[
https://issues.apache.org/jira/browse/YARN-365?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13572249#comment-13572249
]
Siddharth Seth commented on YARN-365:
-------------------------------------
Xuan, I took a look at the patch. Some comments.
The scheduler should really be pulling everything available in the node being
processed. Pulling only a single element doesn't change things too much from
what they are at the moment. The other schedulers will also need to be updated
- since the heartbeat path is common for all of them, i.e. the FifoScheduler
and FairScheduler. Also, some thought needs to be given to handling of cases
where the node may have gone unhealthy etc.
Digging into the patch,
- Don't think RMNode should expose it's internal data structure via
{{getNodeUpdateQueue}}. Instead, it should expose a method give back a List of
ContainerUpdates.
- Do we need an explicit setNextHeartBeat? Instead, the call to get container
updates could be used for now.
- NodeUpdateSchedulerEvent should be changed to remove the container
information, instead of sending nulls.
- Similarly for nodeUpdate in the CapacityScheduler
- Rename UpdateContainerInfo to UpdatedContainerInfo
The code does have some formatting issues - please take a look at
http://wiki.apache.org/hadoop/HowToContribute for code formatting guidelines
and other useful info.
Also, could you please upload another doc with the latest approach, to stay in
sync with the patch.
Thanks!
> Each NM heartbeat should not generate and event for the Scheduler
> -----------------------------------------------------------------
>
> Key: YARN-365
> URL: https://issues.apache.org/jira/browse/YARN-365
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: resourcemanager, scheduler
> Affects Versions: 0.23.5
> Reporter: Siddharth Seth
> Assignee: Xuan Gong
> Attachments: Prototype2.txt, YARN-365.1.patch, YARN-365.2.patch
>
>
> Follow up from YARN-275
> https://issues.apache.org/jira/secure/attachment/12567075/Prototype.txt
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
