[
https://issues.apache.org/jira/browse/MAPREDUCE-4448?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13416268#comment-13416268
]
Daryn Sharp commented on MAPREDUCE-4448:
----------------------------------------
bq. Those subsystems are not (and should not be) aware of whether log
aggregation initialized successfully
Fully agreed! My suggestion was regarding {{LogAggregationService#handle}},
not an external subsystem. I guess there are race conditions where the
aggregator might disappear. Since a failed init is only one case where this
might occur, and it should be rare, I guess a warn log message is ok.
Final suggestion is for the aggregator thread not to remove itself from the
map. Then the existence of the map key + {{thread#isRunning}} would remove the
ambiguity of "did it ever start" and "did it fail unexpectedly".
> Nodemanager crashes upon application cleanup if aggregation failed to start
> ---------------------------------------------------------------------------
>
> Key: MAPREDUCE-4448
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-4448
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: mrv2, nodemanager
> Affects Versions: 0.23.3, 2.0.1-alpha
> Reporter: Jason Lowe
> Assignee: Jason Lowe
> Priority: Critical
> Attachments: MAPREDUCE-4448.patch
>
>
> When log aggregation is enabled, the nodemanager can crash if log aggregation
> for an application failed to start.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
