[SPARK-23128][SQL] A new approach to do adaptive execution in Spark SQL #20303
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,126 @@ | ||
| /* | ||
| * Licensed to the Apache Software Foundation (ASF) under one or more | ||
| * contributor license agreements. See the NOTICE file distributed with | ||
| * this work for additional information regarding copyright ownership. | ||
| * The ASF 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. | ||
| */ | ||
|
|
||
| package org.apache.spark.sql.execution.adaptive | ||
|
|
||
| import java.util.concurrent.CountDownLatch | ||
|
|
||
| import org.apache.spark.rdd.RDD | ||
| import org.apache.spark.sql.SparkSession | ||
| import org.apache.spark.sql.catalyst.InternalRow | ||
| import org.apache.spark.sql.catalyst.expressions.Attribute | ||
| import org.apache.spark.sql.execution.{LeafExecNode, SparkPlan, SparkPlanInfo, SQLExecution} | ||
| import org.apache.spark.sql.execution.ui.SparkListenerSQLAdaptiveExecutionUpdate | ||
|
|
||
| /** | ||
| * A root node to trigger query stages and execute the query plan adaptively. It incrementally | ||
| * updates the query plan when a query stage is materialized and provides accurate runtime | ||
| * statistics. | ||
| */ | ||
| case class AdaptiveSparkPlan(initialPlan: ResultQueryStage, session: SparkSession) | ||
| extends LeafExecNode{ | ||
|
|
||
| override def output: Seq[Attribute] = initialPlan.output | ||
|
|
||
| @volatile private var currentQueryStage: QueryStage = initialPlan | ||
| @volatile private var error: Throwable = null | ||
| private val readyLock = new CountDownLatch(1) | ||
|
||
|
|
||
| private def replaceStage(oldStage: QueryStage, newStage: QueryStage): QueryStage = { | ||
| if (oldStage.id == newStage.id) { | ||
| newStage | ||
| } else { | ||
| val newPlanForOldStage = oldStage.plan.transform { | ||
|
||
| case q: QueryStage => replaceStage(q, newStage) | ||
| } | ||
| oldStage.withNewPlan(newPlanForOldStage) | ||
| } | ||
| } | ||
|
|
||
| private def createCallback(executionId: Option[Long]): QueryStageTriggerCallback = { | ||
| new QueryStageTriggerCallback { | ||
| override def onStageUpdated(stage: QueryStage): Unit = { | ||
| updateCurrentQueryStage(stage, executionId) | ||
| if (stage.isInstanceOf[ResultQueryStage]) readyLock.countDown() | ||
| } | ||
|
|
||
| override def onStagePlanningFailed(stage: QueryStage, e: Throwable): Unit = { | ||
| error = new RuntimeException( | ||
|
||
| s""" | ||
| |Fail to plan stage ${stage.id}: | ||
| |${stage.plan.treeString} | ||
| """.stripMargin, e) | ||
| readyLock.countDown() | ||
| } | ||
|
|
||
| override def onStageMaterializingFailed(stage: QueryStage, e: Throwable): Unit = { | ||
| error = new RuntimeException( | ||
| s""" | ||
| |Fail to materialize stage ${stage.id}: | ||
| |${stage.plan.treeString} | ||
| """.stripMargin, e) | ||
| readyLock.countDown() | ||
| } | ||
|
|
||
| override def onError(e: Throwable): Unit = { | ||
| error = e | ||
| readyLock.countDown() | ||
| } | ||
| } | ||
| } | ||
|
|
||
| private def updateCurrentQueryStage(newStage: QueryStage, executionId: Option[Long]): Unit = { | ||
| currentQueryStage = replaceStage(currentQueryStage, newStage) | ||
| executionId.foreach { id => | ||
| session.sparkContext.listenerBus.post(SparkListenerSQLAdaptiveExecutionUpdate( | ||
| id, | ||
| SQLExecution.getQueryExecution(id).toString, | ||
| SparkPlanInfo.fromSparkPlan(currentQueryStage))) | ||
| } | ||
| } | ||
|
|
||
| def resultStage: ResultQueryStage = { | ||
| if (readyLock.getCount > 0) { | ||
|
There was a problem hiding this comment. Is finalPlan be called/used concurrently in multiple threads? If yes, the use of readyLock.getCount may cause problem. For example, two thread will both find getCount > 0 and each thread will create a QueryStageCreator before readyLock.countDown is called.
Contributor
Author
There was a problem hiding this comment. It should only be called in a single thread. @cloud-fan may confirm it. |
||
| val sc = session.sparkContext | ||
| val executionId = Option(sc.getLocalProperty(SQLExecution.EXECUTION_ID_KEY)).map(_.toLong) | ||
| val trigger = new QueryStageTrigger(session, createCallback(executionId)) | ||
| trigger.start() | ||
| trigger.trigger(initialPlan) | ||
| readyLock.await() | ||
| trigger.stop() | ||
| } | ||
|
|
||
| if (error != null) throw error | ||
| currentQueryStage.asInstanceOf[ResultQueryStage] | ||
| } | ||
|
|
||
| override def executeCollect(): Array[InternalRow] = resultStage.executeCollect() | ||
| override def executeTake(n: Int): Array[InternalRow] = resultStage.executeTake(n) | ||
| override def executeToIterator(): Iterator[InternalRow] = resultStage.executeToIterator() | ||
| override def doExecute(): RDD[InternalRow] = resultStage.execute() | ||
| override def generateTreeString( | ||
| depth: Int, | ||
| lastChildren: Seq[Boolean], | ||
| append: String => Unit, | ||
| verbose: Boolean, | ||
| prefix: String = "", | ||
| addSuffix: Boolean = false, | ||
| maxFields: Int): Unit = { | ||
| currentQueryStage.generateTreeString( | ||
| depth, lastChildren, append, verbose, "", false, maxFields) | ||
| } | ||
| } | ||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,57 @@ | ||
| /* | ||
| * Licensed to the Apache Software Foundation (ASF) under one or more | ||
| * contributor license agreements. See the NOTICE file distributed with | ||
| * this work for additional information regarding copyright ownership. | ||
| * The ASF 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. | ||
| */ | ||
|
|
||
| package org.apache.spark.sql.execution.adaptive | ||
|
|
||
| import org.apache.spark.sql.SparkSession | ||
| import org.apache.spark.sql.catalyst.rules.Rule | ||
| import org.apache.spark.sql.execution.SparkPlan | ||
| import org.apache.spark.sql.execution.exchange.{BroadcastExchangeExec, Exchange, ReusedExchangeExec, ShuffleExchangeExec} | ||
|
|
||
| /** | ||
| * Divide the spark plan into multiple QueryStages. For each Exchange in the plan, it wraps it with | ||
| * a [[QueryStage]]. At the end it adds an [[AdaptiveSparkPlan]] at the top, which will drive the | ||
| * execution of query stages. | ||
| */ | ||
| case class PlanQueryStage(session: SparkSession) extends Rule[SparkPlan] { | ||
|
|
||
| def apply(plan: SparkPlan): SparkPlan = { | ||
| var id = 0 | ||
| val exchangeToQueryStage = new java.util.IdentityHashMap[Exchange, QueryStage] | ||
| val planWithStages = plan.transformUp { | ||
| case e: ShuffleExchangeExec => | ||
| val queryStage = ShuffleQueryStage(id, e) | ||
|
||
| id += 1 | ||
| exchangeToQueryStage.put(e, queryStage) | ||
| queryStage | ||
| case e: BroadcastExchangeExec => | ||
| val queryStage = BroadcastQueryStage(id, e) | ||
| id += 1 | ||
| exchangeToQueryStage.put(e, queryStage) | ||
| queryStage | ||
| // The `ReusedExchangeExec` was added in the rule `ReuseExchange`, via transforming up the | ||
| // query plan. This rule also transform up the query plan, so when we hit `ReusedExchangeExec` | ||
| // here, the exchange being reused must already be hit before and there should be an entry | ||
| // for it in `exchangeToQueryStage`. | ||
| case e: ReusedExchangeExec => | ||
| val existingQueryStage = exchangeToQueryStage.get(e.child) | ||
| assert(existingQueryStage != null, "The exchange being reused should be hit before.") | ||
|
||
| ReusedQueryStage(existingQueryStage, e.output) | ||
|
||
| } | ||
| AdaptiveSparkPlan(ResultQueryStage(id, planWithStages), session) | ||
| } | ||
| } | ||
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
super nit: we can simply write
_ > 0