hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Allen Wittenauer (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-3692) Allow REST API to set a user generated message when killing an application
Date Sun, 04 Sep 2016 21:42:20 GMT

    [ https://issues.apache.org/jira/browse/YARN-3692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15463553#comment-15463553
] 

Allen Wittenauer commented on YARN-3692:
----------------------------------------

bq. I think, we need to run mvn install in all projects..? Allen Wittenauer am I right..?

I'm not really sure what you are suggesting, but "all" is almost always not the best approach.
precommit is configured to only do the minimal amount of work.  If it needs to do more work
in other modules, then it needs to be told do that and only that.  The typical way is upload
a fake patch that also touches those modules.  Just be aware that it will run through the
entire process for all modules touched.

Some other thoughts:

* the YARN tests may not be providing enough adequate API coverage if the unit tests aren't
failing but MR is
* if MR can't run without changes, then the patch in it's current form is likely an incompatible
change not suited for branch-2


> Allow REST API to set a user generated message when killing an application
> --------------------------------------------------------------------------
>
>                 Key: YARN-3692
>                 URL: https://issues.apache.org/jira/browse/YARN-3692
>             Project: Hadoop YARN
>          Issue Type: Improvement
>            Reporter: Rajat Jain
>            Assignee: Rohith Sharma K S
>         Attachments: 0001-YARN-3692.patch, 0002-YARN-3692.patch
>
>
> Currently YARN's REST API supports killing an application without setting a diagnostic
message. It would be good to provide that support.
> *Use Case*
> Usually this helps in workflow management in a multi-tenant environment when the workflow
scheduler (or the hadoop admin) wants to kill a job - and let the user know the reason why
the job was killed. Killing the job by setting a diagnostic message is a very good solution
for that. Ideally, we can set the diagnostic message on all such interface:
> yarn kill -applicationId ... -diagnosticMessage "some message added by admin/workflow"
> REST API { 'state': 'KILLED', 'diagnosticMessage': 'some message added by admin/workflow'}



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org


Mime
View raw message