You are viewing a plain text version of this content. The canonical link for it is here.
Posted to mapreduce-issues@hadoop.apache.org by "Karthik Kambatla (JIRA)" <ji...@apache.org> on 2014/04/11 23:52:17 UTC
[jira] [Updated] (MAPREDUCE-5822) FairScheduler preemption due to
fairshare-starvation not work when fairShare == 1
[ https://issues.apache.org/jira/browse/MAPREDUCE-5822?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Karthik Kambatla updated MAPREDUCE-5822:
----------------------------------------
Summary: FairScheduler preemption due to fairshare-starvation not work when fairShare == 1 (was: FairScheduler isStartvedForFairShare does not work when fairShare == 1)
> FairScheduler preemption due to fairshare-starvation not work when fairShare == 1
> ----------------------------------------------------------------------------------
>
> Key: MAPREDUCE-5822
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-5822
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: scheduler
> Affects Versions: 1.2.1
> Reporter: Anubhav Dhoot
> Assignee: Anubhav Dhoot
> Attachments: MAPREDUCE-5822.001.patch, MAPREDUCE-5822.002.patch
>
>
> If the fair share returned by the scheduler getFairShare() == 1 the pool will never be marked as being starved because of the following calculation:
> {code}
> boolean isStarvedForFairShare(PoolSchedulable sched) {
> int desiredFairShare = (int) Math.floor(Math.min( sched.getFairShare() / 2, sched.getDemand()));
> return (sched.getRunningTasks() < desiredFairShare);
> }
> {code}
> getFairShare() returns 1
> Math.min calculation will return 0.5
> Math.Floor() which will cause the desiredFairShare to be set to 0.
> the return value to be 'false' (0 < 0)
> If you have a small job without a minimum set it will not get scheduled if a large job is hogging the slots.
--
This message was sent by Atlassian JIRA
(v6.2#6252)