[jira] [Commented] (YARN-11643) Skip unnecessary pre-check in Multi Node Placement

["ASF GitHub Bot (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/YARN-11643?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17804587#comment-17804587 ] 

ASF GitHub Bot commented on YARN-11643:
---------------------------------------

laysfire opened a new pull request, #6426:
URL: https://github.com/apache/hadoop/pull/6426

   <!--
     Thanks for sending a pull request!
       1. If this is your first time, please read our contributor guidelines: https://cwiki.apache.org/confluence/display/HADOOP/How+To+Contribute
       2. Make sure your PR title starts with JIRA issue id, e.g., 'HADOOP-17799. Your PR title ...'.
   -->
   
   ### Description of PR
   JIRA: [YARN-11643](https://issues.apache.org/jira/browse/YARN-11643). Skip unnecessary pre-check in Multi Node Placement.
   Scheduler may do useless while loop when Multi Node Placement enable. This lead to poor schedule performance in worst-case scenario.
   
   ### How was this patch tested?
   
   
   ### For code changes:
   
   - [ ] Does the title or this PR starts with the corresponding JIRA issue id (e.g. 'HADOOP-17799. Your PR title ...')?
   - [ ] Object storage: have the integration tests been executed and the endpoint declared according to the connector-specific documentation?
   - [ ] If adding new dependencies to the code, are these dependencies licensed in a way that is compatible for inclusion under [ASF 2.0](http://www.apache.org/legal/resolved.html#category-a)?
   - [ ] If applicable, have you updated the `LICENSE`, `LICENSE-binary`, `NOTICE-binary` files?
   
   




> Skip unnecessary pre-check in Multi Node Placement
> --------------------------------------------------
>
>                 Key: YARN-11643
>                 URL: https://issues.apache.org/jira/browse/YARN-11643
>             Project: Hadoop YARN
>          Issue Type: Improvement
>          Components: capacityscheduler
>            Reporter: Xie YiFan
>            Assignee: Xie YiFan
>            Priority: Minor
>
> When Multi Node Placement enabled, RegularContainerAllocator do a while loop to find one node from candidate set to allocate for a given scheduler key. Before do allocate, pre-check be called to check if current node satisfies check. If this node does not pass all checks, just continue to next node.
> {code:java}
> if (reservedContainer == null) {
>   result = preCheckForNodeCandidateSet(node,
>       schedulingMode, resourceLimits, schedulerKey);
>   if (null != result) {
>     continue;
>   }
> } {code}
> But some checks are related to scheduler Key or Application which return PRIORITY_SKIPPED or APP_SKIPPED. It means that if first node does not pass check, the following nodes also do not pass. 
> If cluster have 5000 nodes in default partition, Scheduler will waste 5000 times loop for just one scheduler key.
>  



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org