hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sunil G (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (YARN-7332) Compute effectiveCapacity per each resource vector
Date Thu, 26 Oct 2017 08:54:00 GMT

     [ https://issues.apache.org/jira/browse/YARN-7332?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Sunil G updated YARN-7332:
--------------------------
    Attachment: YARN-7332.YARN-5881.003.patch

Thanks [~leftnoteasy]. Yes, makes sense.

And that was an interesting value too.

Actual resource: 10, ratio: 1.25, vcores value: 12.5 ~ 12, 
Actual resource: 22, ratio: 1.25, vcores value: 27.5 ~ 27 

Once we increases vcores to 40, it has to be splitted, with calculation we are getting 12+27.
If we use ceil, it ll be 13+28. I chose floor for now. Even if we loose 1 core, it ll be a
delta for parent. Thoughts?

> Compute effectiveCapacity per each resource vector
> --------------------------------------------------
>
>                 Key: YARN-7332
>                 URL: https://issues.apache.org/jira/browse/YARN-7332
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: capacity scheduler
>    Affects Versions: YARN-5881
>            Reporter: Sunil G
>            Assignee: Sunil G
>         Attachments: YARN-7332.YARN-5881.001.patch, YARN-7332.YARN-5881.002.patch, YARN-7332.YARN-5881.003.patch
>
>
> Currently effective capacity uses a generalized approach based on dominance. Hence some
vectors may not be calculated correctly. 



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

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


Mime
View raw message