Skip to content

Commit

Permalink
percolate query implementation in doc-level alerting
Browse files Browse the repository at this point in the history
Signed-off-by: Subhobrata Dey <sbcd90@gmail.com>
  • Loading branch information
sbcd90 committed Apr 14, 2022
1 parent 5f4d9fd commit 6328f06
Show file tree
Hide file tree
Showing 25 changed files with 1,787 additions and 30 deletions.
1 change: 1 addition & 0 deletions alerting/build.gradle
Original file line number Diff line number Diff line change
Expand Up @@ -56,6 +56,7 @@ configurations.testImplementation {

dependencies {
compileOnly "org.opensearch.plugin:opensearch-scripting-painless-spi:${versions.opensearch}"
api "org.opensearch.plugin:percolator-client:${opensearch_version}"

// OpenSearch Nanny state
implementation "org.jetbrains.kotlin:kotlin-stdlib:${kotlin_version}"
Expand Down

Large diffs are not rendered by default.

Large diffs are not rendered by default.

Original file line number Diff line number Diff line change
@@ -0,0 +1,79 @@
/*
* SPDX-License-Identifier: Apache-2.0
*
* The OpenSearch Contributors require contributions made to
* this file be licensed under the Apache-2.0 license or a
* compatible open source license.
*/

/*
* Licensed to Elasticsearch under one or more contributor
* license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright
* ownership. Elasticsearch licenses this file to you under
* the Apache License, Version 2.0 (the "License"); you may
* not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing,
* software distributed under the License is distributed on an
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
* KIND, either express or implied. See the License for the
* specific language governing permissions and limitations
* under the License.
*/

/*
* Modifications Copyright OpenSearch Contributors. See
* GitHub history for details.
*/

package org.opensearch.percolator;

import org.opensearch.common.settings.Setting;
import org.opensearch.index.mapper.Mapper;
import org.opensearch.plugins.ExtensiblePlugin;
import org.opensearch.plugins.MapperPlugin;
import org.opensearch.plugins.Plugin;
import org.opensearch.plugins.SearchPlugin;
import org.opensearch.search.fetch.FetchSubPhase;

import java.util.Arrays;
import java.util.List;
import java.util.Map;

import static java.util.Collections.singletonList;
import static java.util.Collections.singletonMap;

/**
* This is a stop-gap solution & will be removed in future.
*/
// TODO for cleanup
public class PercolatorPluginExt extends Plugin implements MapperPlugin, SearchPlugin, ExtensiblePlugin {
@Override
public List<QuerySpec<?>> getQueries() {
return singletonList(new QuerySpec<>(PercolateQueryBuilderExt.NAME, PercolateQueryBuilderExt::new, PercolateQueryBuilderExt::fromXContent));
}

@Override
public List<FetchSubPhase> getFetchSubPhases(FetchPhaseConstructionContext context) {
return Arrays.asList(new PercolatorMatchedSlotSubFetchPhase(), new PercolatorHighlightSubFetchPhase(context.getHighlighters()));
}

@Override
public List<Setting<?>> getSettings() {
return Arrays.asList(PercolatorFieldMapperExt.INDEX_MAP_UNMAPPED_FIELDS_AS_TEXT_SETTING);
}

@Override
public Map<String, Mapper.TypeParser> getMappers() {
return singletonMap(PercolatorFieldMapperExt.CONTENT_TYPE, new PercolatorFieldMapperExt.TypeParser());
}

@Override
public void loadExtensions(ExtensionLoader loader) {
ExtensiblePlugin.super.loadExtensions(loader);
}
}
Original file line number Diff line number Diff line change
Expand Up @@ -28,6 +28,7 @@ import org.opensearch.alerting.action.SearchEmailGroupAction
import org.opensearch.alerting.action.SearchMonitorAction
import org.opensearch.alerting.aggregation.bucketselectorext.BucketSelectorExtAggregationBuilder
import org.opensearch.alerting.alerts.AlertIndices
import org.opensearch.alerting.core.DocLevelMonitorQueries
import org.opensearch.alerting.core.JobSweeper
import org.opensearch.alerting.core.ScheduledJobIndices
import org.opensearch.alerting.core.action.node.ScheduledJobsStatsAction
Expand Down Expand Up @@ -106,8 +107,8 @@ import org.opensearch.index.IndexModule
import org.opensearch.painless.spi.PainlessExtension
import org.opensearch.painless.spi.Whitelist
import org.opensearch.painless.spi.WhitelistLoader
import org.opensearch.percolator.PercolatorPluginExt
import org.opensearch.plugins.ActionPlugin
import org.opensearch.plugins.Plugin
import org.opensearch.plugins.ReloadablePlugin
import org.opensearch.plugins.ScriptPlugin
import org.opensearch.plugins.SearchPlugin
Expand All @@ -126,7 +127,7 @@ import java.util.function.Supplier
* It also adds [Monitor.XCONTENT_REGISTRY], [SearchInput.XCONTENT_REGISTRY], [QueryLevelTrigger.XCONTENT_REGISTRY],
* [BucketLevelTrigger.XCONTENT_REGISTRY], [ClusterMetricsInput.XCONTENT_REGISTRY] to the [NamedXContentRegistry] so that we are able to deserialize the custom named objects.
*/
internal class AlertingPlugin : PainlessExtension, ActionPlugin, ScriptPlugin, ReloadablePlugin, SearchPlugin, Plugin() {
internal class AlertingPlugin : PainlessExtension, ActionPlugin, ScriptPlugin, ReloadablePlugin, SearchPlugin, PercolatorPluginExt() {

override fun getContextWhitelists(): Map<ScriptContext<*>, List<Whitelist>> {
val whitelist = WhitelistLoader.loadFromResourceFiles(javaClass, "org.opensearch.alerting.txt")
Expand All @@ -152,6 +153,7 @@ internal class AlertingPlugin : PainlessExtension, ActionPlugin, ScriptPlugin, R
lateinit var scheduler: JobScheduler
lateinit var sweeper: JobSweeper
lateinit var scheduledJobIndices: ScheduledJobIndices
lateinit var docLevelMonitorQueries: DocLevelMonitorQueries
lateinit var threadPool: ThreadPool
lateinit var alertIndices: AlertIndices
lateinit var clusterService: ClusterService
Expand Down Expand Up @@ -257,11 +259,12 @@ internal class AlertingPlugin : PainlessExtension, ActionPlugin, ScriptPlugin, R
.registerConsumers()
.registerDestinationSettings()
scheduledJobIndices = ScheduledJobIndices(client.admin(), clusterService)
docLevelMonitorQueries = DocLevelMonitorQueries(client.admin(), clusterService)
scheduler = JobScheduler(threadPool, runner)
sweeper = JobSweeper(environment.settings(), client, clusterService, threadPool, xContentRegistry, scheduler, ALERTING_JOB_TYPES)
this.threadPool = threadPool
this.clusterService = clusterService
return listOf(sweeper, scheduler, runner, scheduledJobIndices)
return listOf(sweeper, scheduler, runner, scheduledJobIndices, docLevelMonitorQueries)
}

override fun getSettings(): List<Setting<*>> {
Expand Down
Original file line number Diff line number Diff line change
@@ -1,3 +1,8 @@
/*
* Copyright OpenSearch Contributors
* SPDX-License-Identifier: Apache-2.0
*/

package org.opensearch.alerting

import kotlinx.coroutines.Dispatchers
Expand Down
Original file line number Diff line number Diff line change
@@ -1,14 +1,21 @@
/*
* Copyright OpenSearch Contributors
* SPDX-License-Identifier: Apache-2.0
*/

package org.opensearch.alerting

import kotlinx.coroutines.Dispatchers
import kotlinx.coroutines.withContext
import org.apache.logging.log4j.LogManager
import org.opensearch.action.index.IndexRequest
import org.opensearch.action.search.SearchAction
import org.opensearch.action.search.SearchRequest
import org.opensearch.action.search.SearchResponse
import org.opensearch.action.support.WriteRequest
import org.opensearch.alerting.core.model.DocLevelMonitorInput
import org.opensearch.alerting.core.model.DocLevelQuery
import org.opensearch.alerting.core.model.ScheduledJob
import org.opensearch.alerting.elasticapi.string
import org.opensearch.alerting.model.ActionRunResult
import org.opensearch.alerting.model.Alert
Expand All @@ -30,11 +37,14 @@ import org.opensearch.client.Client
import org.opensearch.cluster.routing.ShardRouting
import org.opensearch.cluster.service.ClusterService
import org.opensearch.common.Strings
import org.opensearch.common.bytes.BytesReference
import org.opensearch.common.xcontent.ToXContent
import org.opensearch.common.xcontent.XContentBuilder
import org.opensearch.common.xcontent.XContentFactory
import org.opensearch.common.xcontent.XContentType
import org.opensearch.index.query.BoolQueryBuilder
import org.opensearch.index.query.QueryBuilders
import org.opensearch.percolator.PercolateQueryBuilderExt
import org.opensearch.rest.RestStatus
import org.opensearch.search.SearchHits
import org.opensearch.search.builder.SearchSourceBuilder
Expand Down Expand Up @@ -93,22 +103,46 @@ object DocumentReturningMonitorRunner : MonitorRunner {
}
}

val queryToDocIds = mutableMapOf<DocLevelQuery, Set<String>>()
val queryToDocIds = mutableMapOf<DocLevelQuery, MutableSet<String>>()
val docsToQueries = mutableMapOf<String, MutableList<String>>()
val docExecutionContext = DocumentExecutionContext(queries, lastRunContext, updatedLastRunContext)
val idQueryMap = mutableMapOf<String, DocLevelQuery>()
queries.forEach { query ->
val matchingDocIds = runForEachQuery(monitorCtx, docExecutionContext, query, index)
queryToDocIds[query] = matchingDocIds
matchingDocIds.forEach {
docsToQueries.putIfAbsent(it, mutableListOf())
docsToQueries[it]?.add(query.id)

val matchingDocs = getMatchingDocs(monitor, monitorCtx, docExecutionContext, index, dryrun)

if (matchingDocs.isNotEmpty()) {
val matchedQueriesForDocs = getMatchedQueries(monitorCtx, matchingDocs.map { it.second }, monitor)

matchedQueriesForDocs.forEach { hit ->
val (id, query) = Pair(
hit.id.replace("_${monitor.id}", ""),
((hit.sourceAsMap["query"] as HashMap<*, *>)["query_string"] as HashMap<*, *>)["query"]
)
val docLevelQuery = DocLevelQuery(id, id, query.toString())

val docIndices = hit.field("_percolator_document_slot").values.map { it.toString().toInt() }
docIndices.forEach { idx ->
if (queryToDocIds.containsKey(docLevelQuery)) {
queryToDocIds[docLevelQuery]?.add(matchingDocs[idx].first)
} else {
queryToDocIds[docLevelQuery] = mutableSetOf(matchingDocs[idx].first)
}

if (docsToQueries.containsKey(matchingDocs[idx].first)) {
docsToQueries[matchingDocs[idx].first]?.add(id)
} else {
docsToQueries[matchingDocs[idx].first] = mutableListOf(id)
}
}
}
idQueryMap[query.id] = query
}

val queryInputResults = queryToDocIds.mapKeys { it.key.id }
monitorResult = monitorResult.copy(inputResults = InputRunResults(listOf(queryInputResults)))
val queryIds = queries.map { it.id }
val queryIds = queries.map {
idQueryMap[it.id] = it
it.id
}

val triggerResults = mutableMapOf<String, DocumentLevelTriggerRunResult>()
monitor.triggers.forEach {
Expand Down Expand Up @@ -339,40 +373,49 @@ object DocumentReturningMonitorRunner : MonitorRunner {
return allShards.filter { it.primary() }.size
}

private fun runForEachQuery(
private fun getMatchingDocs(
monitor: Monitor,
monitorCtx: MonitorRunnerExecutionContext,
docExecutionCtx: DocumentExecutionContext,
query: DocLevelQuery,
index: String
): Set<String> {
index: String,
dryrun: Boolean
): List<Pair<String, BytesReference>> {
val count: Int = docExecutionCtx.lastRunContext["shards_count"] as Int
val matchingDocs = mutableSetOf<String>()
val matchingDocs = mutableListOf<Pair<String, BytesReference>>()
for (i: Int in 0 until count) {
val shard = i.toString()
try {
logger.info("Monitor execution for shard: $shard")

val prevSeqNo = docExecutionCtx.lastRunContext[shard].toString().toLongOrNull()
val maxSeqNo = docExecutionCtx.updatedLastRunContext[shard].toString().toLong()
logger.info("Shard_$shard has MaxSeqNo: $maxSeqNo and PrevSeqNo: $prevSeqNo")
val maxSeqNo: Long = docExecutionCtx.updatedLastRunContext[shard].toString().toLong()
logger.info("MaxSeqNo of shard_$shard is $maxSeqNo")

// If dryrun, set the previous sequence number as 1 less than the max sequence number or 0
val prevSeqNo = if (dryrun || monitor.id == Monitor.NO_ID)
max(-1, maxSeqNo - 1)
else docExecutionCtx.lastRunContext[shard].toString().toLongOrNull()

if (dryrun) {
logger.info("it is a dryrun")
}

logger.info("prevSeq: $prevSeqNo, maxSeq: $maxSeqNo")

val hits: SearchHits = searchShard(
monitorCtx,
index,
shard,
prevSeqNo,
maxSeqNo,
query.query
null
)
logger.info("Search hits for shard_$shard is: ${hits.hits.size}")

if (hits.hits.isNotEmpty()) {
logger.info("found matches")
matchingDocs.addAll(getAllDocIds(hits))
matchingDocs.addAll(getAllDocs(hits, monitor.id))
}
} catch (e: Exception) {
logger.info("Failed to run for shard $shard. Error: ${e.message}")
logger.debug("Failed to run for shard $shard", e)
}
}
return matchingDocs
Expand All @@ -384,14 +427,17 @@ object DocumentReturningMonitorRunner : MonitorRunner {
shard: String,
prevSeqNo: Long?,
maxSeqNo: Long,
query: String
query: String?
): SearchHits {
if (prevSeqNo?.equals(maxSeqNo) == true && maxSeqNo != 0L) {
return SearchHits.empty()
}
val boolQueryBuilder = BoolQueryBuilder()
boolQueryBuilder.filter(QueryBuilders.rangeQuery("_seq_no").gt(prevSeqNo).lte(maxSeqNo))
boolQueryBuilder.must(QueryBuilders.queryStringQuery(query))

if (query != null) {
boolQueryBuilder.must(QueryBuilders.queryStringQuery(query))
}

val request: SearchRequest = SearchRequest()
.indices(index)
Expand All @@ -410,7 +456,45 @@ object DocumentReturningMonitorRunner : MonitorRunner {
return response.hits
}

private fun getAllDocIds(hits: SearchHits): List<String> {
return hits.map { hit -> hit.id }
private fun getMatchedQueries(
monitorCtx: MonitorRunnerExecutionContext,
docs: List<BytesReference>,
monitor: Monitor
): SearchHits {
val boolQueryBuilder = BoolQueryBuilder()

val percolateQueryBuilder = PercolateQueryBuilderExt("query", docs, XContentType.JSON)
if (monitor.id.isNotEmpty()) {
boolQueryBuilder.filter(QueryBuilders.matchQuery("monitor_id", monitor.id))
}
boolQueryBuilder.filter(percolateQueryBuilder)

val searchRequest = SearchRequest(ScheduledJob.DOC_LEVEL_QUERIES_INDEX)
val searchSourceBuilder = SearchSourceBuilder()
searchSourceBuilder.query(boolQueryBuilder)
searchRequest.source(searchSourceBuilder)

val response: SearchResponse = monitorCtx.client!!.execute(SearchAction.INSTANCE, searchRequest).actionGet()

if (response.status() !== RestStatus.OK) {
throw IOException("Failed to search percolate index: ${ScheduledJob.DOC_LEVEL_QUERIES_INDEX}")
}
return response.hits
}

private fun getAllDocs(hits: SearchHits, monitorId: String): List<Pair<String, BytesReference>> {
return hits.map { hit ->
val sourceMap = hit.sourceAsMap

var xContentBuilder = XContentFactory.jsonBuilder().startObject()
sourceMap.forEach { (k, v) ->
xContentBuilder = xContentBuilder.field("${k}_$monitorId", v)
}
xContentBuilder = xContentBuilder.endObject()

val sourceRef = BytesReference.bytes(xContentBuilder)

Pair(hit.id, sourceRef)
}
}
}
Original file line number Diff line number Diff line change
@@ -1,3 +1,8 @@
/*
* Copyright OpenSearch Contributors
* SPDX-License-Identifier: Apache-2.0
*/

package org.opensearch.alerting

import org.opensearch.action.bulk.BackoffPolicy
Expand Down
Original file line number Diff line number Diff line change
@@ -1,3 +1,8 @@
/*
* Copyright OpenSearch Contributors
* SPDX-License-Identifier: Apache-2.0
*/

package org.opensearch.alerting

import kotlinx.coroutines.Dispatchers
Expand Down
Original file line number Diff line number Diff line change
@@ -1,3 +1,8 @@
/*
* Copyright OpenSearch Contributors
* SPDX-License-Identifier: Apache-2.0
*/

package org.opensearch.alerting.model

import org.opensearch.alerting.core.model.DocLevelQuery
Expand Down
Loading

0 comments on commit 6328f06

Please sign in to comment.