[
https://issues.apache.org/jira/browse/YARN-4599?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15105694#comment-15105694
]
Allen Wittenauer commented on YARN-4599:
----------------------------------------
bq. We should also explicitly set OOM control so that containers are not killed
as soon as they go over their usage.
We need this as something that can be set and I'd propose that the default be
off given that administrators are going to be very confused when they see
container usage go above the limit.
FWIW, we have jobs running on our clusters that the NM's memory control doesn't
stop from breaking nodes because it is too slow.
> Set OOM control for memory cgroups
> ----------------------------------
>
> Key: YARN-4599
> URL: https://issues.apache.org/jira/browse/YARN-4599
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
> Affects Versions: 2.9.0
> Reporter: Karthik Kambatla
> Assignee: Karthik Kambatla
>
> YARN-1856 adds memory cgroups enforcing support. We should also explicitly
> set OOM control so that containers are not killed as soon as they go over
> their usage. Today, one could set the swappiness to control this, but
> clusters with swap turned off exist.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
