hive-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Chao (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HIVE-8913) Make SparkMapJoinResolver handle runtime skew join [Spark Branch]
Date Thu, 11 Dec 2014 04:51:12 GMT

    [ https://issues.apache.org/jira/browse/HIVE-8913?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14242148#comment-14242148
] 

Chao commented on HIVE-8913:
----------------------------

Looking at the log, something interesting..

{noformat}
2014-12-10 05:28:26,822 INFO  [Executor task launch worker-0]: exec.GroupByOperator (Operator.java:close(613))
- 44 Close done
2014-12-10 05:28:26,822 INFO  [Executor task launch worker-0]: exec.FilterOperator (Operator.java:close(613))
- 43 Close done
2014-12-10 05:28:26,822 INFO  [Executor task launch worker-0]: exec.ForwardOperator (Operator.java:close(613))
- 38 Close done
2014-12-10 05:28:26,828 INFO  [Executor task launch worker-1]: executor.Executor (Logging.scala:logInfo(59))
- Finished task 2.0 in stage 4.0 (TID 6). 1170 bytes result sent to driver
2014-12-10 05:28:26,828 INFO  [Executor task launch worker-0]: executor.Executor (Logging.scala:logInfo(59))
- Finished task 1.0 in stage 4.0 (TID 5). 1170 bytes result sent to driver
....
2014-12-10 05:28:26,979 INFO  [Executor task launch worker-0]: log.PerfLogger (PerfLogger.java:PerfLogEnd(135))
- </PERFLOG method=deserializePlan start=1418218106762 end=1418218106979 duration=217 from=org.apache.hadoop.hive.ql.exec.Utilities>
2014-12-10 05:28:26,980 INFO  [Executor task launch worker-0]: mr.ObjectCache (ObjectCache.java:cache(36))
- Ignoring cache key: __MAP_PLAN__
2014-12-10 05:28:26,982 ERROR [Executor task launch worker-0]: executor.Executor (Logging.scala:logError(96))
- Exception in task 0.0 in stage 2.0 (TID 2)
java.lang.RuntimeException: Map operator initialization failed
{noformat}

At first, worker-0 is running a reduce work. After finishing that, it is switched to this
ShuffleMapTask, which apparently doesn't go through {{HadoopRDD#compute}}, and thus IOContext
is left uninitialized.

> Make SparkMapJoinResolver handle runtime skew join [Spark Branch]
> -----------------------------------------------------------------
>
>                 Key: HIVE-8913
>                 URL: https://issues.apache.org/jira/browse/HIVE-8913
>             Project: Hive
>          Issue Type: Improvement
>          Components: Spark
>            Reporter: Rui Li
>            Assignee: Rui Li
>         Attachments: HIVE-8913.1-spark.patch, HIVE-8913.2-spark.patch
>
>
> Sub-task of HIVE-8406.
> Now we have {{SparkMapJoinResolver}} in place. But at the moment, it doesn't handle the
map join task created by upstream SkewJoinResolver, i.e. those wrapped in a ConditionalTask.
We have to implement this part for runtime skew join to work on spark. To do so, we can borrow
logic from {{MapJoinResolver}}.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message