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 "Sunil G (JIRA)" <ji...@apache.org> on 2015/07/21 17:22:04 UTC

[jira] [Updated] (YARN-3941) Proportional Preemption policy should try to avoid sending duplicate PREEMPT_CONTAINER event to scheduler

     [ https://issues.apache.org/jira/browse/YARN-3941?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Sunil G updated YARN-3941:
--------------------------
    Attachment: 0001-YARN-3941.patch

Hi [~leftnoteasy]
Uploading an initial version of patch.

*testExpireKill* already handles the verification of this change. Multiple PREEMPT_CONTAINER event will not be raised for same container for each interval check of ProportionalCPP.

> Proportional Preemption policy should try to avoid sending duplicate PREEMPT_CONTAINER event to scheduler
> ---------------------------------------------------------------------------------------------------------
>
>                 Key: YARN-3941
>                 URL: https://issues.apache.org/jira/browse/YARN-3941
>             Project: Hadoop YARN
>          Issue Type: Bug
>          Components: capacityscheduler
>    Affects Versions: 2.7.1
>            Reporter: Sunil G
>            Assignee: Sunil G
>         Attachments: 0001-YARN-3941.patch
>
>
> Currently ProportionalCPP tries to send multiple PREEMPT_CONTAINER events to scheduler during every cycle of preemption check till the container is either forcefully killed or preempted by AM. 
> This can be throttled from ProportionalPreemptionPolicy to avoid excess of events to scheduler.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)