Wangda Tan commented on YARN-3275:

In addition, parent queue can go beyond max capacity as described in YARN-3243 
no matter if cluster resource changed or not. But child queue can only go 
beyond max capacity when cluster resource reduced.

Eric, could you verify is this any of cases I mentioned above?
bq. No, I don't think going over the absolute max capacity should happen, in 
one case, we saw a small queue in a very busy and large cluster go over by 5%

I feel maybe we need preempt container in your described case 
 capacity and disable-preemption are just two options, we haven't defined 
"disable-preemption" is more important than "max-capacity". IMO, if we should 
do this JIRA or not is still discussable.


> Preemption happening on non-preemptable queues
> ----------------------------------------------
>                 Key: YARN-3275
>                 URL: https://issues.apache.org/jira/browse/YARN-3275
>             Project: Hadoop YARN
>          Issue Type: Bug
>    Affects Versions: 2.7.0
>            Reporter: Eric Payne
>            Assignee: Eric Payne
>         Attachments: YARN-3275.v1.txt
> YARN-2056 introduced the ability to turn preemption on and off at the queue 
> level. In cases where a queue goes over its absolute max capacity (YARN-3243, 
> for example), containers can be preempted from that queue, even though the 
> queue is marked as non-preemptable.
> We are using this feature in large, busy clusters and seeing this behavior.

This message was sent by Atlassian JIRA

Reply via email to