MENG DING commented on YARN-1651:

I think that should work. 
We probably need to make sure to properly log warning messages so that user 
will find out what is going on and why under these circumstances. I know that 
both of my examples are edge cases (mostly through incorrect logic in AM), but 
it could be frustrating to the user when they think their container has been 
granted certain resource in NM, only to find out at a later time that it is not.

> CapacityScheduler side changes to support increase/decrease container 
> resource.
> -------------------------------------------------------------------------------
>                 Key: YARN-1651
>                 URL: https://issues.apache.org/jira/browse/YARN-1651
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: resourcemanager, scheduler
>            Reporter: Wangda Tan
>            Assignee: Wangda Tan
>         Attachments: YARN-1651-1.YARN-1197.patch

This message was sent by Atlassian JIRA

Reply via email to