[
https://issues.apache.org/jira/browse/MAPREDUCE-5542?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14168952#comment-14168952
]
Rohith commented on MAPREDUCE-5542:
-----------------------------------
Apoligies [~jlowe] [~vinodkv] [~jira.shegalov] for not looking back into jira
after long time.
I agree, before kill job need to get some state from AM/RM/HistoryServer. I did
some more investigations on this issue
* Job Status can be null on below cases.
** MRAppMaster is fail to write history and JobClient is redirected to
historyserver to get job status.
** JobClient has different done/done-intermediate directory location from
historyserver. Jobcilent is redirected to historyserver to get job status.
And job status is null means it is getting null status from HistoryServer only.
If null is set from MRAppMaster then JobClient is accessing unauthorised
MRAppMaster. Better getting status from RM when it hit null that ensures job is
finished with succeeded/killed/failed state.
> Killing a job just as it finishes can generate an NPE in client
> ---------------------------------------------------------------
>
> Key: MAPREDUCE-5542
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-5542
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: client, mrv2
> Affects Versions: 2.1.0-beta, 0.23.9
> Reporter: Jason Lowe
> Assignee: Rohith
> Attachments: MAPREDUCE-5542.1.patch, MAPREDUCE-5542.2.patch
>
>
> If a client tries to kill a job just as the job is finishing then the client
> can crash with an NPE.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
