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 "Varun Vasudev (JIRA)" <ji...@apache.org> on 2016/03/30 11:18:26 UTC

[jira] [Resolved] (YARN-4776) Report container metrics back to the RM

     [ https://issues.apache.org/jira/browse/YARN-4776?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Varun Vasudev resolved YARN-4776.
---------------------------------
    Resolution: Won't Fix

After talking with [~vinodkv] offline - it turns out that this has been discussed earlier(with respect to scheduling based on actual resource usage) and there are fears that it will put too much load on the RM. With ATS v2, the aggregators running on the node can store the resource usage in the ATS itself. Closing this as won't fix.

> Report container metrics back to the RM
> ---------------------------------------
>
>                 Key: YARN-4776
>                 URL: https://issues.apache.org/jira/browse/YARN-4776
>             Project: Hadoop YARN
>          Issue Type: Improvement
>            Reporter: Varun Vasudev
>            Assignee: Varun Vasudev
>
> Today we generate some pretty useful container metrics(like actual memory and cpu usage) which are discarded when the container finishes. We should report these metrics back to the RM so that they can be collected and analyzed while the app is running/after it finishes.



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