Junping Du commented on YARN-41:

Thanks [~devaraj.k] for providing truth table above. Contents looks mostly good 
to me. 
One interesting case is when  "yarn.nodemanager.recovery.enabled=true" but 
"yarn.nodemanager.recovery.supervised=false", the shutdown behavior after 
YARN-2331 is: running containers will get killed, but recovery work will 
continue after NM get restarted. Theoretically, I think we should unregister NM 
from RM because we don't expect app/container get recovered on this NM. 

bq. As per my understanding I assumed here that NM is under supervision enabled 
only when the NM recovery is enabled. 
Agree. In practical, if NM is under supervision while NM recovery is disabled, 
the behavior should be exactly the same as both config are set to false.

[~jlowe], any comments here?

> The RM should handle the graceful shutdown of the NM.
> -----------------------------------------------------
>                 Key: YARN-41
>                 URL: https://issues.apache.org/jira/browse/YARN-41
>             Project: Hadoop YARN
>          Issue Type: New Feature
>          Components: nodemanager, resourcemanager
>            Reporter: Ravi Teja Ch N V
>            Assignee: Devaraj K
>              Labels: BB2015-05-TBR
>         Attachments: MAPREDUCE-3494.1.patch, MAPREDUCE-3494.2.patch, 
> MAPREDUCE-3494.patch, YARN-41-1.patch, YARN-41-2.patch, YARN-41-3.patch, 
> YARN-41-4.patch, YARN-41.patch
> Instead of waiting for the NM expiry, RM should remove and handle the NM, 
> which is shutdown gracefully.

This message was sent by Atlassian JIRA

Reply via email to