[
https://issues.apache.org/jira/browse/YARN-3769?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14709856#comment-14709856
]
MENG DING commented on YARN-3769:
---------------------------------
[~leftnoteasy], for better tracking purposes, would it be better to update the
title of this JIRA to something more general, e.g., *CapacityScheduler: Improve
preemption to preempt only those containers that would satisfy the incoming
request* (similar to YARN-2154)? This ticket can then be used to address
preemption ping-pong issue for both new container request and container
resource increase request.
Besides the proposal that you have presented, an alternative solution to
consider is: once we collect the list of preemptable containers, we immediately
have a *dry run* of the scheduling algorithm to match the preemptable resources
against outstanding new/increase resource requests. We then only preempt the
resources that can find a match.
Thoughts?
Meng
> Preemption occurring unnecessarily because preemption doesn't consider user
> limit
> ---------------------------------------------------------------------------------
>
> Key: YARN-3769
> URL: https://issues.apache.org/jira/browse/YARN-3769
> Project: Hadoop YARN
> Issue Type: Bug
> Components: capacityscheduler
> Affects Versions: 2.6.0, 2.7.0, 2.8.0
> Reporter: Eric Payne
> Assignee: Wangda Tan
>
> We are seeing the preemption monitor preempting containers from queue A and
> then seeing the capacity scheduler giving them immediately back to queue A.
> This happens quite often and causes a lot of churn.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
