[
https://issues.apache.org/jira/browse/HIVE-7651?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14093083#comment-14093083
]
Na Yang commented on HIVE-7651:
-------------------------------
This issue was caused by a single jobConf instance are used by multiple
MapTrans. The fix is included in the patch of HIVE-7541.
> Investigate why union two RDDs generated from two MapTrans does not get the
> right result
> ----------------------------------------------------------------------------------------
>
> Key: HIVE-7651
> URL: https://issues.apache.org/jira/browse/HIVE-7651
> Project: Hive
> Issue Type: Bug
> Components: Spark
> Reporter: Na Yang
>
> If the SparkWork has two map works as root, then use the current
> generate(basework) API to generate two mapTran. union the RDDs processed by
> the two mapTrans does not generate the correct result.
> If two input RDDs come from different data tables, then the union result is
> empty.
> if two input RDDs come from the same data table, then the union result is not
> correct. The same row of data happen 4 times in the union result.
> Need to investigate why this happen and how to fix it.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
