[
https://issues.apache.org/jira/browse/MAPREDUCE-7126?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16584113#comment-16584113
]
Jason Lowe commented on MAPREDUCE-7126:
---------------------------------------
There needs to be more information on this report to diagnose if there's a real
issue here. You will need to check the RM logs to see why it decided to
respond to the AM as being unrecognized. One theory is the AM could have been
running on a node where the nodemanager crashed. If that occurs then the RM
will eventually expire the nodemanager due to lack of heartbeats and consider
all of the containers on that node lost. When the AM proceeds to heartbeat to
the RM I would expect the RM to reply that the AM is no longer recognized since
the RM considers that app attempt dead (being on a lost node). Having the AM
shut down (without unregistering!) is appropriate in that case.
> Error communicating with RM: Resource Manager doesn't recognize AttemptId:
> appattempt_idxx
> ------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-7126
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-7126
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: mr-am
> Affects Versions: 2.6.0
> Reporter: Avdhesh kumar
> Priority: Major
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
