You are viewing a plain text version of this content. The canonical link for it is here.
Posted to yarn-issues@hadoop.apache.org by "Sunil Govindan (JIRA)" <ji...@apache.org> on 2019/07/02 06:57:00 UTC
[jira] [Commented] (YARN-9644) First RMContext object is always
leaked during switch over
[ https://issues.apache.org/jira/browse/YARN-9644?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16876730#comment-16876730 ]
Sunil Govindan commented on YARN-9644:
--------------------------------------
Thanks [~bibinchundatt]. Pushed to trunk. Conflicts in branch-3.2
Cud u pls share a branch-3.2/3.1 patch.
> First RMContext object is always leaked during switch over
> ----------------------------------------------------------
>
> Key: YARN-9644
> URL: https://issues.apache.org/jira/browse/YARN-9644
> Project: Hadoop YARN
> Issue Type: Bug
> Reporter: Bibin A Chundatt
> Assignee: Bibin A Chundatt
> Priority: Critical
> Attachments: YARN-9644.001.patch, YARN-9644.002.patch, YARN-9644.003.patch
>
>
> As per my understanding following 2 issues causes the issue.
> * WebApp holds the reference to First applicationMasterServer instance, which has rmcontext with ActiveServiceContext (holds RMApps + nodes map). WebApp remains to life time of RM process.
> * On transistion to active RMNMInfo object is registered in MBean and never unregistered on transitionToStandBy
> On transistion to Standby and again based to active new RMContext gets created, but above 2 issues causes first RMcontext persist still RMShutdown.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org