[
https://issues.apache.org/jira/browse/MAPREDUCE-6776?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15537275#comment-15537275
]
Hitesh Shah commented on MAPREDUCE-6776:
----------------------------------------
>From a practical sense, this is not really an incompatible change as there is
>some internal behavioral aspects that are being changed to retry 3 times
>instead of no retries.
However, from a pure theoretical compat perspective, a public default value is
being changed as well as the value in mapred-default.xml. Tests which might be
earlier doing some verification would expect immediate failures whereas now it
might be reconnect or fail after 6 seconds or so.
I suggest pushing this to trunk for sure as we are still in the alpha stage of
releases. As for branch-2, I would check with the 2.8 release manager.
> yarn.app.mapreduce.client.job.max-retries should have a more useful default
> ---------------------------------------------------------------------------
>
> Key: MAPREDUCE-6776
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-6776
> Project: Hadoop Map/Reduce
> Issue Type: Improvement
> Components: client
> Affects Versions: 2.8.0
> Reporter: Daniel Templeton
> Assignee: Miklos Szegedi
> Attachments: MAPREDUCE-6776.001.patch, MAPREDUCE-6776.002.patch,
> MAPREDUCE-6776.003.patch
>
>
> The default is 0, so any communication failure results in a client failure.
> Oozie doesn't like that. If the RM is failing over and Oozie gets a
> communication failure, it assumes the target job has failed. I propose
> raising the default to something modest like 3 or 5. The default retry
> interval is 2s.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
