[
https://issues.apache.org/jira/browse/YARN-9756?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16909636#comment-16909636
]
Manikandan R commented on YARN-9756:
------------------------------------
[~eepayne] I spent some time on understanding this.
ProportionalCapacityPreemptionPolicy#preemptOrkillSelectedContainerAfterWait
triggers an pre-emption event for each container based on max limit allowed per
round. I think if we can do sum of memory/vcores of all containers going to be
pre-empt for each round and call appropriate metrics methods here. Is this
correct?
Also, since metrics is going to be per round, assuming there would be so many
rounds, Wouldn't be difficult for users to derive value out of it? Do you have
any JMX o/p structure in your mind?
> Create metric that sums total memory/vcores preempted per round
> ---------------------------------------------------------------
>
> Key: YARN-9756
> URL: https://issues.apache.org/jira/browse/YARN-9756
> Project: Hadoop YARN
> Issue Type: Improvement
> Components: capacity scheduler
> Affects Versions: 3.2.0, 2.9.2, 3.0.3, 2.8.5, 3.1.2
> Reporter: Eric Payne
> Assignee: Eric Payne
> Priority: Major
>
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
