[
https://issues.apache.org/jira/browse/YARN-460?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13597226#comment-13597226
]
Thomas Graves commented on YARN-460:
------------------------------------
We think this might be a race between when the application gets removed
(LeafQueue.removeApplication) and between a possibly in flight allocate call
from the AM. If LeafQueue.removeApplication is called it removes the user from
the list of active users. If a CapacityScheduler.allocate call comes in before
the application gets removed from application datastructures in
CapacityScheduler.doneApplication, it could add it back to the activeUser list
because the allocate() call just checks to make sure application isn't null.
We either need to make the check in allocate better or prevent the race between
finish and removing.
This is basically in CapacityScheduler.doneApplication:
} else {
queue.finishApplication(application, queue.getQueueName());
}
// Remove from our data-structure
applications.remove(applicationAttemptId);
> CS user left in list of active users for the queue even when application
> finished
> ---------------------------------------------------------------------------------
>
> Key: YARN-460
> URL: https://issues.apache.org/jira/browse/YARN-460
> Project: Hadoop YARN
> Issue Type: Bug
> Components: capacityscheduler
> Affects Versions: 0.23.7, 2.0.4-alpha
> Reporter: Thomas Graves
> Assignee: Thomas Graves
> Priority: Critical
>
> We have seen a user get left in the queues list of active users even though
> the application was removed. This can cause everyone else in the queue to get
> less resources if using the minimum user limit percent config.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
