spark-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Thomas Graves (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (SPARK-15671) performance regression CoalesceRDD large # partitions
Date Wed, 01 Jun 2016 15:34:59 GMT

    [ https://issues.apache.org/jira/browse/SPARK-15671?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15310487#comment-15310487
] 

Thomas Graves commented on SPARK-15671:
---------------------------------------

Note the performance impact is in the 10's of minutes with large # of partitions.  With the
change to pickBin reverted it takes about 10 seconds.  For now I think we revert it and we
can look at possible other optimizations later.

> performance regression CoalesceRDD large # partitions
> -----------------------------------------------------
>
>                 Key: SPARK-15671
>                 URL: https://issues.apache.org/jira/browse/SPARK-15671
>             Project: Spark
>          Issue Type: Bug
>          Components: Spark Core
>    Affects Versions: 2.0.0
>            Reporter: Thomas Graves
>            Priority: Critical
>
> I was running a 15TB join job with 202000 partitions. It looks like the changes I made
to CoalesceRDD in pickBin() are really slow with that large of partitions.  The array filter
with that many elements just takes to long.
>  It took about an hour for it to pickBins for all the partitions.
> original change:
> https://github.com/apache/spark/commit/83ee92f60345f016a390d61a82f1d924f64ddf90
> Just reverting the pickBin code back to get currpreflocs fixes the issue



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

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org


Mime
View raw message