[ 
https://issues.apache.org/jira/browse/YARN-4090?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15856692#comment-15856692
 ] 

Yufei Gu commented on YARN-4090:
--------------------------------

Hi [~zsl2007], thanks for woking on this. I could misunderstand your patch. To 
make sure we are in the same page, let me clarify this: we are trying to remove 
the recursive computation of {{getResourceUsage()}} to improve the performance. 
Right? Base on that, here are some thoughts:
1. FSParentQueue don’t do the recursive {{getResourceUsage()}} in your patch 
and resource usage updates while allocating, recovering and moving. Does 
FSParentQueue get updated on resource usage when an application finishes or its 
tasks finish? Resource usage is a critical metrics of queues, it would be nice 
to including some unit tests (or modification of existing tests) to make sure 
we do it right. 
2. Any reason we don't refactor the {{getResourceUsage()}} in FSLeafQueue?
Some minor nits:
# Use {{fsQueue}} instead of {{queue}} so you don’t need to cast. 
# Need one empty line after function move() 


> Make Collections.sort() more efficient in FSParentQueue.java
> ------------------------------------------------------------
>
>                 Key: YARN-4090
>                 URL: https://issues.apache.org/jira/browse/YARN-4090
>             Project: Hadoop YARN
>          Issue Type: Improvement
>          Components: fairscheduler
>            Reporter: Xianyin Xin
>            Assignee: zhangshilong
>         Attachments: sampling1.jpg, sampling2.jpg, YARN-4090.001.patch, 
> YARN-4090.002.patch, YARN-4090.003.patch, YARN-4090.004.patch, 
> YARN-4090.005.patch, YARN-4090.006.patch, YARN-4090-preview.patch, 
> YARN-4090-TestResult.pdf
>
>
> Collections.sort() consumes too much time in a scheduling round.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

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

Reply via email to