MENG DING commented on YARN-4138:

Hi, [~sunilg]

The case you mentioned is covered. When RM rolls back container size to 2G, it 
does the equivalent of decreasing the container size to 2G, and will 
subsequently send a decrease message to NM through heartbeat response. NM 
(currently using 4G per your example), upon receiving the response, will 
decrease its enforcement to 2G. So eventually the resource view is consistent 
between RM and NM.

> Roll back container resource allocation after resource increase token expires
> -----------------------------------------------------------------------------
>                 Key: YARN-4138
>                 URL: https://issues.apache.org/jira/browse/YARN-4138
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: api, nodemanager, resourcemanager
>            Reporter: MENG DING
>            Assignee: MENG DING
>         Attachments: YARN-4138-YARN-1197.1.patch
> In YARN-1651, after container resource increase token expires, the running 
> container is killed.
> This ticket will change the behavior such that when a container resource 
> increase token expires, the resource allocation of the container will be 
> reverted back to the value before the increase.

This message was sent by Atlassian JIRA

Reply via email to