[ 
https://issues.apache.org/jira/browse/YARN-3405?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14385124#comment-14385124
 ] 

Peng Zhang commented on YARN-3405:
----------------------------------

[~kasha]
What the real problem I met in our cluster is livelock. After checking code, I 
think it has some other bad cases like we talked above.
I list them together because I think they has the same basic problem: 
"calculation of preemption request and preemption of container are separated as 
two phases", lot of necessary info is lost between these two phases.

For less confusion, I created YARN-3414 to discuss livelock problem.
And I'll update this description to show non-livelock case that I think.


> FairScheduler's preemption cannot happen between sibling in some case
> ---------------------------------------------------------------------
>
>                 Key: YARN-3405
>                 URL: https://issues.apache.org/jira/browse/YARN-3405
>             Project: Hadoop YARN
>          Issue Type: Bug
>          Components: fairscheduler
>    Affects Versions: 2.7.0
>            Reporter: Peng Zhang
>            Priority: Critical
>
> Queue hierarchy described as below:
> {noformat}
>              root
>               |         
>            queue-1        
>           /      \
> queue-1-1        queue-1-2
> {noformat}
> 1. When queue-1-1 is active and it has been assigned with all resources.
> 2. When queue-1-2 is active, and it cause some new preemption request.
> 3. But when do preemption, it now starts from root, and found queue-1 is not 
> over fairshare, so no recursion preemption to queue-1-1.
> 4. Finally queue-1-2 will be waiting for resource release form queue-1-1 
> itself.



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

Reply via email to