hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ashwin Shankar (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-3111) Fix ratio problem on FairScheduler page
Date Tue, 17 Mar 2015 18:29:38 GMT

    [ https://issues.apache.org/jira/browse/YARN-3111?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14365722#comment-14365722

Ashwin Shankar commented on YARN-3111:

bq. What do you guys think of overlaying CPU and memory usage, the way steady and instantaneous
fairshares are laid out today?

I agree with Peng, it is going to become pretty cluttered displaying shares/max of each of
the resources on the same bar.
Also, if we go by this approach, it would be ambiguous as to which resource is the usage bar
representing, since the usage bar shows usage of dominant resource ie max(memoryRatio, vCoresRatio).
Usage bar turns orange when its above fairshare, if we represent all the resources in one
bar, how do we
know if we are above fair share due to memory or cpu or disk ?

Here is my proposal :
For each queue bar :
1. represent steady/instant/max of only the dominant resource in the bar.
2. usage, like in the patch, will be again usage of dominant resource.
3. In the tooltip, we mention what is the dominant resource we are representing in that queue([memory,cpu]).
Note that dominant resource displayed can be memory in one queue and something else in another.
4. We already display steady/instant memory percentage in the tool tip, we could add cpu %
of steady/instant/max in there as well, so that we know details
of each of the resource.

Thoughts ?

> Fix ratio problem on FairScheduler page
> ---------------------------------------
>                 Key: YARN-3111
>                 URL: https://issues.apache.org/jira/browse/YARN-3111
>             Project: Hadoop YARN
>          Issue Type: Bug
>          Components: fairscheduler
>    Affects Versions: 2.6.0
>            Reporter: Peng Zhang
>            Assignee: Peng Zhang
>            Priority: Minor
>         Attachments: YARN-3111.1.patch, YARN-3111.png
> Found 3 problems on FairScheduler page:
> 1. Only compute memory for ratio even when queue schedulingPolicy is DRF.
> 2. When min resources is configured larger than real resources, the steady fair share
ratio is so long that it is out the page.
> 3. When cluster resources is 0(no nodemanager start), ratio is displayed as "NaN% used"
> Attached image shows the snapshot of above problems. 

This message was sent by Atlassian JIRA

View raw message