cloud-fan commented on a change in pull request #24706: [SPARK-23128][SQL] A new approach to do adaptive execution in Spark SQL URL: https://github.com/apache/spark/pull/24706#discussion_r289421523
########## File path: sql/core/src/test/scala/org/apache/spark/sql/execution/adaptive/AdaptiveQueryExecSuite.scala ########## @@ -0,0 +1,275 @@ +/* + * 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.QueryTest +import org.apache.spark.sql.execution.{ReusedSubqueryExec, SparkPlan} +import org.apache.spark.sql.execution.joins.{BroadcastHashJoinExec, SortMergeJoinExec} +import org.apache.spark.sql.internal.SQLConf +import org.apache.spark.sql.test.SharedSQLContext + +class AdaptiveQueryExecSuite extends QueryTest with SharedSQLContext { + + setupTestData() + + private def runAdaptiveAndVerifyResult(query: String): (SparkPlan, SparkPlan) = { + val dfAdaptive = sql(query) + val result = dfAdaptive.collect() + withSQLConf(SQLConf.RUNTIME_REOPTIMIZATION_ENABLED.key -> "false") { + val df = sql(query) + QueryTest.sameRows(result.toSeq, df.collect().toSeq) + } + val plan = dfAdaptive.queryExecution.executedPlan + assert(plan.isInstanceOf[AdaptiveSparkPlanExec]) + val adaptivePlan = plan.asInstanceOf[AdaptiveSparkPlanExec].executedPlan + (dfAdaptive.queryExecution.sparkPlan, adaptivePlan) + } + + private def findTopLevelBroadcastHashJoin(plan: SparkPlan): Seq[BroadcastHashJoinExec] = { + plan.collect { + case j: BroadcastHashJoinExec => Seq(j) + case s: QueryStageExec => findTopLevelBroadcastHashJoin(s.plan) + }.flatten + } + + private def findTopLevelSortMergeJoin(plan: SparkPlan): Seq[SortMergeJoinExec] = { + plan.collect { + case j: SortMergeJoinExec => Seq(j) + case s: QueryStageExec => findTopLevelSortMergeJoin(s.plan) + }.flatten + } + + private def findReusedExchange(plan: SparkPlan): Seq[ReusedQueryStageExec] = { + plan.collect { + case e: ReusedQueryStageExec => Seq(e) + case a: AdaptiveSparkPlanExec => findReusedExchange(a.executedPlan) + case s: QueryStageExec => findReusedExchange(s.plan) + case p: SparkPlan => p.subqueries.flatMap(findReusedExchange) + }.flatten + } + + private def findReusedSubquery(plan: SparkPlan): Seq[ReusedSubqueryExec] = { + plan.collect { + case e: ReusedSubqueryExec => Seq(e) + case s: QueryStageExec => findReusedSubquery(s.plan) + case p: SparkPlan => p.subqueries.flatMap(findReusedSubquery) + }.flatten + } + + test("Change merge join to broadcast join") { + withSQLConf( + SQLConf.RUNTIME_REOPTIMIZATION_ENABLED.key -> "true", + SQLConf.AUTO_BROADCASTJOIN_THRESHOLD.key -> "80") { + val (plan, adaptivePlan) = runAdaptiveAndVerifyResult( + "SELECT * FROM testData join testData2 ON key = a where value = '1'") + val smj = findTopLevelSortMergeJoin(plan) + assert(smj.size == 1) + val bhj = findTopLevelBroadcastHashJoin(adaptivePlan) + assert(bhj.size == 1) + } + } + + test("Scalar subquery") { + withSQLConf( + SQLConf.RUNTIME_REOPTIMIZATION_ENABLED.key -> "true", + SQLConf.AUTO_BROADCASTJOIN_THRESHOLD.key -> "80") { + val (plan, adaptivePlan) = runAdaptiveAndVerifyResult( + "SELECT * FROM testData join testData2 ON key = a " + + "where value = (SELECT max(a) from testData3)") + val smj = findTopLevelSortMergeJoin(plan) + assert(smj.size == 1) + val bhj = findTopLevelBroadcastHashJoin(adaptivePlan) + assert(bhj.size == 1) + } + } + + test("Scalar subquery in later stages") { + withSQLConf( + SQLConf.RUNTIME_REOPTIMIZATION_ENABLED.key -> "true", + SQLConf.AUTO_BROADCASTJOIN_THRESHOLD.key -> "80") { + val (plan, adaptivePlan) = runAdaptiveAndVerifyResult( + "SELECT * FROM testData join testData2 ON key = a " + + "where (value + a) = (SELECT max(a) from testData3)") + val smj = findTopLevelSortMergeJoin(plan) + assert(smj.size == 1) + val bhj = findTopLevelBroadcastHashJoin(adaptivePlan) + assert(bhj.size == 1) + } + } + + test("multiple joins") { + withSQLConf( + SQLConf.RUNTIME_REOPTIMIZATION_ENABLED.key -> "true", + SQLConf.AUTO_BROADCASTJOIN_THRESHOLD.key -> "80") { + val (plan, adaptivePlan) = runAdaptiveAndVerifyResult( + """ + |WITH t4 AS ( + | SELECT * FROM lowercaseData t2 JOIN testData3 t3 ON t2.n = t3.a + |) + |SELECT * FROM testData + |JOIN testData2 t2 ON key = t2.a + |JOIN t4 ON key = t4.a + |WHERE value = 1 + """.stripMargin) + val smj = findTopLevelSortMergeJoin(plan) + assert(smj.size == 3) + val bhj = findTopLevelBroadcastHashJoin(adaptivePlan) + assert(bhj.size == 2) + } + } + + test("multiple joins with aggregate") { + withSQLConf( + SQLConf.RUNTIME_REOPTIMIZATION_ENABLED.key -> "true", + SQLConf.AUTO_BROADCASTJOIN_THRESHOLD.key -> "80") { + val (plan, adaptivePlan) = runAdaptiveAndVerifyResult( + """ + |WITH t4 AS ( + | SELECT * FROM lowercaseData t2 JOIN ( + | select a, sum(b) from testData3 group by a + | ) t3 ON t2.n = t3.a + |) + |SELECT * FROM testData + |JOIN testData2 t2 ON key = t2.a + |JOIN t4 ON key = t4.a + |WHERE value = 1 + """.stripMargin) + val smj = findTopLevelSortMergeJoin(plan) + assert(smj.size == 3) + val bhj = findTopLevelBroadcastHashJoin(adaptivePlan) + assert(bhj.size == 2) + } + } + + test("multiple joins with aggregate 2") { + withSQLConf( + SQLConf.RUNTIME_REOPTIMIZATION_ENABLED.key -> "true", + SQLConf.AUTO_BROADCASTJOIN_THRESHOLD.key -> "500") { + val (plan, adaptivePlan) = runAdaptiveAndVerifyResult( + """ + |WITH t4 AS ( + | SELECT * FROM lowercaseData t2 JOIN ( + | select a, max(b) b from testData2 group by a + | ) t3 ON t2.n = t3.b + |) + |SELECT * FROM testData + |JOIN testData2 t2 ON key = t2.a + |JOIN t4 ON value = t4.a + |WHERE value = 1 + """.stripMargin) + val smj = findTopLevelSortMergeJoin(plan) + assert(smj.size == 3) + val bhj = findTopLevelBroadcastHashJoin(adaptivePlan) + assert(bhj.size == 3) + } + } + + test("Exchange reuse") { + withSQLConf( + SQLConf.RUNTIME_REOPTIMIZATION_ENABLED.key -> "true", + SQLConf.AUTO_BROADCASTJOIN_THRESHOLD.key -> "80") { + val (plan, adaptivePlan) = runAdaptiveAndVerifyResult( + "SELECT value FROM testData join testData2 ON key = a " + + "join (SELECT value v from testData join testData3 ON key = a) on value = v") + val smj = findTopLevelSortMergeJoin(plan) + assert(smj.size == 3) + val bhj = findTopLevelBroadcastHashJoin(adaptivePlan) + assert(bhj.size == 2) + val ex = findReusedExchange(adaptivePlan) Review comment: the non-adaptive plan should also trigger exchange reuse? ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
