[
https://issues.apache.org/jira/browse/YARN-1389?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13923280#comment-13923280
]
Mayank Bansal commented on YARN-1389:
-------------------------------------
bq .I realize there will be an issue that may not have immediate solution.
Currently, if an application is finished, we can get all the finished
containers of it from the history store. However,if an application is still
running, YarnScheduler is going to remove the container out of its cache once
the container is done. Therefore, we're unable to get the finished containers
of a running application.
It seems that we need to cache RMContainer until the application is finished.
Thoughts?
yes we should have this as right now there is inconsistency in finished
containers for running apps, I will create another JIRA to track that.
Thanks,
Mayank
> ApplicationClientProtocol and ApplicationHistoryProtocol should expose
> analogous APIs
> -------------------------------------------------------------------------------------
>
> Key: YARN-1389
> URL: https://issues.apache.org/jira/browse/YARN-1389
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Mayank Bansal
> Assignee: Mayank Bansal
> Attachments: YARN-1389-1.patch, YARN-1389-2.patch, YARN-1389-3.patch,
> YARN-1389-4.patch, YARN-1389-5.patch
>
>
> As we plan to have the APIs in ApplicationHistoryProtocol to expose the
> reports of *finished* application attempts and containers, we should do the
> same for ApplicationClientProtocol, which will return the reports of
> *running* attempts and containers.
> Later on, we can improve YarnClient to direct the query of running instance
> to ApplicationClientProtocol, while that of finished instance to
> ApplicationHistoryProtocol, making it transparent to the users.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
