You are viewing a plain text version of this content. The canonical link for it is here.
Posted to yarn-issues@hadoop.apache.org by "ASF GitHub Bot (Jira)" <ji...@apache.org> on 2023/01/19 10:46:00 UTC

[jira] [Commented] (YARN-11412) Create a Concurrent Users Manager to enable thread safe concurrent resource usage tracking of user

    [ https://issues.apache.org/jira/browse/YARN-11412?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17678612#comment-17678612 ] 

ASF GitHub Bot commented on YARN-11412:
---------------------------------------

krishan1390 opened a new pull request, #5318:
URL: https://github.com/apache/hadoop/pull/5318

   JIRA: [YARN-11412](https://issues.apache.org/jira/browse/YARN-11412).
   
   Create a Concurrent Users Manager to enable thread safe concurrent resource usage tracking of user
   
   Low level design doc - https://docs.google.com/document/d/1czUh2XU3_X_eRIJAsSM40hRuHYSHw4ymyfky39dIj4s/edit




> Create a Concurrent Users Manager to enable thread safe concurrent resource usage tracking of user
> --------------------------------------------------------------------------------------------------
>
>                 Key: YARN-11412
>                 URL: https://issues.apache.org/jira/browse/YARN-11412
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>            Reporter: Krishan Goyal
>            Assignee: Krishan Goyal
>            Priority: Major
>
> We will build ConcurrentUsersManager whose objective is similar to UsersManager but whose APIs provide better concurrent performance with different class invariants
>  
> Both ConcurrentUsersManager & UsersManager will extend a new class AbstractCSUsersManager which implements AbstractUsersManager
>  
> The primary behavioural change of ConcurrentUsersManager is that it will provide user resource limits which are eventually consistent wrt containers (allocation or release) / users (active or inactive) - These limits affect scheduling and preemption but at massive scale small variances in individual decisions for a container are fine as long as the scheduler progresses in an acceptable way (i.e - users are unconcerned about these variances)
>  
> More details of low level details - [https://docs.google.com/document/d/1czUh2XU3_X_eRIJAsSM40hRuHYSHw4ymyfky39dIj4s/edit] 



--
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