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 "Sandy Ryza (JIRA)" <ji...@apache.org> on 2013/04/15 21:18:16 UTC

[jira] [Commented] (YARN-482) FS: Extend SchedulingMode to intermediate queues

    [ https://issues.apache.org/jira/browse/YARN-482?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13632054#comment-13632054 ] 

Sandy Ryza commented on YARN-482:
---------------------------------

Karthik, the patch is looking good.  A few comments:

* throwPolicyError would be clearer as something like throwPolicyDoesNotApplyError
* Does FSQueue#assignReservedContainer get called anywhere? It seems to me that this method doesn't need to belong to the Schedulable interface and should only be in AppSchedulable.
* In FSSchedulerNode, I think renaming appSchedulable to reservedAppSchedulable would be clearer.  It would also be possible to avoid storing the reserved AppSchedulable in the node and instead get the reserved AppSchedulable using scheduler.applications.get(reservedContainer.getApplicationAttemptId()).getAppSchedulable(). But if you've considered this and think it's too circuitous, I think that's reasonable.
* Unnecessary whitespace change in FairScheduler under maxAssign's declaration.
* FairSharePolicy looks better to me than FairsharePolicy.
                
> FS: Extend SchedulingMode to intermediate queues
> ------------------------------------------------
>
>                 Key: YARN-482
>                 URL: https://issues.apache.org/jira/browse/YARN-482
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: scheduler
>    Affects Versions: 2.0.3-alpha
>            Reporter: Karthik Kambatla
>            Assignee: Karthik Kambatla
>         Attachments: yarn-482.patch, yarn-482.patch
>
>
> FS allows setting {{SchedulingMode}} for leaf queues. Extending this to non-leaf queues allows using different kinds of fairness: e.g., root can have three child queues - fair-mem, drf-cpu-mem, drf-cpu-disk-mem taking different number of resources into account. In turn, this allows users to decide on the scheduling latency vs sophistication of the scheduling mode.

--
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