[
https://issues.apache.org/jira/browse/YARN-8932?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16669331#comment-16669331
]
Haibo Chen commented on YARN-8932:
----------------------------------
Thanks for the review, [~rkanter]. I have updated the patch to include a few
more comments to explain what the numbers mean after the change.
The cpu utilization numbers involved, before this patch, were percentage
numbers. By multiple them by the total number of vcores, they are changed to
absolute cpu utilization in terms of the number of vcores used.
> ResourceUtilization cpu is misused in oversubscription as a percentage
> ----------------------------------------------------------------------
>
> Key: YARN-8932
> URL: https://issues.apache.org/jira/browse/YARN-8932
> Project: Hadoop YARN
> Issue Type: Sub-task
> Affects Versions: YARN-1011
> Reporter: Haibo Chen
> Assignee: Haibo Chen
> Priority: Major
> Attachments: YARN-8932-YARN-1011.00.patch,
> YARN-8932-YARN-1011.01.patch, YARN-8932-YARN-1011.02.patch
>
>
> The ResourceUtilization javadoc mistakenly documents the cpu as a percentage
> represented by a float number in [0, 1.0f], however it is used as the # of
> vcores used in reality.
> See javadoc and discussion in YARN-8911.
> /**
> * Get <em>CPU</em> utilization.
> *
> * @return <em>CPU utilization</em> normalized to 1 CPU
> */
> @Public
> @Unstable
> public abstract float getCPU();
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
