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 "Bikas Saha (JIRA)" <ji...@apache.org> on 2013/06/18 07:17:22 UTC

[jira] [Created] (YARN-847) Fix loss of size information in capacity scheduler

Bikas Saha created YARN-847:
-------------------------------

             Summary: Fix loss of size information in capacity scheduler
                 Key: YARN-847
                 URL: https://issues.apache.org/jira/browse/YARN-847
             Project: Hadoop YARN
          Issue Type: Bug
    Affects Versions: 2.1.0-beta
            Reporter: Bikas Saha
            Assignee: Zhijie Shen


Capacity scheduler currently uses the last resource size set on a resource request as the resource size for requests at a given priority. If the last request has smaller size than an earlier requests then those earlier requests will be assigned containers of size < requested. This is incorrect and can stall/starve apps. Using the largest resource size results in at least always being correct.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira