[
https://issues.apache.org/jira/browse/YARN-3839?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15967899#comment-15967899
]
Manikandan R commented on YARN-3839:
------------------------------------
[~jianhe], [~jlowe], [~kasha]
I am trying to understand the changes required for this jira with the help of
YARN-3842 discussion/comments and to see if I can able to contribute to this
jira.
My understanding is, ensuring that NMNotReadyException is not being used in any
places and making sure other exceptions (for ex, invalid token exception) is
being thrown in case of situations like NM restarts etc and can be validated by
running corresponding test cases. Since NMNotReadyException is not useful,
there is no use in having blockNewContainerRequests atomic boolean variable as
well because start/stop/increase containers methods depends on
blockNewContainerRequests value to throw NMNotReadyException or not. Hence,
corresponding setter and getter methods also can be removed. In addition,
corresponding test cases also needs to be cleaned up.
Currently, failed containers adds up in case of InvalidToken and YARN
exceptions (in startContainers() method) and retries doesn't happen as those
are system errors, which should be retained as it is.
Can you please validate this and provide suggestions?
> Quit throwing NMNotYetReadyException
> ------------------------------------
>
> Key: YARN-3839
> URL: https://issues.apache.org/jira/browse/YARN-3839
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
> Reporter: Karthik Kambatla
>
> Quit throwing NMNotYetReadyException when NM has not yet registered with the
> RM.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org
