hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jason Lowe (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-4839) ResourceManager deadlock between RMAppAttemptImpl and SchedulerApplicationAttempt
Date Fri, 18 Mar 2016 16:00:34 GMT

    [ https://issues.apache.org/jira/browse/YARN-4839?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15201687#comment-15201687
] 

Jason Lowe commented on YARN-4839:
----------------------------------

bq. Could this be the same issue as pointed out by YARN-4247?

It is essentially the same core issue, but it wasn't caused by YARN-2005.  We don't have that
change in our build, but we do have YARN-3116.  That's the first time getMasterContainer was
called from SchedulerApplicationAttempt.  Without the side-effect of YARN-3361 it leads to
a deadlock since getMasterContainer tries to grab the lock.


> ResourceManager deadlock between RMAppAttemptImpl and SchedulerApplicationAttempt
> ---------------------------------------------------------------------------------
>
>                 Key: YARN-4839
>                 URL: https://issues.apache.org/jira/browse/YARN-4839
>             Project: Hadoop YARN
>          Issue Type: Bug
>          Components: resourcemanager
>    Affects Versions: 2.8.0
>            Reporter: Jason Lowe
>            Priority: Blocker
>
> Hit a deadlock in the ResourceManager as one thread was holding the SchedulerApplicationAttempt
lock and trying to call RMAppAttemptImpl.getMasterContainer while another thread had the RMAppAttemptImpl
lock and was trying to call SchedulerApplicationAttempt.getResourceUsageReport.



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

Mime
View raw message