; private final SearchPhaseController searchPhaseController; private final SearchService searchService; private final ProjectResolver projectResolver; private final ResponseCollectorService responseCollectorService; private final IndexNameExpressionResolver indexNameExpressionResolver; private final NamedWriteableRegistry namedWriteableRegistry; private final CircuitBreaker circuitBreaker; private final ExecutorSelector executorSelector; private final int defaultPreFilterShardSize; private final boolean ccsCheckCompatibility; private final SearchResponseMetrics searchResponseMetrics; private final Client client; private final UsageService usageService; private final boolean collectTelemetry; @Inject public TransportSearchAction( ThreadPool threadPool, CircuitBreakerService circuitBreakerService, TransportService transportService, SearchService searchService, ResponseCollectorService responseCollectorService, SearchTransportService searchTransportService, SearchPhaseController searchPhaseController, ClusterService clusterService, ActionFilters actionFilters, ProjectResolver projectResolver, IndexNameExpressionResolver indexNameExpressionResolver, NamedWriteableRegistry namedWriteableRegistry, ExecutorSelector executorSelector, SearchResponseMetrics searchResponseMetrics, Client client, UsageService usageService ) { super(TYPE.name(), transportService, actionFilters, SearchRequest::new, EsExecutors.DIRECT_EXECUTOR_SERVICE); this.threadPool = threadPool; this.circuitBreaker = circuitBreakerService.getBreaker(CircuitBreaker.REQUEST); this.searchPhaseController = searchPhaseController; this.searchTransportService = searchTransportService; this.remoteClusterService = searchTransportService.getRemoteClusterService(); SearchTransportService.registerRequestHandler(transportService, searchService); SearchQueryThenFetchAsyncAction.registerNodeSearchAction( searchTransportService, searchService, searchPhaseController, namedWriteableRegistry ); this.clusterService = clusterService; this.transportService = transportService; this.searchService = searchService; this.projectResolver = projectResolver; this.responseCollectorService = responseCollectorService; this.indexNameExpressionResolver = indexNameExpressionResolver; this.namedWriteableRegistry = namedWriteableRegistry; this.executorSelector = executorSelector; var settings = clusterService.getSettings(); this.defaultPreFilterShardSize = DEFAULT_PRE_FILTER_SHARD_SIZE.get(settings); this.ccsCheckCompatibility = SearchService.CCS_VERSION_CHECK_SETTING.get(settings); this.collectTelemetry = SearchService.CCS_COLLECT_TELEMETRY.get(settings); this.searchResponseMetrics = searchResponseMetrics; this.client = client; this.usageService = usageService; } private Map buildPerIndexOriginalIndices( ProjectState projectState, Set indicesAndAliases, String[] indices, IndicesOptions indicesOptions ) { Map res = Maps.newMapWithExpectedSize(indices.length); var blocks = projectState.blocks(); var projectId = projectState.projectId(); // optimization: mostly we do not have any blocks so there's no point in the expensive per-index checking boolean hasBlocks = blocks.global(projectId).isEmpty() == false || blocks.indices(projectState.projectId()).isEmpty() == false; // Get a distinct set of index abstraction names present from the resolved expressions to help with the reverse resolution from // concrete index to the expression that produced it. Set indicesAndAliasesResources = indicesAndAliases.stream().map(ResolvedExpression::resource).collect(Collectors.toSet()); for (String index : indices) { if (hasBlocks) { blocks.indexBlockedRaiseException(projectId, ClusterBlockLevel.READ, index); } String[] aliases = indexNameExpressionResolver.allIndexAliases(projectState.metadata(), index, indicesAndAliases); String[] finalIndices = Strings.EMPTY_ARRAY; if (aliases == null || aliases.length == 0 || indicesAndAliasesResources.contains(index) || hasDataStreamRef(projectState.metadata(), indicesAndAliasesResources, index)) { finalIndices = new String[] { index }; } if (aliases != null) { finalIndices = finalIndices.length == 0 ? aliases : ArrayUtils.concat(finalIndices, aliases); } res.put(index, new OriginalIndices(finalIndices, indicesOptions)); } return res; } private static boolean hasDataStreamRef(ProjectMetadata project, Set indicesAndAliases, String index) { IndexAbstraction ret = project.getIndicesLookup().get(index); if (ret == null || ret.getParentDataStream() == null) { return false; } return indicesAndAliases.contains(ret.getParentDataStream().getName()); } Map buildIndexAliasFilters( ProjectState projectState, Set indicesAndAliases, Index[] concreteIndices ) { final Map aliasFilterMap = new HashMap<>(); for (Index index : concreteIndices) { projectState.blocks().indexBlockedRaiseException(projectState.projectId(), ClusterBlockLevel.READ, index.getName()); AliasFilter aliasFilter = searchService.buildAliasFilter(projectState, index.getName(), indicesAndAliases); assert aliasFilter != null; aliasFilterMap.put(index.getUUID(), aliasFilter); } return aliasFilterMap; } private Map resolveIndexBoosts(SearchRequest searchRequest, ClusterState clusterState) { if (searchRequest.source() == null) { return Collections.emptyMap(); } SearchSourceBuilder source = searchRequest.source(); if (source.indexBoosts() == null) { return Collections.emptyMap(); } Map concreteIndexBoosts = new HashMap<>(); for (SearchSourceBuilder.IndexBoost ib : source.indexBoosts()) { Index[] concreteIndices = indexNameExpressionResolver.concreteIndices( clusterState, searchRequest.indicesOptions(), ib.getIndex() ); for (Index concreteIndex : concreteIndices) { concreteIndexBoosts.putIfAbsent(concreteIndex.getUUID(), ib.getBoost()); } } return Collections.unmodifiableMap(concreteIndexBoosts); } /** * Search operations need two clocks. One clock is to fulfill real clock needs (e.g., resolving * "now" to an index name). Another clock is needed for measuring how long a search operation * took. These two uses are at odds with each other. There are many issues with using a real * clock for measuring how long an operation took (they often lack precision, they are subject * to moving backwards due to NTP and other such complexities, etc.). There are also issues with * using a relative clock for reporting real time. Thus, we simply separate these two uses. */ public record SearchTimeProvider(long absoluteStartMillis, long relativeStartNanos, LongSupplier relativeCurrentNanosProvider) { /** * Instantiates a new search time provider. The absolute start time is the real clock time * used for resolving index expressions that include dates. The relative start time is the * start of the search operation according to a relative clock. The total time the search * operation took can be measured against the provided relative clock and the relative start * time. * * @param absoluteStartMillis the absolute start time in milliseconds since the epoch * @param relativeStartNanos the relative start time in nanoseconds * @param relativeCurrentNanosProvider provides the current relative time */ public SearchTimeProvider {} public long buildTookInMillis() { return TimeUnit.NANOSECONDS.toMillis(relativeCurrentNanosProvider.getAsLong() - relativeStartNanos); } } @Override protected void doExecute(Task task, SearchRequest searchRequest, ActionListener listener) { executeRequest((SearchTask) task, searchRequest, new SearchResponseActionListener(listener), AsyncSearchActionProvider::new); } void executeRequest( SearchTask task, SearchRequest original, ActionListener listener, Function, SearchPhaseProvider> searchPhaseProvider ) { final long relativeStartNanos = System.nanoTime(); final SearchTimeProvider timeProvider = new SearchTimeProvider( original.getOrCreateAbsoluteStartMillis(), relativeStartNanos, System::nanoTime ); final ClusterState clusterState = clusterService.state(); clusterState.blocks().globalBlockedRaiseException(projectResolver.getProjectId(), ClusterBlockLevel.READ); ProjectState projectState = projectResolver.getProjectState(clusterState); final ResolvedIndices resolvedIndices; if (original.pointInTimeBuilder() != null) { resolvedIndices = ResolvedIndices.resolveWithPIT( original.pointInTimeBuilder(), original.indicesOptions(), projectState.metadata(), namedWriteableRegistry ); } else { resolvedIndices = ResolvedIndices.resolveWithIndicesRequest( original, projectState.metadata(), indexNameExpressionResolver, remoteClusterService, timeProvider.absoluteStartMillis() ); frozenIndexCheck(resolvedIndices); } ActionListener rewriteListener = listener.delegateFailureAndWrap((delegate, rewritten) -> { if (ccsCheckCompatibility) { checkCCSVersionCompatibility(rewritten); } if (resolvedIndices.getRemoteClusterIndices().isEmpty()) { executeLocalSearch( task, timeProvider, rewritten, resolvedIndices, projectState, SearchResponse.Clusters.EMPTY, searchPhaseProvider.apply(delegate) ); } else { if (delegate instanceof TelemetryListener tl) { tl.setRemotes(resolvedIndices.getRemoteClusterIndices().size()); if (task.isAsync()) { tl.setFeature(CCSUsageTelemetry.ASYNC_FEATURE); } if (original.pointInTimeBuilder() != null) { tl.setFeature(CCSUsageTelemetry.PIT_FEATURE); } tl.setClient(task); // Check if any of the index patterns are wildcard patterns var localIndices = resolvedIndices.getLocalIndices(); if (localIndices != null && Arrays.stream(localIndices.indices()).anyMatch(Regex::isSimpleMatchPattern)) { tl.setFeature(CCSUsageTelemetry.WILDCARD_FEATURE); } if (resolvedIndices.getRemoteClusterIndices() .values() .stream() .anyMatch(indices -> Arrays.stream(indices.indices()).anyMatch(Regex::isSimpleMatchPattern))) { tl.setFeature(CCSUsageTelemetry.WILDCARD_FEATURE); } } final TaskId parentTaskId = task.taskInfo(clusterService.localNode().getId(), false).taskId(); if (shouldMinimizeRoundtrips(rewritten)) { if (delegate instanceof TelemetryListener tl) { tl.setFeature(CCSUsageTelemetry.MRT_FEATURE); } final AggregationReduceContext.Builder aggregationReduceContextBuilder = rewritten.source() != null && rewritten.source().aggregations() != null ? searchService.aggReduceContextBuilder(task::isCancelled, rewritten.source().aggregations()) : null; SearchResponse.Clusters clusters = new SearchResponse.Clusters( resolvedIndices.getLocalIndices(), resolvedIndices.getRemoteClusterIndices(), true, remoteClusterService::isSkipUnavailable ); if (resolvedIndices.getLocalIndices() == null) { // Notify the progress listener that a CCS with minimize_roundtrips is happening remote-only (no local shards) task.getProgressListener() .notifyListShards(Collections.emptyList(), Collections.emptyList(), clusters, false, timeProvider); } ccsRemoteReduce( task, parentTaskId, rewritten, resolvedIndices, clusters, timeProvider, aggregationReduceContextBuilder, remoteClusterService, threadPool, delegate, (r, l) -> executeLocalSearch( task, timeProvider, r, resolvedIndices, projectState, clusters, searchPhaseProvider.apply(l) ) ); } else { final SearchContextId searchContext = resolvedIndices.getSearchContextId(); SearchResponse.Clusters clusters = new SearchResponse.Clusters( resolvedIndices.getLocalIndices(), resolvedIndices.getRemoteClusterIndices(), false, remoteClusterService::isSkipUnavailable ); // TODO: pass parentTaskId collectSearchShards( rewritten.indicesOptions(), rewritten.preference(), rewritten.routing(), rewritten.source() != null ? rewritten.source().query() : null, Objects.requireNonNullElse(rewritten.allowPartialSearchResults(), searchService.defaultAllowPartialSearchResults()), searchContext, resolvedIndices.getRemoteClusterIndices(), clusters, timeProvider, transportService, delegate.delegateFailureAndWrap((finalDelegate, searchShardsResponses) -> { final BiFunction clusterNodeLookup = getRemoteClusterNodeLookup( searchShardsResponses ); final Map remoteAliasFilters; final List remoteShardIterators; if (searchContext != null) { remoteAliasFilters = searchContext.aliasFilter(); remoteShardIterators = getRemoteShardsIteratorFromPointInTime( searchShardsResponses, searchContext, rewritten.pointInTimeBuilder().getKeepAlive(), resolvedIndices.getRemoteClusterIndices() ); } else { remoteAliasFilters = new HashMap<>(); for (SearchShardsResponse searchShardsResponse : searchShardsResponses.values()) { remoteAliasFilters.putAll(searchShardsResponse.getAliasFilters()); } remoteShardIterators = getRemoteShardsIterator( searchShardsResponses, resolvedIndices.getRemoteClusterIndices(), remoteAliasFilters ); } executeSearch( task, timeProvider, rewritten, resolvedIndices, remoteShardIterators, clusterNodeLookup, projectState, remoteAliasFilters, clusters, searchPhaseProvider.apply(finalDelegate) ); }) ); } } }); final SearchSourceBuilder source = original.source(); final boolean isExplain = source != null && source.explain() != null && source.explain(); if (shouldOpenPIT(source)) { // disabling shard reordering for request original.setPreFilterShardSize(Integer.MAX_VALUE); openPIT(client, original, searchService.getDefaultKeepAliveInMillis(), listener.delegateFailureAndWrap((delegate, resp) -> { // We set the keep alive to -1 to indicate that we don't need the pit id in the response. // This is needed since we delete the pit prior to sending the response so the id doesn't exist anymore. source.pointInTimeBuilder(new PointInTimeBuilder(resp.getPointInTimeId()).setKeepAlive(TimeValue.MINUS_ONE)); var pitListener = new SearchResponseActionListener(delegate) { @Override public void onResponse(SearchResponse response) { // we need to close the PIT first so we delay the release of the response to after the closing response.incRef(); closePIT( client, original.source().pointInTimeBuilder(), () -> ActionListener.respondAndRelease(delegate, response) ); } @Override public void onFailure(Exception e) { closePIT(client, original.source().pointInTimeBuilder(), () -> delegate.onFailure(e)); } }; executeRequest(task, original, pitListener, searchPhaseProvider); })); } else { Rewriteable.rewriteAndFetch( original, searchService.getRewriteContext( timeProvider::absoluteStartMillis, resolvedIndices, original.pointInTimeBuilder(), isExplain ), rewriteListener ); } } /** * Returns true if the provided source needs to open a shared point in time prior to executing the request. */ private boolean shouldOpenPIT(SearchSourceBuilder source) { if (source == null) { return false; } if (source.pointInTimeBuilder() != null) { return false; } var retriever = source.retriever(); return retriever != null && retriever.isCompound(); } static void openPIT(Client client, SearchRequest request, long keepAliveMillis, ActionListener listener) { OpenPointInTimeRequest pitReq = new OpenPointInTimeRequest(request.indices()).indicesOptions(request.indicesOptions()) .preference(request.preference()) .routing(request.routing()) .keepAlive(TimeValue.timeValueMillis(keepAliveMillis)); client.execute(TransportOpenPointInTimeAction.TYPE, pitReq, listener); } static void closePIT(Client client, PointInTimeBuilder pit, Runnable next) { client.execute( TransportClosePointInTimeAction.TYPE, new ClosePointInTimeRequest(pit.getEncodedId()), ActionListener.runAfter(new ActionListener<>() { @Override public void onResponse(ClosePointInTimeResponse closePointInTimeResponse) {} @Override public void onFailure(Exception e) {} }, next) ); } static void adjustSearchType(SearchRequest searchRequest, boolean singleShard) { // if there's a kNN search, always use DFS_QUERY_THEN_FETCH if (searchRequest.hasKnnSearch()) { searchRequest.searchType(DFS_QUERY_THEN_FETCH); return; } // if there's only suggest, disable request cache and always use QUERY_THEN_FETCH if (searchRequest.isSuggestOnly()) { searchRequest.requestCache(false); searchRequest.searchType(QUERY_THEN_FETCH); return; } // optimize search type for cases where there is only one shard group to search on if (singleShard) { // if we only have one group, then we always want Q_T_F, no need for DFS, and no need to do THEN since we hit one shard searchRequest.searchType(QUERY_THEN_FETCH); } } public static boolean shouldMinimizeRoundtrips(SearchRequest searchRequest) { if (searchRequest.isCcsMinimizeRoundtrips() == false) { return false; } if (searchRequest.scroll() != null) { return false; } if (searchRequest.pointInTimeBuilder() != null) { return false; } if (searchRequest.searchType() == DFS_QUERY_THEN_FETCH) { return false; } if (searchRequest.hasKnnSearch()) { return false; } SearchSourceBuilder source = searchRequest.source(); return source == null || source.collapse() == null || source.collapse().getInnerHits() == null || source.collapse().getInnerHits().isEmpty(); } /** * Handles ccs_minimize_roundtrips=true */ static void ccsRemoteReduce( SearchTask task, TaskId parentTaskId, SearchRequest searchRequest, ResolvedIndices resolvedIndices, SearchResponse.Clusters clusters, SearchTimeProvider timeProvider, AggregationReduceContext.Builder aggReduceContextBuilder, RemoteClusterService remoteClusterService, ThreadPool threadPool, ActionListener listener, BiConsumer> localSearchConsumer ) { final var remoteClientResponseExecutor = threadPool.executor(ThreadPool.Names.SEARCH_COORDINATION); if (resolvedIndices.getLocalIndices() == null && resolvedIndices.getRemoteClusterIndices().size() == 1) { // if we are searching against a single remote cluster, we simply forward the original search request to such cluster // and we directly perform final reduction in the remote cluster Map.Entry entry = resolvedIndices.getRemoteClusterIndices().entrySet().iterator().next(); String clusterAlias = entry.getKey(); boolean skipUnavailable = remoteClusterService.isSkipUnavailable(clusterAlias); OriginalIndices indices = entry.getValue(); SearchRequest ccsSearchRequest = SearchRequest.subSearchRequest( parentTaskId, searchRequest, indices.indices(), clusterAlias, timeProvider.absoluteStartMillis(), true ); var remoteClusterClient = remoteClusterService.getRemoteClusterClient( clusterAlias, remoteClientResponseExecutor, RemoteClusterService.DisconnectedStrategy.RECONNECT_UNLESS_SKIP_UNAVAILABLE ); remoteClusterClient.execute(TransportSearchAction.REMOTE_TYPE, ccsSearchRequest, new ActionListener<>() { @Override public void onResponse(SearchResponse searchResponse) { // overwrite the existing cluster entry with the updated one ccsClusterInfoUpdate(searchResponse, clusters, clusterAlias, skipUnavailable); Map profileResults = searchResponse.getProfileResults(); SearchProfileResults profile = profileResults == null || profileResults.isEmpty() ? null : new SearchProfileResults(profileResults); ActionListener.respondAndRelease( listener, new SearchResponse( searchResponse.getHits(), searchResponse.getAggregations(), searchResponse.getSuggest(), searchResponse.isTimedOut(), searchResponse.isTerminatedEarly(), profile, searchResponse.getNumReducePhases(), searchResponse.getScrollId(), searchResponse.getTotalShards(), searchResponse.getSuccessfulShards(), searchResponse.getSkippedShards(), timeProvider.buildTookInMillis(), searchResponse.getShardFailures(), clusters, searchResponse.pointInTimeId() ) ); } @Override public void onFailure(Exception e) { ShardSearchFailure failure = new ShardSearchFailure(e); logCCSError(failure, clusterAlias, skipUnavailable); ccsClusterInfoUpdate(failure, clusters, clusterAlias, skipUnavailable); if (skipUnavailable) { ActionListener.respondAndRelease(listener, SearchResponse.empty(timeProvider::buildTookInMillis, clusters)); } else { listener.onFailure(wrapRemoteClusterFailure(clusterAlias, e)); } } }); } else { SearchResponseMerger searchResponseMerger = createSearchResponseMerger( searchRequest.source(), timeProvider, aggReduceContextBuilder ); task.setSearchResponseMergerSupplier( () -> createSearchResponseMerger(searchRequest.source(), timeProvider, aggReduceContextBuilder) ); final AtomicReference exceptions = new AtomicReference<>(); int totalClusters = resolvedIndices.getRemoteClusterIndices().size() + (resolvedIndices.getLocalIndices() == null ? 0 : 1); final CountDown countDown = new CountDown(totalClusters); for (Map.Entry entry : resolvedIndices.getRemoteClusterIndices().entrySet()) { String clusterAlias = entry.getKey(); boolean skipUnavailable = remoteClusterService.isSkipUnavailable(clusterAlias); OriginalIndices indices = entry.getValue(); SearchRequest ccsSearchRequest = SearchRequest.subSearchRequest( parentTaskId, searchRequest, indices.indices(), clusterAlias, timeProvider.absoluteStartMillis(), false ); ActionListener ccsListener = createCCSListener( clusterAlias, skipUnavailable, countDown, exceptions, searchResponseMerger, clusters, task.getProgressListener(), listener ); final var remoteClusterClient = remoteClusterService.getRemoteClusterClient( clusterAlias, remoteClientResponseExecutor, RemoteClusterService.DisconnectedStrategy.RECONNECT_UNLESS_SKIP_UNAVAILABLE ); remoteClusterClient.execute(TransportSearchAction.REMOTE_TYPE, ccsSearchRequest, ccsListener); } if (resolvedIndices.getLocalIndices() != null) { ActionListener ccsListener = createCCSListener( RemoteClusterAware.LOCAL_CLUSTER_GROUP_KEY, false, countDown, exceptions, searchResponseMerger, clusters, task.getProgressListener(), listener ); SearchRequest ccsLocalSearchRequest = SearchRequest.subSearchRequest( parentTaskId, searchRequest, resolvedIndices.getLocalIndices().indices(), RemoteClusterAware.LOCAL_CLUSTER_GROUP_KEY, timeProvider.absoluteStartMillis(), false ); localSearchConsumer.accept(ccsLocalSearchRequest, ccsListener); } } } static SearchResponseMerger createSearchResponseMerger( SearchSourceBuilder source, SearchTimeProvider timeProvider, AggregationReduceContext.Builder aggReduceContextBuilder ) { final int from; final int size; final int trackTotalHitsUpTo; if (source == null) { from = SearchService.DEFAULT_FROM; size = SearchService.DEFAULT_SIZE; trackTotalHitsUpTo = SearchContext.DEFAULT_TRACK_TOTAL_HITS_UP_TO; } else { from = source.from() == -1 ? SearchService.DEFAULT_FROM : source.from(); size = source.size() == -1 ? SearchService.DEFAULT_SIZE : source.size(); trackTotalHitsUpTo = source.trackTotalHitsUpTo() == null ? SearchContext.DEFAULT_TRACK_TOTAL_HITS_UP_TO : source.trackTotalHitsUpTo(); // here we modify the original source so we can re-use it by setting it to each outgoing search request source.from(0); source.size(from + size); } return new SearchResponseMerger(from, size, trackTotalHitsUpTo, timeProvider, aggReduceContextBuilder); } /** * Collect remote search shards that we need to search for potential matches. * Used for ccs_minimize_roundtrips=false */ static void collectSearchShards( IndicesOptions indicesOptions, String preference, String routing, QueryBuilder query, boolean allowPartialResults, SearchContextId searchContext, Map remoteIndicesByCluster, SearchResponse.Clusters clusters, SearchTimeProvider timeProvider, TransportService transportService, ActionListener> listener ) { RemoteClusterService remoteClusterService = transportService.getRemoteClusterService(); final CountDown responsesCountDown = new CountDown(remoteIndicesByCluster.size()); final Map searchShardsResponses = new ConcurrentHashMap<>(); final AtomicReference exceptions = new AtomicReference<>(); for (Map.Entry entry : remoteIndicesByCluster.entrySet()) { final String clusterAlias = entry.getKey(); boolean skipUnavailable = remoteClusterService.isSkipUnavailable(clusterAlias); CCSActionListener> singleListener = new CCSActionListener<>( clusterAlias, skipUnavailable, responsesCountDown, exceptions, clusters, listener ) { @Override void innerOnResponse(SearchShardsResponse searchShardsResponse) { assert ThreadPool.assertCurrentThreadPool(ThreadPool.Names.SEARCH_COORDINATION); ccsClusterInfoUpdate(searchShardsResponse, clusters, clusterAlias, timeProvider); searchShardsResponses.put(clusterAlias, searchShardsResponse); } @Override Map createFinalResponse() { return searchShardsResponses; } }; remoteClusterService.maybeEnsureConnectedAndGetConnection( clusterAlias, skipUnavailable == false, singleListener.delegateFailureAndWrap((delegate, connection) -> { final String[] indices = entry.getValue().indices(); final Executor responseExecutor = transportService.getThreadPool().executor(ThreadPool.Names.SEARCH_COORDINATION); // TODO: support point-in-time if (searchContext == null && connection.getTransportVersion().onOrAfter(TransportVersions.V_8_9_X)) { SearchShardsRequest searchShardsRequest = new SearchShardsRequest( indices, indicesOptions, query, routing, preference, allowPartialResults, clusterAlias ); transportService.sendRequest( connection, TransportSearchShardsAction.TYPE.name(), searchShardsRequest, TransportRequestOptions.EMPTY, new ActionListenerResponseHandler<>(delegate, SearchShardsResponse::new, responseExecutor) ); } else { // does not do a can-match ClusterSearchShardsRequest searchShardsRequest = new ClusterSearchShardsRequest( MasterNodeRequest.INFINITE_MASTER_NODE_TIMEOUT, indices ).indicesOptions(indicesOptions).local(true).preference(preference).routing(routing); transportService.sendRequest( connection, TransportClusterSearchShardsAction.TYPE.name(), searchShardsRequest, TransportRequestOptions.EMPTY, new ActionListenerResponseHandler<>( delegate.map(SearchShardsResponse::fromLegacyResponse), ClusterSearchShardsResponse::new, responseExecutor ) ); } }) ); } } /** * Only used for ccs_minimize_roundtrips=true pathway */ private static ActionListener createCCSListener( String clusterAlias, boolean skipUnavailable, CountDown countDown, AtomicReference exceptions, SearchResponseMerger searchResponseMerger, SearchResponse.Clusters clusters, SearchProgressListener progressListener, ActionListener originalListener ) { return new CCSActionListener<>( clusterAlias, skipUnavailable, countDown, exceptions, clusters, ActionListener.releaseAfter(originalListener, searchResponseMerger) ) { @Override void innerOnResponse(SearchResponse searchResponse) { ccsClusterInfoUpdate(searchResponse, clusters, clusterAlias, skipUnavailable); searchResponseMerger.add(searchResponse); progressListener.notifyClusterResponseMinimizeRoundtrips(clusterAlias, searchResponse); } @Override SearchResponse createFinalResponse() { return searchResponseMerger.getMergedResponse(clusters); } @Override protected void releaseResponse(SearchResponse searchResponse) { searchResponse.decRef(); } }; } /** * Creates a new Cluster object using the {@link ShardSearchFailure} info and skip_unavailable * flag to set Status. Then it swaps it in the clusters CHM at key clusterAlias */ static void ccsClusterInfoUpdate( ShardSearchFailure failure, SearchResponse.Clusters clusters, String clusterAlias, boolean skipUnavailable ) { clusters.swapCluster(clusterAlias, (k, v) -> { SearchResponse.Cluster.Status status; if (skipUnavailable) { status = SearchResponse.Cluster.Status.SKIPPED; } else { status = SearchResponse.Cluster.Status.FAILED; } return new SearchResponse.Cluster.Builder(v).setStatus(status) .setFailures(CollectionUtils.appendToCopy(v.getFailures(), failure)) .build(); }); } /** * Helper method common to multiple ccs_minimize_roundtrips=true code paths. * Used to update a specific SearchResponse.Cluster object state based upon * the SearchResponse coming from the cluster coordinator the search was performed on. * @param searchResponse SearchResponse from cluster sub-search * @param clusters Clusters that the search was executed on * @param clusterAlias Alias of the cluster to be updated */ private static void ccsClusterInfoUpdate( SearchResponse searchResponse, SearchResponse.Clusters clusters, String clusterAlias, boolean skipUnavailable ) { /* * Cluster Status logic: * 1) FAILED if total_shards > 0 && all shards failed && skip_unavailable=false * 2) SKIPPED if total_shards > 0 && all shards failed && skip_unavailable=true * 3) PARTIAL if it timed out * 4) PARTIAL if it at least one of the shards succeeded but not all * 5) SUCCESSFUL if no shards failed (and did not time out) */ clusters.swapCluster(clusterAlias, (k, v) -> { SearchResponse.Cluster.Status status; int totalShards = searchResponse.getTotalShards(); if (totalShards > 0 && searchResponse.getFailedShards() >= totalShards) { if (skipUnavailable) { status = SearchResponse.Cluster.Status.SKIPPED; } else { status = SearchResponse.Cluster.Status.FAILED; } } else if (searchResponse.isTimedOut()) { status = SearchResponse.Cluster.Status.PARTIAL; } else if (searchResponse.getFailedShards() > 0) { status = SearchResponse.Cluster.Status.PARTIAL; } else { status = SearchResponse.Cluster.Status.SUCCESSFUL; } return new SearchResponse.Cluster.Builder(v).setStatus(status) .setTotalShards(totalShards) .setSuccessfulShards(searchResponse.getSuccessfulShards()) .setSkippedShards(searchResponse.getSkippedShards()) .setFailedShards(searchResponse.getFailedShards()) .setFailures(Arrays.asList(searchResponse.getShardFailures())) .setTook(searchResponse.getTook()) .setTimedOut(searchResponse.isTimedOut()) .build(); }); } /** * Edge case --- * Typically we don't need to update a Cluster object after the SearchShards API call, since the * skipped shards will be passed into SearchProgressListener.onListShards. * However, there is an edge case where the remote SearchShards API call returns no shards at all. * So in that case, nothing for this cluster will be passed to onListShards, so we need to update * the Cluster object to SUCCESSFUL status with shard counts of 0 and a filled in 'took' value. * * @param response from SearchShards API call to remote cluster * @param clusters Clusters that the search was executed on * @param clusterAlias Alias of the cluster to be updated * @param timeProvider search time provider (for setting took value) */ private static void ccsClusterInfoUpdate( SearchShardsResponse response, SearchResponse.Clusters clusters, String clusterAlias, SearchTimeProvider timeProvider ) { if (response.getGroups().isEmpty()) { clusters.swapCluster( clusterAlias, (k, v) -> new SearchResponse.Cluster.Builder(v).setStatus(SearchResponse.Cluster.Status.SUCCESSFUL) .setTotalShards(0) .setSuccessfulShards(0) .setSkippedShards(0) .setFailedShards(0) .setFailures(Collections.emptyList()) .setTook(new TimeValue(timeProvider.buildTookInMillis())) .setTimedOut(false) .build() ); } } void executeLocalSearch( Task task, SearchTimeProvider timeProvider, SearchRequest searchRequest, ResolvedIndices resolvedIndices, ProjectState projectState, SearchResponse.Clusters clusterInfo, SearchPhaseProvider searchPhaseProvider ) { executeSearch( (SearchTask) task, timeProvider, searchRequest, resolvedIndices, Collections.emptyList(), (clusterName, nodeId) -> null, projectState, Collections.emptyMap(), clusterInfo, searchPhaseProvider ); } static BiFunction getRemoteClusterNodeLookup(Map searchShardsResp) { Map> clusterToNode = new HashMap<>(); for (Map.Entry entry : searchShardsResp.entrySet()) { String clusterAlias = entry.getKey(); for (DiscoveryNode remoteNode : entry.getValue().getNodes()) { clusterToNode.computeIfAbsent(clusterAlias, k -> new HashMap<>()).put(remoteNode.getId(), remoteNode); } } return (clusterAlias, nodeId) -> { Map clusterNodes = clusterToNode.get(clusterAlias); if (clusterNodes == null) { throw new IllegalArgumentException("unknown remote cluster: " + clusterAlias); } return clusterNodes.get(nodeId); }; } /** * Produce a list of {@link SearchShardIterator}s from the set of responses from remote clusters. * Used for ccs_minimize_roundtrips=false. */ static List getRemoteShardsIterator( Map searchShardsResponses, Map remoteIndicesByCluster, Map aliasFilterMap ) { final List remoteShardIterators = new ArrayList<>(); for (Map.Entry entry : searchShardsResponses.entrySet()) { for (SearchShardsGroup searchShardsGroup : entry.getValue().getGroups()) { // add the cluster name to the remote index names for indices disambiguation // this ends up in the hits returned with the search response ShardId shardId = searchShardsGroup.shardId(); AliasFilter aliasFilter = aliasFilterMap.get(shardId.getIndex().getUUID()); String[] aliases = aliasFilter.getAliases(); String clusterAlias = entry.getKey(); String[] finalIndices = aliases.length == 0 ? new String[] { shardId.getIndexName() } : aliases; final OriginalIndices originalIndices = remoteIndicesByCluster.get(clusterAlias); assert originalIndices != null : "original indices are null for clusterAlias: " + clusterAlias; SearchShardIterator shardIterator = new SearchShardIterator( clusterAlias, shardId, searchShardsGroup.allocatedNodes(), new OriginalIndices(finalIndices, originalIndices.indicesOptions()), null, null, searchShardsGroup.preFiltered(), searchShardsGroup.skipped() ); remoteShardIterators.add(shardIterator); } } return remoteShardIterators; } static List getRemoteShardsIteratorFromPointInTime( Map searchShardsResponses, SearchContextId searchContextId, TimeValue searchContextKeepAlive, Map remoteClusterIndices ) { final List remoteShardIterators = new ArrayList<>(); for (Map.Entry entry : searchShardsResponses.entrySet()) { for (SearchShardsGroup group : entry.getValue().getGroups()) { final ShardId shardId = group.shardId(); final SearchContextIdForNode perNode = searchContextId.shards().get(shardId); if (perNode == null) { // the shard was skipped after can match, hence it is not even part of the pit id continue; } final String clusterAlias = entry.getKey(); assert clusterAlias.equals(perNode.getClusterAlias()) : clusterAlias + " != " + perNode.getClusterAlias(); final List targetNodes = new ArrayList<>(group.allocatedNodes().size()); if (perNode.getNode() != null) { // If the shard was available when the PIT was created, it's included. // Otherwise, we add the shard iterator without a target node, allowing a partial search failure to // be thrown when a search phase attempts to access it. targetNodes.add(perNode.getNode()); if (perNode.getSearchContextId().getSearcherId() != null) { for (String node : group.allocatedNodes()) { if (node.equals(perNode.getNode()) == false) { targetNodes.add(node); } } } } assert remoteClusterIndices.get(clusterAlias) != null : "original indices are null for clusterAlias: " + clusterAlias; final OriginalIndices finalIndices = new OriginalIndices( new String[] { shardId.getIndexName() }, remoteClusterIndices.get(clusterAlias).indicesOptions() ); SearchShardIterator shardIterator = new SearchShardIterator( clusterAlias, shardId, targetNodes, finalIndices, perNode.getSearchContextId(), searchContextKeepAlive, false, false ); remoteShardIterators.add(shardIterator); } } assert checkAllRemotePITShardsWereReturnedBySearchShards(searchContextId.shards(), searchShardsResponses) : "search shards did not return remote shards that PIT included: " + searchContextId.shards(); return remoteShardIterators; } private static boolean checkAllRemotePITShardsWereReturnedBySearchShards( Map searchContextIdShards, Map searchShardsResponses ) { Map searchContextIdForNodeMap = new HashMap<>(searchContextIdShards); for (SearchShardsResponse searchShardsResponse : searchShardsResponses.values()) { for (SearchShardsGroup group : searchShardsResponse.getGroups()) { searchContextIdForNodeMap.remove(group.shardId()); } } return searchContextIdForNodeMap.values() .stream() .allMatch(searchContextIdForNode -> searchContextIdForNode.getClusterAlias() == null); } /** * If any of the indices we are searching are frozen, issue deprecation warning. */ void frozenIndexCheck(ResolvedIndices resolvedIndices) { List frozenIndices = new ArrayList<>(); Map indexMetadataMap = resolvedIndices.getConcreteLocalIndicesMetadata(); for (var entry : indexMetadataMap.entrySet()) { if (entry.getValue().getSettings().getAsBoolean("index.frozen", false)) { frozenIndices.add(entry.getKey().getName()); } } if (frozenIndices.isEmpty() == false) { DEPRECATION_LOGGER.warn( DeprecationCategory.INDICES, "search-frozen-indices", FROZEN_INDICES_DEPRECATION_MESSAGE, String.join(",", frozenIndices) ); } } /** * Execute search locally and for all given remote shards. * Used when minimize_roundtrips=false or for local search. */ private void executeSearch( SearchTask task, SearchTimeProvider timeProvider, SearchRequest searchRequest, ResolvedIndices resolvedIndices, List remoteShardIterators, BiFunction remoteConnections, ProjectState projectState, Map remoteAliasMap, SearchResponse.Clusters clusters, SearchPhaseProvider searchPhaseProvider ) { if (searchRequest.allowPartialSearchResults() == null) { // No user preference defined in search request - apply cluster service default searchRequest.allowPartialSearchResults(searchService.defaultAllowPartialSearchResults()); } // TODO: I think startTime() should become part of ActionRequest and that should be used both for index name // date math expressions and $now in scripts. This way all apis will deal with now in the same way instead // of just for the _search api final List localShardIterators; final Map aliasFilter; final String[] concreteLocalIndices; if (resolvedIndices.getSearchContextId() != null) { assert searchRequest.pointInTimeBuilder() != null; aliasFilter = resolvedIndices.getSearchContextId().aliasFilter(); concreteLocalIndices = resolvedIndices.getLocalIndices() == null ? new String[0] : resolvedIndices.getLocalIndices().indices(); localShardIterators = getLocalShardsIteratorFromPointInTime( projectState, searchRequest.indicesOptions(), searchRequest.getLocalClusterAlias(), resolvedIndices.getSearchContextId(), searchRequest.pointInTimeBuilder().getKeepAlive(), searchRequest.allowPartialSearchResults() ); } else { final Index[] indices = resolvedIndices.getConcreteLocalIndices(); concreteLocalIndices = Arrays.stream(indices).map(Index::getName).toArray(String[]::new); final Set indicesAndAliases = indexNameExpressionResolver.resolveExpressions( projectState.metadata(), searchRequest.indices() ); aliasFilter = buildIndexAliasFilters(projectState, indicesAndAliases, indices); aliasFilter.putAll(remoteAliasMap); localShardIterators = getLocalShardsIterator( projectState, searchRequest, searchRequest.getLocalClusterAlias(), indicesAndAliases, concreteLocalIndices ); // localShardIterators is empty since there are no matching indices. In such cases, // we update the local cluster's status from RUNNING to SUCCESSFUL right away. Before // we attempt to do that, we must ensure that the local cluster was specified in the user's // search request. This is done by trying to fetch the local cluster via getCluster() and // checking for a non-null return value. If the local cluster was never specified, its status // update can be skipped. if (localShardIterators.isEmpty() && clusters != SearchResponse.Clusters.EMPTY && clusters.getCluster(RemoteClusterAware.LOCAL_CLUSTER_GROUP_KEY) != null) { clusters.swapCluster( RemoteClusterAware.LOCAL_CLUSTER_GROUP_KEY, (alias, v) -> new SearchResponse.Cluster.Builder(v).setStatus(SearchResponse.Cluster.Status.SUCCESSFUL) .setTotalShards(0) .setSuccessfulShards(0) .setSkippedShards(0) .setFailedShards(0) .setFailures(Collections.emptyList()) .setTook(TimeValue.timeValueMillis(0)) .setTimedOut(false) .build() ); } } final List shardIterators = mergeShardsIterators(localShardIterators, remoteShardIterators); failIfOverShardCountLimit(clusterService, shardIterators.size()); if (searchRequest.getWaitForCheckpoints().isEmpty() == false) { if (remoteShardIterators.isEmpty() == false) { throw new IllegalArgumentException("Cannot use wait_for_checkpoints parameter with cross-cluster searches."); } else { validateAndResolveWaitForCheckpoint(projectState, indexNameExpressionResolver, searchRequest, concreteLocalIndices); } } Map concreteIndexBoosts = resolveIndexBoosts(searchRequest, projectState.cluster()); adjustSearchType(searchRequest, shardIterators.size() == 1); final DiscoveryNodes nodes = projectState.cluster().nodes(); BiFunction connectionLookup = buildConnectionLookup( searchRequest.getLocalClusterAlias(), nodes::get, remoteConnections, searchTransportService::getConnection ); final Executor asyncSearchExecutor = asyncSearchExecutor(concreteLocalIndices); final boolean preFilterSearchShards = shouldPreFilterSearchShards( projectState, searchRequest, concreteLocalIndices, localShardIterators.size() + remoteShardIterators.size(), defaultPreFilterShardSize ); searchPhaseProvider.runNewSearchPhase( task, searchRequest, asyncSearchExecutor, shardIterators, timeProvider, connectionLookup, projectState.cluster(), Collections.unmodifiableMap(aliasFilter), concreteIndexBoosts, preFilterSearchShards, threadPool, clusters ); } Executor asyncSearchExecutor(final String[] indices) { boolean seenSystem = false; boolean seenCritical = false; for (String index : indices) { final String executorName = executorSelector.executorForSearch(index); switch (executorName) { case SYSTEM_READ -> seenSystem = true; case SYSTEM_CRITICAL_READ -> seenCritical = true; default -> { return threadPool.executor(executorName); } } } final String executor; if (seenSystem == false && seenCritical) { executor = SYSTEM_CRITICAL_READ; } else if (seenSystem) { executor = SYSTEM_READ; } else { executor = ThreadPool.Names.SEARCH; } return threadPool.executor(executor); } static BiFunction buildConnectionLookup( String requestClusterAlias, Function localNodes, BiFunction remoteNodes, BiFunction nodeToConnection ) { return (clusterAlias, nodeId) -> { final DiscoveryNode discoveryNode; final boolean remoteCluster; if (clusterAlias == null || requestClusterAlias != null) { assert requestClusterAlias == null || requestClusterAlias.equals(clusterAlias); discoveryNode = localNodes.apply(nodeId); remoteCluster = false; } else { discoveryNode = remoteNodes.apply(clusterAlias, nodeId); remoteCluster = true; } if (discoveryNode == null) { throw new IllegalStateException("no node found for id: " + nodeId); } return nodeToConnection.apply(remoteCluster ? clusterAlias : null, discoveryNode); }; } static boolean shouldPreFilterSearchShards( ProjectState projectState, SearchRequest searchRequest, String[] indices, int numShards, int defaultPreFilterShardSize ) { if (searchRequest.searchType() != QUERY_THEN_FETCH) { // we can't do this for DFS it needs to fan out to all shards all the time return false; } SearchSourceBuilder source = searchRequest.source(); Integer preFilterShardSize = searchRequest.getPreFilterShardSize(); if (preFilterShardSize == null) { if (hasReadOnlyIndices(indices, projectState) || hasPrimaryFieldSort(source)) { preFilterShardSize = 1; } else { preFilterShardSize = defaultPreFilterShardSize; } } return preFilterShardSize < numShards && (SearchService.canRewriteToMatchNone(source) || hasPrimaryFieldSort(source)); } private static boolean hasReadOnlyIndices(String[] indices, ProjectState projectState) { var blocks = projectState.blocks(); if (blocks.global(projectState.projectId()).isEmpty() && blocks.indices(projectState.projectId()).isEmpty()) { // short circuit optimization because block check below is relatively expensive for many indices return false; } for (String index : indices) { ClusterBlockException writeBlock = blocks.indexBlockedException(projectState.projectId(), ClusterBlockLevel.WRITE, index); if (writeBlock != null) { return true; } } return false; } // package private for testing static List mergeShardsIterators( List localShardIterators, List remoteShardIterators ) { final List shards; if (remoteShardIterators.isEmpty()) { shards = localShardIterators; } else { shards = CollectionUtils.concatLists(remoteShardIterators, localShardIterators); } shards.sort(SearchShardIterator::compareTo); return shards; } interface SearchPhaseProvider { void runNewSearchPhase( SearchTask task, SearchRequest searchRequest, Executor executor, List shardIterators, SearchTimeProvider timeProvider, BiFunction connectionLookup, ClusterState clusterState, Map aliasFilter, Map concreteIndexBoosts, boolean preFilter, ThreadPool threadPool, SearchResponse.Clusters clusters ); } private class AsyncSearchActionProvider implements SearchPhaseProvider { private final ActionListener listener; AsyncSearchActionProvider(ActionListener listener) { this.listener = listener; } @Override public void runNewSearchPhase( SearchTask task, SearchRequest searchRequest, Executor executor, List shardIterators, SearchTimeProvider timeProvider, BiFunction connectionLookup, ClusterState clusterState, Map aliasFilter, Map concreteIndexBoosts, boolean preFilter, ThreadPool threadPool, SearchResponse.Clusters clusters ) { if (preFilter) { // only for aggs we need to contact shards even if there are no matches boolean requireAtLeastOneMatch = searchRequest.source() != null && searchRequest.source().aggregations() != null; CanMatchPreFilterSearchPhase.execute( logger, searchTransportService, connectionLookup, aliasFilter, concreteIndexBoosts, threadPool.executor(ThreadPool.Names.SEARCH_COORDINATION), searchRequest, shardIterators, timeProvider, task, requireAtLeastOneMatch, searchService.getCoordinatorRewriteContextProvider(timeProvider::absoluteStartMillis) ) .addListener( listener.delegateFailureAndWrap( (l, iters) -> runNewSearchPhase( task, searchRequest, executor, iters, timeProvider, connectionLookup, clusterState, aliasFilter, concreteIndexBoosts, false, threadPool, clusters ) ) ); return; } // for synchronous CCS minimize_roundtrips=false, use the CCSSingleCoordinatorSearchProgressListener // (AsyncSearchTask will not return SearchProgressListener.NOOP, since it uses its own progress listener // which delegates to CCSSingleCoordinatorSearchProgressListener when minimizing roundtrips) if (clusters.isCcsMinimizeRoundtrips() == false && clusters.hasRemoteClusters() && task.getProgressListener() == SearchProgressListener.NOOP) { task.setProgressListener(new CCSSingleCoordinatorSearchProgressListener()); } final SearchPhaseResults queryResultConsumer = searchPhaseController.newSearchPhaseResults( executor, circuitBreaker, task::isCancelled, task.getProgressListener(), searchRequest, shardIterators.size(), exc -> searchTransportService.cancelSearchTask(task, "failed to merge result [" + exc.getMessage() + "]") ); boolean success = false; try { final AbstractSearchAsyncAction searchPhase; if (searchRequest.searchType() == DFS_QUERY_THEN_FETCH) { searchPhase = new SearchDfsQueryThenFetchAsyncAction( logger, namedWriteableRegistry, searchTransportService, connectionLookup, aliasFilter, concreteIndexBoosts, executor, queryResultConsumer, searchRequest, listener, shardIterators, timeProvider, clusterState, task, clusters, client ); } else { assert searchRequest.searchType() == QUERY_THEN_FETCH : searchRequest.searchType(); searchPhase = new SearchQueryThenFetchAsyncAction( logger, namedWriteableRegistry, searchTransportService, connectionLookup, aliasFilter, concreteIndexBoosts, executor, queryResultConsumer, searchRequest, listener, shardIterators, timeProvider, clusterState, task, clusters, client, searchService.batchQueryPhase() ); } success = true; searchPhase.start(); } finally { if (success == false) { queryResultConsumer.close(); } } } } private static void validateAndResolveWaitForCheckpoint( ProjectState projectState, IndexNameExpressionResolver resolver, SearchRequest searchRequest, String[] concreteLocalIndices ) { HashSet searchedIndices = new HashSet<>(Arrays.asList(concreteLocalIndices)); Map newWaitForCheckpoints = Maps.newMapWithExpectedSize(searchRequest.getWaitForCheckpoints().size()); for (Map.Entry waitForCheckpointIndex : searchRequest.getWaitForCheckpoints().entrySet()) { long[] checkpoints = waitForCheckpointIndex.getValue(); int checkpointsProvided = checkpoints.length; String target = waitForCheckpointIndex.getKey(); Index resolved; try { resolved = resolver.concreteSingleIndex(projectState.cluster(), new IndicesRequest() { @Override public String[] indices() { return new String[] { target }; } @Override public IndicesOptions indicesOptions() { return IndicesOptions.strictSingleIndexNoExpandForbidClosed(); } }); } catch (Exception e) { throw new IllegalArgumentException( "Failed to resolve wait_for_checkpoints target [" + target + "]. Configured target " + "must resolve to a single open index.", e ); } String index = resolved.getName(); IndexMetadata indexMetadata = projectState.metadata().index(index); if (searchedIndices.contains(index) == false) { throw new IllegalArgumentException( "Target configured with wait_for_checkpoints must be a concrete index resolved in " + "this search. Target [" + target + "] is not a concrete index resolved in this search." ); } else if (indexMetadata == null) { throw new IllegalArgumentException("Cannot find index configured for wait_for_checkpoints parameter [" + index + "]."); } else if (indexMetadata.getNumberOfShards() != checkpointsProvided) { throw new IllegalArgumentException( "Target configured with wait_for_checkpoints must search the same number of shards as " + "checkpoints provided. [" + checkpointsProvided + "] checkpoints provided. Target [" + target + "] which resolved to " + "index [" + index + "] has " + "[" + indexMetadata.getNumberOfShards() + "] shards." ); } newWaitForCheckpoints.put(index, checkpoints); } searchRequest.setWaitForCheckpoints(Collections.unmodifiableMap(newWaitForCheckpoints)); } private static void failIfOverShardCountLimit(ClusterService clusterService, int shardCount) { final long shardCountLimit = clusterService.getClusterSettings().get(SHARD_COUNT_LIMIT_SETTING); if (shardCount > shardCountLimit) { throw new IllegalArgumentException( "Trying to query " + shardCount + " shards, which is over the limit of " + shardCountLimit + ". This limit exists because querying many shards at the same time can make the " + "job of the coordinating node very CPU and/or memory intensive. It is usually a better idea to " + "have a smaller number of larger shards. Update [" + SHARD_COUNT_LIMIT_SETTING.getKey() + "] to a greater value if you really want to query that many shards at the same time." ); } } /** * {@link ActionListener} suitable for collecting cross-cluster responses. * @param Response type we're getting as intermediate per-cluster results. * @param Response type that the final listener expects. */ abstract static class CCSActionListener implements ActionListener { protected final String clusterAlias; protected final boolean skipUnavailable; private final CountDown countDown; private final AtomicReference exceptions; protected final SearchResponse.Clusters clusters; private final ActionListener originalListener; /** * Used by both minimize_roundtrips true and false */ CCSActionListener( String clusterAlias, boolean skipUnavailable, CountDown countDown, AtomicReference exceptions, SearchResponse.Clusters clusters, ActionListener originalListener ) { this.clusterAlias = clusterAlias; this.skipUnavailable = skipUnavailable; this.countDown = countDown; this.exceptions = exceptions; this.clusters = clusters; this.originalListener = originalListener; } @Override public final void onResponse(Response response) { innerOnResponse(response); maybeFinish(); } /** * Specific listener type will implement this method to process its specific partial response. */ abstract void innerOnResponse(Response response); @Override public final void onFailure(Exception e) { ShardSearchFailure f = new ShardSearchFailure(e); logCCSError(f, clusterAlias, skipUnavailable); SearchResponse.Cluster cluster = clusters.getCluster(clusterAlias); if (skipUnavailable && ExceptionsHelper.isTaskCancelledException(e) == false) { if (cluster != null) { ccsClusterInfoUpdate(f, clusters, clusterAlias, true); } } else { if (cluster != null) { ccsClusterInfoUpdate(f, clusters, clusterAlias, false); } Exception exception = e; if (RemoteClusterAware.LOCAL_CLUSTER_GROUP_KEY.equals(clusterAlias) == false && ExceptionsHelper.isTaskCancelledException(e) == false) { exception = wrapRemoteClusterFailure(clusterAlias, e); } if (exceptions.compareAndSet(null, exception) == false) { exceptions.accumulateAndGet(exception, (previous, current) -> { current.addSuppressed(previous); return current; }); } } maybeFinish(); } private void maybeFinish() { if (countDown.countDown()) { Exception exception = exceptions.get(); if (exception == null) { FinalResponse response; try { response = createFinalResponse(); } catch (Exception e) { originalListener.onFailure(e); return; } try { originalListener.onResponse(response); } finally { releaseResponse(response); } } else { originalListener.onFailure(exceptions.get()); } } } protected void releaseResponse(FinalResponse response) {} abstract FinalResponse createFinalResponse(); } /** * In order to gather data on what types of CCS errors happen in the field, we will log * them using the ShardSearchFailure XContent (JSON), which supplies information about underlying * causes of shard failures. * @param f ShardSearchFailure to log * @param clusterAlias cluster on which the failure occurred * @param skipUnavailable the skip_unavailable setting of the cluster with the search error */ private static void logCCSError(ShardSearchFailure f, String clusterAlias, boolean skipUnavailable) { String errorInfo; try { errorInfo = Strings.toString(f.toXContent(XContentFactory.jsonBuilder(), ToXContent.EMPTY_PARAMS)); } catch (IOException ex) { // use the toString as a fallback if for some reason the XContent conversion to JSON fails errorInfo = f.toString(); } logger.debug( "CCS remote cluster failure. Cluster [{}]. skip_unavailable: [{}]. Error: {}", clusterAlias, skipUnavailable, errorInfo ); } private static RemoteTransportException wrapRemoteClusterFailure(String clusterAlias, Exception e) { return new RemoteTransportException("error while communicating with remote cluster [" + clusterAlias + "]", e); } static List getLocalShardsIteratorFromPointInTime( ProjectState projectState, IndicesOptions indicesOptions, String localClusterAlias, SearchContextId searchContext, TimeValue keepAlive, boolean allowPartialSearchResults ) { final List iterators = new ArrayList<>(searchContext.shards().size()); for (Map.Entry entry : searchContext.shards().entrySet()) { final SearchContextIdForNode perNode = entry.getValue(); if (Strings.isEmpty(perNode.getClusterAlias())) { final ShardId shardId = entry.getKey(); final List targetNodes = new ArrayList<>(2); if (perNode.getNode() != null) { // If the shard was available when the PIT was created, it's included. // Otherwise, we add the shard iterator without a target node, allowing a partial search failure to // be thrown when a search phase attempts to access it. try { final ShardIterator shards = OperationRouting.getShards(projectState.routingTable(), shardId); // Prefer executing shard requests on nodes that are part of PIT first. if (projectState.cluster().nodes().nodeExists(perNode.getNode())) { targetNodes.add(perNode.getNode()); } if (perNode.getSearchContextId().getSearcherId() != null) { for (ShardRouting shard : shards) { if (shard.currentNodeId().equals(perNode.getNode()) == false) { targetNodes.add(shard.currentNodeId()); } } } } catch (IndexNotFoundException | ShardNotFoundException e) { // We can hit these exceptions if the index was deleted after creating PIT or the cluster state on // this coordinating node is outdated. It's fine to ignore these extra "retry-able" target shards // when allowPartialSearchResults is false if (allowPartialSearchResults == false) { throw e; } } } OriginalIndices finalIndices = new OriginalIndices(new String[] { shardId.getIndexName() }, indicesOptions); iterators.add( new SearchShardIterator( localClusterAlias, shardId, targetNodes, finalIndices, perNode.getSearchContextId(), keepAlive, false, false ) ); } } return iterators; } /** * Create a list of {@link SearchShardIterator}s for the local indices we are searching. * This resolves aliases and index expressions. */ List getLocalShardsIterator( ProjectState projectState, SearchRequest searchRequest, String clusterAlias, Set indicesAndAliases, String[] concreteIndices ) { var routingMap = indexNameExpressionResolver.resolveSearchRouting( projectState.metadata(), searchRequest.routing(), searchRequest.indices() ); List shardRoutings = clusterService.operationRouting() .searchShards( projectState, concreteIndices, routingMap, searchRequest.preference(), responseCollectorService, searchTransportService.getPendingSearchRequests() ); final Map originalIndices = buildPerIndexOriginalIndices( projectState, indicesAndAliases, concreteIndices, searchRequest.indicesOptions() ); SearchShardIterator[] list = new SearchShardIterator[shardRoutings.size()]; int i = 0; for (ShardIterator shardRouting : shardRoutings) { final ShardId shardId = shardRouting.shardId(); OriginalIndices finalIndices = originalIndices.get(shardId.getIndex().getName()); assert finalIndices != null; list[i++] = new SearchShardIterator(clusterAlias, shardId, shardRouting.getShardRoutings(), finalIndices); } // the returned list must support in-place sorting, so this is the most memory efficient we can do here return Arrays.asList(list); } private interface TelemetryListener { void setRemotes(int count); void setFeature(String feature); void setClient(Task task); } private class SearchResponseActionListener extends DelegatingActionListener implements TelemetryListener { private final CCSUsage.Builder usageBuilder; SearchResponseActionListener(ActionListener listener) { super(listener); if (listener instanceof SearchResponseActionListener srListener) { usageBuilder = srListener.usageBuilder; } else { usageBuilder = new CCSUsage.Builder(); } } /** * Should we collect telemetry for this search? */ private boolean collectTelemetry() { return collectTelemetry && usageBuilder.getRemotesCount() > 0; } public void setRemotes(int count) { usageBuilder.setRemotesCount(count); } @Override public void setFeature(String feature) { usageBuilder.setFeature(feature); } @Override public void setClient(Task task) { usageBuilder.setClientFromTask(task); } @Override public void onResponse(SearchResponse searchResponse) { try { searchResponseMetrics.recordTookTime(searchResponse.getTookInMillis()); SearchResponseMetrics.ResponseCountTotalStatus responseCountTotalStatus = SearchResponseMetrics.ResponseCountTotalStatus.SUCCESS; if (searchResponse.getShardFailures() != null && searchResponse.getShardFailures().length > 0) { // Deduplicate failures by exception message and index ShardOperationFailedException[] groupedFailures = ExceptionsHelper.groupBy(searchResponse.getShardFailures()); for (ShardOperationFailedException f : groupedFailures) { boolean causeHas500Status = false; if (f.getCause() != null) { causeHas500Status = ExceptionsHelper.status(f.getCause()).getStatus() >= 500; } if ((f.status().getStatus() >= 500 || causeHas500Status) && ExceptionsHelper.isNodeOrShardUnavailableTypeException(f.getCause()) == false) { logger.warn("TransportSearchAction shard failure (partial results response)", f); responseCountTotalStatus = SearchResponseMetrics.ResponseCountTotalStatus.PARTIAL_FAILURE; } } } searchResponseMetrics.incrementResponseCount(responseCountTotalStatus); if (collectTelemetry()) { extractCCSTelemetry(searchResponse); recordTelemetry(); } } catch (Exception e) { onFailure(e); return; } // This is last because we want to collect telemetry before returning the response. delegate.onResponse(searchResponse); } @Override public void onFailure(Exception e) { searchResponseMetrics.incrementResponseCount(SearchResponseMetrics.ResponseCountTotalStatus.FAILURE); if (collectTelemetry()) { usageBuilder.setFailure(e); recordTelemetry(); } super.onFailure(e); } private void recordTelemetry() { usageService.getCcsUsageHolder().updateUsage(usageBuilder.build()); } /** * Extract telemetry data from the search response. * @param searchResponse The final response from the search. */ private void extractCCSTelemetry(SearchResponse searchResponse) { usageBuilder.took(searchResponse.getTookInMillis()); for (String clusterAlias : searchResponse.getClusters().getClusterAliases()) { SearchResponse.Cluster cluster = searchResponse.getClusters().getCluster(clusterAlias); if (cluster.getStatus() == SearchResponse.Cluster.Status.SKIPPED) { usageBuilder.skippedRemote(clusterAlias); } else { usageBuilder.perClusterUsage(clusterAlias, cluster.getTook()); } } } } }