[
https://issues.apache.org/jira/browse/YARN-6384?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15943545#comment-15943545
]
dengkai commented on YARN-6384:
-------------------------------
I think it's different from
[YARN-5936|https://issues.apache.org/jira/browse/YARN-5936]. YARN-5936 talked
about the fairness of cpu-scheduling between containers. In my cluster, this
works well because of the same cpu.shares. The same cpu.shares assures the
lower limit of cpu-usage between containers. What I saying is the upper limit
of cpu-usage for a container. It's unreasonable a container with few vcores
take too much cpu from cluster than promised but using
yarn.nodemanager.resource.percentage-physical-cpu-limit is inflexible.
> Add configuratin to set max cpu usage when strict-resource-usage is false
> with cgroups
> --------------------------------------------------------------------------------------
>
> Key: YARN-6384
> URL: https://issues.apache.org/jira/browse/YARN-6384
> Project: Hadoop YARN
> Issue Type: Improvement
> Reporter: dengkai
>
> When using cgroups on yarn, if
> yarn.nodemanager.linux-container-executor.cgroups.strict-resource-usage is
> false, user may get very more cpu time than expected based on the vcores.
> There should be a upper limit even resource-usage is not strict, just like a
> percentage which user can get more than promised by vcores. I think it's
> important in a shared cluster.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
