[jira] [Commented] (YARN-5614) NMClient APIs do not need to take NodeId

["Jian He (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/YARN-5614?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15457793#comment-15457793 ] 

Jian He commented on YARN-5614:
-------------------------------

Actually, this may not be that straightfoward.. if the cache in NMClient is somehow cleaned, e.g. due to restart etc. then it lost the NodeId.
Should the ContainerId contain NodeId?

> NMClient APIs do not need to take NodeId
> ----------------------------------------
>
>                 Key: YARN-5614
>                 URL: https://issues.apache.org/jira/browse/YARN-5614
>             Project: Hadoop YARN
>          Issue Type: Improvement
>            Reporter: Jian He
>
> Currently, all the NMClient APIs, e.g. startContainer, stopContainer, getContainerStatus force caller to pass in the NodeId as wells as the ContainerId.  Hence, caller has to maintain both NodeId and ContainerId mapping on its own, which becomes a bit inconvenient sometimes.
> In fact, NMClient internally already cached the NodeId for the ContainerId, so the NodeId in the API could be avoided.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org