You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@crunch.apache.org by "Andrew Olson (JIRA)" <ji...@apache.org> on 2019/01/23 18:59:00 UTC
[jira] [Commented] (CRUNCH-675) HFileTarget should use DistCp when
source and destination are in different filesystems
[ https://issues.apache.org/jira/browse/CRUNCH-675?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16750327#comment-16750327 ]
Andrew Olson commented on CRUNCH-675:
-------------------------------------
Pull request,
https://github.com/apache/crunch/pull/17
> HFileTarget should use DistCp when source and destination are in different filesystems
> --------------------------------------------------------------------------------------
>
> Key: CRUNCH-675
> URL: https://issues.apache.org/jira/browse/CRUNCH-675
> Project: Crunch
> Issue Type: Improvement
> Reporter: Andrew Olson
> Priority: Major
>
> For jobs using the HFileTarget we have found that running a DistCp to handleOutputs at the conclusion of a job, when the source and destination are in different filesystems, provides a substantial performance benefit compared to a single-threaded data copy. This enhancement is similar to what was suggested in CRUNCH-660.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)