[ 
https://issues.apache.org/jira/browse/YARN-365?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13573592#comment-13573592
 ] 

Thomas Graves commented on YARN-365:
------------------------------------

Sorry Sid. I missed your comments and a few important points yesterday upon 
quick review.  

By aggregating I meant the information in the heartbeat aggregated with all 
previous heartbeats for that single node and then handled all at once in a 
single pass by the scheduler before it tries to do any allocations.  Really its 
the same as your comment (which I missed yesterday) "scheduler should really be 
pulling everything available in the node being processed.

I was originally thinking something along the lines of it having a single list 
for each completed and launched containers that it would just add to rather 
then having the queue of the individual completed and launched lists (one per 
heartbeat). But as long as the scheduler handles all the updates in the queue 
before it tries to schedule you get the same affect.  I'll review the current 
patch in more detail.

A few comments on the current patch:
- we don't need to add an update to the queue if there were no changes
- I don't think the current patch is handling all the updates in a single 
scheduler pass.


                
> 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, Prototype3.txt, YARN-365.1.patch, 
> YARN-365.2.patch, YARN-365.3.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

Reply via email to