[ 
https://issues.apache.org/jira/browse/YARN-4225?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15038195#comment-15038195
 ] 

Jason Lowe commented on YARN-4225:
----------------------------------

With deployments of multiple clusters it's easy to get into situations where 
newer clients end up talking to older clusters.  Clusters are rarely upgraded 
at the same time, and remote access of one cluster for a job's input is not 
rare at least in our setups.  That's a perfect example of a newer client 
talking to an older server.  It may not be officially supported, but it's not 
going to be a rare occurrence (for HDFS, at least).

For this specific case I think the need for the feature would be rare, hence my 
hedging about whether we really need it.  It's not that the new client 
completely breaks talking to the older server, it just would be capable of 
returning misleading information about the preemption status.  I'm OK if we 
decide this scenario isn't worth supporting.  (It's not hard to do so, just 
tedious and a bit messy with the API.)  But in general there will be cases 
where breaking compatibility between a newer client and an older server is 
going to be problematic even if it isn't officially supported because of the 
multiple cluster scenarios.

> Add preemption status to yarn queue -status for capacity scheduler
> ------------------------------------------------------------------
>
>                 Key: YARN-4225
>                 URL: https://issues.apache.org/jira/browse/YARN-4225
>             Project: Hadoop YARN
>          Issue Type: Bug
>          Components: capacity scheduler, yarn
>    Affects Versions: 2.7.1
>            Reporter: Eric Payne
>            Assignee: Eric Payne
>            Priority: Minor
>         Attachments: YARN-4225.001.patch, YARN-4225.002.patch, 
> YARN-4225.003.patch
>
>




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

Reply via email to