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 "Karthik Kambatla (JIRA)" <ji...@apache.org> on 2015/05/08 19:46:01 UTC

[jira] [Commented] (YARN-3259) FairScheduler: Update to fairShare could be triggered early on node events instead of waiting for update interval

    [ https://issues.apache.org/jira/browse/YARN-3259?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14535067#comment-14535067 ] 

Karthik Kambatla commented on YARN-3259:
----------------------------------------

The patch doesn't compile because we are attempting to catch InterruptedException even when there is no code calling it. 

Comments on the patch:
# On InterrruptedException on {{wait()}}, we should actually quit the thread and not do nothing. May be, remove the inner try-catch altogether.
# Would be nice to add a unit test to verify it is called either on the interval or on a qualifying condition, whichever is earlier. 

> FairScheduler: Update to fairShare could be triggered early on node events instead of waiting for update interval 
> ------------------------------------------------------------------------------------------------------------------
>
>                 Key: YARN-3259
>                 URL: https://issues.apache.org/jira/browse/YARN-3259
>             Project: Hadoop YARN
>          Issue Type: Improvement
>          Components: fairscheduler
>            Reporter: Anubhav Dhoot
>            Assignee: Anubhav Dhoot
>         Attachments: YARN-3259.001.patch
>
>
> Instead of waiting for update interval unconditionally, we can trigger early updates on important events - for eg node join and leave.



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