[
https://issues.apache.org/jira/browse/YARN-10930?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Szilard Nemeth updated YARN-10930:
----------------------------------
Fix Version/s: 3.4.0
> Introduce universal configured capacity vector
> ----------------------------------------------
>
> Key: YARN-10930
> URL: https://issues.apache.org/jira/browse/YARN-10930
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: capacity scheduler
> Reporter: Andras Gyori
> Assignee: Andras Gyori
> Priority: Major
> Labels: pull-request-available
> Fix For: 3.4.0
>
> Attachments: capacity_scheduler_queue_capacity.html
>
> Time Spent: 6h 10m
> Remaining Estimate: 0h
>
> The proposal is to introduce a capacity resource vector that is universally
> parsed for every queue. CapacityResourceVector is a way to unite the current
> capacity modes (weight, percentage, absolute), while maintaining flexibility
> and extendability.
> CapacityResourceVector is a good fit for the existing capacity configs, for
> example:
> * percentage mode: root.example.capacity 50 is a syntactic sugar for
> [memory=50%, vcores=50%, ....]
> * absolute mode: root.example.capacity [memory=1024, vcores=2] is a natural
> fit for the vector, there is no need for additional settings
> CapacityResourceVector will be used in a future refactor, to unify the
> resource calculation and lift the limitation imposed on the queue hierarchy
> capacity settings (eg. can not use both absolute resource and percentage in
> the same hierarchy etc...)
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org
