[
https://issues.apache.org/jira/browse/HIVE-10458?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14551902#comment-14551902
]
Rui Li commented on HIVE-10458:
-------------------------------
Hi [~xuefuz], I believe there's nothing special for MR when fetching the
results. Anyway, the fetch task doesn't even need an execution engine to run.
And if I mess up the order of the splits, we always get incorrect results, no
matter whether we're on MR or spark.
Did a little bit more search and found this may be related to HADOOP-10798. In
the comments, guys did mention that local filesystem doesn't guarantee order on
Linux. I think it's because local filesystem uses {{File::list}} whose JavaDoc
specifies there's no specific order of the returned files.
> Enable parallel order by for spark [Spark Branch]
> -------------------------------------------------
>
> Key: HIVE-10458
> URL: https://issues.apache.org/jira/browse/HIVE-10458
> Project: Hive
> Issue Type: Sub-task
> Components: Spark
> Reporter: Rui Li
> Assignee: Rui Li
> Attachments: HIVE-10458.1-spark.patch, HIVE-10458.2-spark.patch,
> HIVE-10458.3-spark.patch
>
>
> We don't have to force reducer# to 1 as spark supports parallel sorting.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
