[
https://issues.apache.org/jira/browse/MAPREDUCE-2762?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13117738#comment-13117738
]
Mahadev konar commented on MAPREDUCE-2762:
------------------------------------------
You are right Vinod.
Thinking abt this more, I think we should move the unregister call out of the
stop() methods. We could make sure that as soon as unregister is called, we
send an event back to JobImpl saying done with the RM, and then we can go ahead
to delete the staging dir.
So, there'd be a state transition from
RUNNING->SUCCEEDED that would sent an unregister event to RMCommunicator.
and then
SUCCEEDED -> SUCCEEDED in JobImpl, that would happen on an event triggered
after unregister is called, and the same for Killed and Failed, which would
delete the staging directory.
Do you guys see any issues with that?
> [MR-279] - Cleanup staging dir after job completion
> ---------------------------------------------------
>
> Key: MAPREDUCE-2762
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-2762
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: mrv2
> Affects Versions: 0.23.0
> Reporter: Ramya Sunil
> Assignee: Mahadev konar
> Priority: Blocker
> Fix For: 0.23.0
>
> Attachments: MAPREDUCE-2762.patch, MAPREDUCE-2762.patch,
> MAPREDUCE-2762.patch
>
>
> The files created under the staging dir have to be deleted after job
> completion. Currently, all job.* files remain forever in the
> ${yarn.apps.stagingDir}
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
