You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@kylin.apache.org by xx...@apache.org on 2023/07/06 05:53:24 UTC
[kylin] 07/19: KYLIN-5601 Temporarily Close AQE for dict build job (#30251)
This is an automated email from the ASF dual-hosted git repository.
xxyu pushed a commit to branch kylin5
in repository https://gitbox.apache.org/repos/asf/kylin.git
commit 931854ad7cf7053366eb54877f6358eeae0d7343
Author: huangsheng <hu...@163.com>
AuthorDate: Fri Apr 7 14:04:10 2023 +0800
KYLIN-5601 Temporarily Close AQE for dict build job (#30251)
* KYLIN-5601 Temporarily Close AQE for dict build job
---
.../org/apache/kylin/common/KylinConfigBase.java | 4 +++
.../engine/spark/builder/DFDictionaryBuilder.scala | 18 ++++++++++
.../engine/spark/builder/TestGlobalDictBuild.scala | 41 ++++++++++++++++++++++
3 files changed, 63 insertions(+)
diff --git a/src/core-common/src/main/java/org/apache/kylin/common/KylinConfigBase.java b/src/core-common/src/main/java/org/apache/kylin/common/KylinConfigBase.java
index 4a852aa1e4..9557ede4d5 100644
--- a/src/core-common/src/main/java/org/apache/kylin/common/KylinConfigBase.java
+++ b/src/core-common/src/main/java/org/apache/kylin/common/KylinConfigBase.java
@@ -1632,6 +1632,10 @@ public abstract class KylinConfigBase implements Serializable {
return Boolean.parseBoolean(getOptional("kylin.engine.global-dict-check-enabled", FALSE));
}
+ public boolean isGlobalDictAQEEnabled() {
+ return Boolean.parseBoolean(getOptional("kylin.engine.global-dict-aqe-enabled", FALSE));
+ }
+
public String getJdbcSourceName() {
return getOptional(KYLIN_SOURCE_JDBC_SOURCE_NAME_KEY);
}
diff --git a/src/spark-project/engine-spark/src/main/scala/org/apache/kylin/engine/spark/builder/DFDictionaryBuilder.scala b/src/spark-project/engine-spark/src/main/scala/org/apache/kylin/engine/spark/builder/DFDictionaryBuilder.scala
index 992bd6b70c..5dfc35eeac 100644
--- a/src/spark-project/engine-spark/src/main/scala/org/apache/kylin/engine/spark/builder/DFDictionaryBuilder.scala
+++ b/src/spark-project/engine-spark/src/main/scala/org/apache/kylin/engine/spark/builder/DFDictionaryBuilder.scala
@@ -44,9 +44,12 @@ class DFDictionaryBuilder(
@throws[IOException]
def buildDictSet(): Unit = {
colRefSet.asScala.foreach(col => safeBuild(col))
+ changeAQEConfig(true)
}
private val YARN_CLUSTER: String = "cluster"
+ private val AQE = "spark.sql.adaptive.enabled";
+ private val originalAQE = ss.conf.get(AQE)
private def tryZKJaasConfiguration(): Unit = {
val config = KylinConfig.getInstanceFromEnv
@@ -79,6 +82,18 @@ class DFDictionaryBuilder(
} finally lock.unlock()
}
+ // Workaround: https://olapio.atlassian.net/browse/KE-41645
+ private[builder] def changeAQEConfig(isDictBuildFinished: Boolean = false) : Boolean = {
+ if (!seg.getConfig.isGlobalDictAQEEnabled && !isDictBuildFinished) {
+ logInfo("Temporarily Close AQE for dict build job")
+ ss.conf.set(AQE, false)
+ return false
+ }
+ logInfo(s"Restore AQE to its initial config: $originalAQE")
+ ss.conf.set(AQE, originalAQE)
+ originalAQE.toBoolean
+ }
+
@throws[IOException]
private[builder] def build(ref: TblColRef, bucketPartitionSize: Int,
afterDistinct: Dataset[Row]): Unit = logTime(s"building global dictionaries V2 for ${ref.getIdentity}") {
@@ -86,7 +101,9 @@ class DFDictionaryBuilder(
globalDict.prepareWrite()
val broadcastDict = ss.sparkContext.broadcast(globalDict)
+ changeAQEConfig(false)
ss.sparkContext.setJobDescription("Build dict " + ref.getIdentity)
+
val dictCol = col(afterDistinct.schema.fields.head.name)
afterDistinct.filter(dictCol.isNotNull)
.repartition(bucketPartitionSize, dictCol)
@@ -118,6 +135,7 @@ class DFDictionaryBuilder(
}
logInfo(s"Global dict correctness check completed, table: ${ref.getTableAlias}, col: ${ref.getName}")
}
+ changeAQEConfig(true)
}
private def getLockPath(pathName: String) = s"/${seg.getProject}${HadoopUtil.GLOBAL_DICT_STORAGE_ROOT}/$pathName/lock"
diff --git a/src/spark-project/engine-spark/src/test/scala/org/apache/kylin/engine/spark/builder/TestGlobalDictBuild.scala b/src/spark-project/engine-spark/src/test/scala/org/apache/kylin/engine/spark/builder/TestGlobalDictBuild.scala
index 30dcd4a961..764229ff44 100644
--- a/src/spark-project/engine-spark/src/test/scala/org/apache/kylin/engine/spark/builder/TestGlobalDictBuild.scala
+++ b/src/spark-project/engine-spark/src/test/scala/org/apache/kylin/engine/spark/builder/TestGlobalDictBuild.scala
@@ -164,6 +164,47 @@ class TestGlobalDictBuild extends SparderBaseFunSuite with SharedSparkSession wi
Assert.assertEquals(shufflePartitionSizeInt, dictDirSize)
}
+
+ test("global dict build and close aqe") {
+ val dsMgr: NDataflowManager = NDataflowManager.getInstance(getTestConfig, DEFAULT_PROJECT)
+ val df: NDataflow = dsMgr.getDataflow(CUBE_NAME)
+ val seg = df.getLastSegment
+ val nSpanningTree = NSpanningTreeFactory.fromLayouts(seg.getIndexPlan.getAllLayouts, df.getUuid)
+ val dictColSet = DictionaryBuilderHelper.extractTreeRelatedGlobalDicts(seg, nSpanningTree.getAllIndexEntities)
+ val randomDataSet = generateOriginData(200, 10)
+
+ val dictionaryBuilder = new DFDictionaryBuilder(randomDataSet, seg, spark, dictColSet)
+ val col = dictColSet.iterator().next()
+ val ds = randomDataSet.select("26").distinct()
+ val bucketPartitionSize = DictionaryBuilderHelper.calculateBucketSize(seg, col, ds)
+
+ val originalAQE = spark.conf.get("spark.sql.adaptive.enabled")
+
+ // false false
+ seg.getConfig.setProperty("kylin.engine.global-dict-aqe-enabled", "FALSE")
+ dictionaryBuilder.changeAQEConfig(false)
+ Assert.assertFalse(spark.conf.get("spark.sql.adaptive.enabled").toBoolean)
+ dictionaryBuilder.build(col, bucketPartitionSize, ds)
+ Assert.assertTrue(spark.conf.get("spark.sql.adaptive.enabled").equals(originalAQE))
+
+ // false true
+ dictionaryBuilder.changeAQEConfig(true)
+ Assert.assertTrue(spark.conf.get("spark.sql.adaptive.enabled").equals(originalAQE))
+
+ // true false
+ seg.getConfig.setProperty("kylin.engine.global-dict-aqe-enabled", "TRUE")
+ dictionaryBuilder.changeAQEConfig(false)
+ Assert.assertTrue(spark.conf.get("spark.sql.adaptive.enabled").equals(originalAQE))
+
+ dictionaryBuilder.build(col, bucketPartitionSize, ds)
+ Assert.assertTrue(spark.conf.get("spark.sql.adaptive.enabled").equals(originalAQE))
+
+ // true true
+ dictionaryBuilder.changeAQEConfig(true)
+ Assert.assertTrue(spark.conf.get("spark.sql.adaptive.enabled").equals(originalAQE))
+
+ }
+
def buildDict(seg: NDataSegment, randomDataSet: Dataset[Row], dictColSet: Set[TblColRef]): NGlobalDictMetaInfo = {
val dictionaryBuilder = new DFDictionaryBuilder(randomDataSet, seg, randomDataSet.sparkSession, dictColSet)
val col = dictColSet.iterator().next()