[
https://issues.apache.org/jira/browse/YARN-4747?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15248512#comment-15248512
]
Varun Saxena commented on YARN-4747:
------------------------------------
[~ebadger], if this has to go in 2.7.3 then sure please go ahead and take it
up. Because I am currently travelling and wont have much bandwidth till next
week.
And it seems for 2.7.3, we will have a RC out by then.
I had however done some fix already when I took it up i.e. I reported host and
port info in container finished event from RM.
Fix in ApplicationHistoryManagerOnTimelineStore was also done but this fix is
not necessary. It was added just to make AHS more robust. I hadn't added test
cases for fix in ApplicationHistoryManagerOnTimelineStore though. I was also
checking if we report assigned node as null in container report, what would be
the impact.
It seems in CLI related code, we do not expect it to be null.
However as I said if host and port is reported from SMP, host and port wont be
null in ApplicationHistoryManagerOnTimelineStore and hence assigned node wont
be null.
What I will do is that I can rebase and update the patch which I have, with the
part already done by me.
And you can take it from there.
If you can wait till start of next week though, I can handle it then. I am fine
with you taking it up as well.
> AHS error 500 due to NPE when container start event is missing
> --------------------------------------------------------------
>
> Key: YARN-4747
> URL: https://issues.apache.org/jira/browse/YARN-4747
> Project: Hadoop YARN
> Issue Type: Bug
> Components: timelineserver
> Affects Versions: 2.7.2
> Reporter: Jason Lowe
> Assignee: Varun Saxena
> Attachments: YARN-4747.01.patch
>
>
> Saw an error 500 due to a NullPointerException caused by a missing host for
> an AM container. Stacktrace to follow.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
