[ 
https://issues.apache.org/jira/browse/YARN-5216?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15768540#comment-15768540
 ] 

Arun Suresh commented on YARN-5216:
-----------------------------------

bq. I'm not so sure about adding anything into the Container interface as 
pause/resume is only for opportunistic containers.
I think that discussion is orthogonal. I would like to keep the code consistent 
to kill / launch etc. and I feel exposing 
"dispatcher.getEventHandler().handle(..)" is a bad idea, since:
# it limits testability and 
# This requires the caller (CapacityScheduler in this case) to know what type 
event and which dispatcher to use, both of which become tightly coupled to the 
caller.. I understand there exists code, but I would like to keep new code as 
clean as possible.
Given the above the "need to enforce only opportunistic containers can be 
paused" argument seems too week to not perform refactor.

+1 to the patch, pending the above and one minor nit:
The "use-pause-for-preemption" is an NM level config, so you should not use 
RM_PREFIX


> Expose configurable preemption policy for OPPORTUNISTIC containers running on 
> the NM
> ------------------------------------------------------------------------------------
>
>                 Key: YARN-5216
>                 URL: https://issues.apache.org/jira/browse/YARN-5216
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: distributed-scheduling
>            Reporter: Arun Suresh
>            Assignee: Hitesh Sharma
>              Labels: oct16-hard
>         Attachments: YARN-5216-YARN-5972.001.patch, 
> YARN-5216-YARN-5972.002.patch, YARN-5216-YARN-5972.003.patch, 
> YARN-5216-YARN-5972.004.patch, YARN-5216-YARN-5972.005.patch, 
> YARN5216.001.patch, yarn5216.002.patch
>
>
> Currently, the default action taken by the QueuingContainerManager, 
> introduced in YARN-2883, when a GUARANTEED Container is scheduled on an NM 
> with OPPORTUNISTIC containers using up resources, is to KILL the running 
> OPPORTUNISTIC containers.
> This JIRA proposes to expose a configurable hook to allow the NM to take a 
> different action.



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

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to