[
https://issues.apache.org/jira/browse/YARN-4511?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16198915#comment-16198915
]
Haibo Chen commented on YARN-4511:
----------------------------------
Thanks a lot for the review, [~asuresh]!
bq. in the context of container updates, generally, a container stays on the
same node - so was surprised that you added a newNode and oldNode arguments to
the method
I was not sure about whether container update is always node-local, hence I
added a newNode and another oldNode argument. Will remove one in the follow up
patch. The change to swapContainer(), IIRC, is necessary to fix some of the
unit tests that the rest of the patch causes. The revelant change in the patch
is that SchedulerNode is now modified to distinguish between Opportunistic and
Guaranteed containers. Since swapContainer only swaps the internal container
objects rather than RMContainers which SchedulerNod gets notified of, we need
to update the resource accounting explicitly on the node every time there is a
container swap. Otherwise, the incorrect resource accounting would cause
existing unit tests to fail. I hope it makes sense to you.
> Common scheduler changes supporting scheduler-specific implementations
> ----------------------------------------------------------------------
>
> Key: YARN-4511
> URL: https://issues.apache.org/jira/browse/YARN-4511
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Wangda Tan
> Assignee: Haibo Chen
> Attachments: YARN-4511-YARN-1011.00.patch,
> YARN-4511-YARN-1011.01.patch, YARN-4511-YARN-1011.02.patch,
> YARN-4511-YARN-1011.03.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org
