[
https://issues.apache.org/jira/browse/OOZIE-157?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13099843#comment-13099843
]
Hadoop QA commented on OOZIE-157:
---------------------------------
brookwc remarked:
Along the way, found another problem for purging workflow jobs/actions. There
is no limit imposed at all. Therefore if you have lots of qualified workflow
jobs/actions to purge (say > 10,000) , memory usage just blows up.
In light of this, change the issue subject to be more general.
> GH-141: Oozie uses excessive memory when doing purging
> ------------------------------------------------------
>
> Key: OOZIE-157
> URL: https://issues.apache.org/jira/browse/OOZIE-157
> Project: Oozie
> Issue Type: Bug
> Reporter: Hadoop QA
>
> When there are many coordinator jobs/actions eligible for purge, Oozie
> becomes a memory hog when purging these jobs/actions. Oozie will purge
> coordinator actions first then coordinator jobs within one transaction. First
> we delete some actions and then we have select count queries on actions, this
> causes OpenJPA to create a large number of CoordinatorActionBean objects -
> eating up lots of memory.
> To fix this, we need to have two transactions, one for deleting actions and
> one for deleting jobs. Then we will not run into this scenario where you have
> select count queries on some table with uncommitted changes.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
