[HIVE-8610] Compile time skew join optimization doesn't work with auto map join

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 1.1.0
• Assignee: Rui Li
• Reporter: Rui Li
• Created: 27/Oct/14 03:34
• Updated: 12/Feb/15 23:41
• Resolved: 30/Oct/14 22:24

Description

NPE is thrown if both hive.optimize.skewjoin.compiletime and hive.auto.convert.join are enabled:

java.lang.NullPointerException
	at org.apache.hadoop.hive.ql.optimizer.MapJoinProcessor.convertMapJoin(MapJoinProcessor.java:329)
	at org.apache.hadoop.hive.ql.optimizer.MapJoinProcessor.genMapJoinOpAndLocalWork(MapJoinProcessor.java:236)
	at org.apache.hadoop.hive.ql.optimizer.physical.CommonJoinTaskDispatcher.convertTaskToMapJoinTask(CommonJoinTaskDispatcher.java:181)
	at org.apache.hadoop.hive.ql.optimizer.physical.CommonJoinTaskDispatcher.processCurrentTask(CommonJoinTaskDispatcher.java:463)
	at org.apache.hadoop.hive.ql.optimizer.physical.AbstractJoinTaskDispatcher.dispatch(AbstractJoinTaskDispatcher.java:182)
	at org.apache.hadoop.hive.ql.lib.TaskGraphWalker.dispatch(TaskGraphWalker.java:111)
	at org.apache.hadoop.hive.ql.lib.TaskGraphWalker.walk(TaskGraphWalker.java:194)
	at org.apache.hadoop.hive.ql.lib.TaskGraphWalker.startWalking(TaskGraphWalker.java:139)
	at org.apache.hadoop.hive.ql.optimizer.physical.CommonJoinResolver.resolve(CommonJoinResolver.java:79)
	at org.apache.hadoop.hive.ql.optimizer.physical.PhysicalOptimizer.optimize(PhysicalOptimizer.java:107)
	at org.apache.hadoop.hive.ql.parse.MapReduceCompiler.optimizeTaskPlan(MapReduceCompiler.java:275)
	at org.apache.hadoop.hive.ql.parse.TaskCompiler.compile(TaskCompiler.java:223)
	at org.apache.hadoop.hive.ql.parse.SemanticAnalyzer.analyzeInternal(SemanticAnalyzer.java:10028)
	at org.apache.hadoop.hive.ql.parse.BaseSemanticAnalyzer.analyze(BaseSemanticAnalyzer.java:221)
	at org.apache.hadoop.hive.ql.parse.ExplainSemanticAnalyzer.analyzeInternal(ExplainSemanticAnalyzer.java:74)
	at org.apache.hadoop.hive.ql.parse.BaseSemanticAnalyzer.analyze(BaseSemanticAnalyzer.java:221)
	at org.apache.hadoop.hive.ql.Driver.compile(Driver.java:415)
	at org.apache.hadoop.hive.ql.Driver.compile(Driver.java:303)
	at org.apache.hadoop.hive.ql.Driver.compileInternal(Driver.java:1068)
	at org.apache.hadoop.hive.ql.Driver.runInternal(Driver.java:1130)
	at org.apache.hadoop.hive.ql.Driver.run(Driver.java:1005)
	at org.apache.hadoop.hive.ql.Driver.run(Driver.java:995)
	at org.apache.hadoop.hive.cli.CliDriver.processLocalCmd(CliDriver.java:246)
	at org.apache.hadoop.hive.cli.CliDriver.processCmd(CliDriver.java:198)
	at org.apache.hadoop.hive.cli.CliDriver.processLine(CliDriver.java:408)
	at org.apache.hadoop.hive.cli.CliDriver.executeDriver(CliDriver.java:781)
	at org.apache.hadoop.hive.cli.CliDriver.run(CliDriver.java:675)
	at org.apache.hadoop.hive.cli.CliDriver.main(CliDriver.java:614)
	at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
	at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:57)
	at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
	at java.lang.reflect.Method.invoke(Method.java:606)
	at org.apache.hadoop.util.RunJar.main(RunJar.java:212)

A simple way to reproduce this issue is to add set hive.auto.convert.join=true to one of the skew join qfile, e.g. skewjoinopt2.q.

While reduce side join can produce correct results, we kind of lost the point of skew join optimization - join skewed data via a map join to avoid one reducer getting too much records.

Issue Links

• is related to: HIVE-8535 Enable compile time skew join optimization for spark [Spark Branch]
• links to: RB request