hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sandy Ryza (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-758) Fair scheduler has some bug that causes TestRMRestart to fail
Date Tue, 04 Jun 2013 02:07:21 GMT

    [ https://issues.apache.org/jira/browse/YARN-758?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13673944#comment-13673944
] 

Sandy Ryza commented on YARN-758:
---------------------------------

It looks like the problem stems from that, after YARN-326, the fair scheduler takes a node's
CPU capacity into account.  The MockNMs only get instantiated with 1 vcore, so in the fair
scheduler's eyes they don't have space for the non-AM containers that the test expects them
to.  This is the expected behavior for the fair scheduler.

It might be good to add a timeout to the test so that scheduler-side issues like this don't
cause it to spin indefinitely in the future?  Also, maybe we should modify MockNM to take
a CPU argument and/or have its number of vcores default to memoryArg * (default yarn.nodemanager.resource.cpu-vcores)
/ (default yarn.nodemanager.resource.memory-mb).
                
> Fair scheduler has some bug that causes TestRMRestart to fail
> -------------------------------------------------------------
>
>                 Key: YARN-758
>                 URL: https://issues.apache.org/jira/browse/YARN-758
>             Project: Hadoop YARN
>          Issue Type: Bug
>    Affects Versions: 2.1.0-beta
>            Reporter: Bikas Saha
>            Assignee: Sandy Ryza
>
> YARN-757 got fixed by changing the scheduler from Fair to default (which is capacity).

--
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

Mime
View raw message