[jira] [Commented] (MAPREDUCE-2441) regression: maximum limit of -1 + user-lmit math appears to be off

["Allen Wittenauer (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/MAPREDUCE-2441?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13032165#comment-13032165 ] 

Allen Wittenauer commented on MAPREDUCE-2441:
---------------------------------------------

Actually, it looks like queue spillage/task stealing doesn't work at all, whether it is -1 or not.  The problem code appears to be in assignSlotsToJob which appears to have replaced the two-phase system in previous versions with a single phase.  This single phase does this check to determine the limit:

{code}
int limit =
      Math.min(
          Math.max(divideAndCeil(currentCapacity, activeUsers),
                   divideAndCeil(ulMin*currentCapacity, 100)),
          (int)(queueCapacity * ulMinFactor)
          );

{code}

In a two queue system where one is -1 and the other is a number, the maximum queue capacity ends up being set to the remainder.  Without a second pass, any additional slots from other queues are essentially ignored.

> regression: maximum limit of -1 + user-lmit math appears to be off
> ------------------------------------------------------------------
>
>                 Key: MAPREDUCE-2441
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-2441
>             Project: Hadoop Map/Reduce
>          Issue Type: Bug
>          Components: contrib/capacity-sched
>    Affects Versions: 0.20.203.0
>            Reporter: Allen Wittenauer
>            Priority: Blocker
>         Attachments: capsched.xml
>
>
> The math around the slot usage when maximum-capacity=-1 appears to be faulty.  See comments.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira