[HIVE-12629] hive.auto.convert.join=true makes lateral view join sql failed on spark engine on yarn

• Type: Bug
• Status: Patch Available
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 1.2.1
• Component/s: Spark
• Assignee: Chao Sun
• Reporter: 吴子美
• Created: 09/Dec/15 14:00
• Updated: 25/Jan/16 20:54

Description

I am using hive1.2 on spark on yarn.

I found
select count(1) from
(select user_id from xxx group by user_id ) a join
(select user_id from yyy lateral view json_tuple(u, 'h') v1 as h) b
on a.user_id=b.user_id ;
failed in hive on spark on yarn, but OK in hive on MR.

I tried the following sql on spark. It was OK.
select count(1) from
(select user_id from xxx group by user_id ) a left join
(select user_id from yyy lateral view json_tuple(u, 'h') v1 as h) b
on a.user_id=b.user_id ;

When I turn hive.auto.convert.join from true to false. Everything goes OK.

The error message in hive.log was :

2015-12-09 21:10:17,190 INFO  [stderr-redir-1]: client.SparkClientImpl (SparkClientImpl.java:run(569)) - 15/12/09 21:10:17 INFO log.PerfLogger: <PERFLOG method=serializePlan from=org.apache.hadoop.hive.ql.exec.Utilities>
2015-12-09 21:10:17,190 INFO  [stderr-redir-1]: client.SparkClientImpl (SparkClientImpl.java:run(569)) - 15/12/09 21:10:17 INFO exec.Utilities: Serializing ReduceWork via kryo
2015-12-09 21:10:17,214 INFO  [stderr-redir-1]: client.SparkClientImpl (SparkClientImpl.java:run(569)) - 15/12/09 21:10:17 INFO log.PerfLogger: </PERFLOG method=serializePlan start=1449666617190 end=1449666617214 duration=24 from=org.apache.hadoop.hive.ql.exec.Utilities>
2015-12-09 21:10:17,261 INFO  [stderr-redir-1]: client.SparkClientImpl (SparkClientImpl.java:run(569)) - 15/12/09 21:10:17 INFO client.RemoteDriver: Failed to run job 8fed1ca8-834f-497f-b189-eab343440a9f
2015-12-09 21:10:17,261 INFO  [stderr-redir-1]: client.SparkClientImpl (SparkClientImpl.java:run(569)) - java.lang.IllegalStateException: Connection already exists
2015-12-09 21:10:17,261 INFO  [stderr-redir-1]: client.SparkClientImpl (SparkClientImpl.java:run(569)) - 	at org.apache.hadoop.hive.ql.exec.spark.SparkPlan.connect(SparkPlan.java:142)
2015-12-09 21:10:17,261 INFO  [stderr-redir-1]: client.SparkClientImpl (SparkClientImpl.java:run(569)) - 	at org.apache.hadoop.hive.ql.exec.spark.SparkPlanGenerator.generateParentTran(SparkPlanGenerator.java:142)
2015-12-09 21:10:17,261 INFO  [stderr-redir-1]: client.SparkClientImpl (SparkClientImpl.java:run(569)) - 	at org.apache.hadoop.hive.ql.exec.spark.SparkPlanGenerator.generate(SparkPlanGenerator.java:106)
2015-12-09 21:10:17,261 INFO  [stderr-redir-1]: client.SparkClientImpl (SparkClientImpl.java:run(569)) - 	at org.apache.hadoop.hive.ql.exec.spark.RemoteHiveSparkClient$JobStatusJob.call(RemoteHiveSparkClient.java:252)
2015-12-09 21:10:17,261 INFO  [stderr-redir-1]: client.SparkClientImpl (SparkClientImpl.java:run(569)) - 	at org.apache.hive.spark.client.RemoteDriver$JobWrapper.call(RemoteDriver.java:366)
2015-12-09 21:10:17,261 INFO  [stderr-redir-1]: client.SparkClientImpl (SparkClientImpl.java:run(569)) - 	at org.apache.hive.spark.client.RemoteDriver$JobWrapper.call(RemoteDriver.java:335)
2015-12-09 21:10:17,261 INFO  [stderr-redir-1]: client.SparkClientImpl (SparkClientImpl.java:run(569)) - 	at java.util.concurrent.FutureTask.run(FutureTask.java:262)
2015-12-09 21:10:17,262 INFO  [stderr-redir-1]: client.SparkClientImpl (SparkClientImpl.java:run(569)) - 	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
2015-12-09 21:10:17,262 INFO  [stderr-redir-1]: client.SparkClientImpl (SparkClientImpl.java:run(569)) - 	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
2015-12-09 21:10:17,262 INFO  [stderr-redir-1]: client.SparkClientImpl (SparkClientImpl.java:run(569)) - 	at java.lang.Thread.run(Thread.java:745)
2015-12-09 21:10:17,266 INFO  [RPC-Handler-3]: client.SparkClientImpl (SparkClientImpl.java:handle(522)) - Received result for 8fed1ca8-834f-497f-b189-eab343440a9f
2015-12-09 21:10:18,054 ERROR [HiveServer2-Background-Pool: Thread-43]: status.SparkJobMonitor (SessionState.java:printError(960)) - Status: Failed
2015-12-09 21:10:18,055 INFO  [HiveServer2-Background-Pool: Thread-43]: log.PerfLogger (PerfLogger.java:PerfLogEnd(148)) - </PERFLOG method=SparkRunJob start=1449666615051 end=1449666618055 duration=3004 from=org.apache.hadoop.hive.ql.exec.spark.status.SparkJobMonitor>
2015-12-09 21:10:18,076 ERROR [HiveServer2-Background-Pool: Thread-43]: ql.Driver (SessionState.java:printError(960)) - FAILED: Execution Error, return code 3 from org.apache.hadoop.hive.ql.exec.spark.SparkTask

Is it the bug of hive on spark?