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 "Steve Loughran (JIRA)" <ji...@apache.org> on 2016/01/14 21:50:39 UTC

[jira] [Commented] (YARN-4593) Deadlock in AbstractService.getConfig()

    [ https://issues.apache.org/jira/browse/YARN-4593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15098825#comment-15098825 ] 

Steve Loughran commented on YARN-4593:
--------------------------------------

I don't see why we need this to be synchronized (it's that way in branch-205); the actual type is {{volatile}} so access is thread safe anyway.


> Deadlock in AbstractService.getConfig()
> ---------------------------------------
>
>                 Key: YARN-4593
>                 URL: https://issues.apache.org/jira/browse/YARN-4593
>             Project: Hadoop YARN
>          Issue Type: Bug
>          Components: yarn
>    Affects Versions: 2.7.2
>         Environment: AM restarting on kerberized cluster
>            Reporter: Steve Loughran
>            Assignee: Steve Loughran
>
> SLIDER-1052 has found a deadlock which can arise in it during AM restart. Looking at the thread trace, one of the blockages is actually {{AbstractService.getConfig()}} —this is synchronized and so blocked.



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