[
https://issues.apache.org/jira/browse/MAPREDUCE-2749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13096628#comment-13096628
]
Vinod Kumar Vavilapalli commented on MAPREDUCE-2749:
----------------------------------------------------
Thanks for the patch Thomas!
One minor comment I have with this patch is the setting of keys via the event.
While this works, my concern is that if we do it this way, by the time
{{NodeManager.start()}} returns, the server isn't still guaranteed to be ready
for service. I think we can do this another way. If we move
{{ContainerTokenSecretManager}} to the top level {{NodeManager}} composite, we
can set the keys as the last step in {{NodeManager.start()}}. Thoughts?
Also, we can add a test for this - mock {{rpc.getServer()}} to throw an
exception and verify that ResourceManager doesn't have any registered NM.
> [MR-279] NM registers with RM even before it starts various servers
> -------------------------------------------------------------------
>
> Key: MAPREDUCE-2749
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-2749
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: mrv2
> Affects Versions: 0.23.0
> Reporter: Vinod Kumar Vavilapalli
> Assignee: Thomas Graves
> Fix For: 0.23.0
>
> Attachments: MAPREDUCE-2749-v2.patch, MAPREDUCE-2749.patch
>
>
> In case NM eventually fails to start the ContainerManager server because of
> say a port clash, RM will have to wait for expiry to detect the NM crash.
> It is desirable to make NM register with RM only after it can start all of
> its components successfully.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
