hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Wangda Tan (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-4280) CapacityScheduler reservations may not prevent indefinite postponement on a busy cluster
Date Mon, 02 May 2016 22:23:13 GMT

    [ https://issues.apache.org/jira/browse/YARN-4280?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15267623#comment-15267623
] 

Wangda Tan commented on YARN-4280:
----------------------------------

Thanks [~jlowe], your comments are all make sense to me.

[~kshukla], some thoughts (majorly brain dump) might help your prototype:

- Existing reserved container allocation is go to leaf queue directly, which assumes no more
checks required in parent queues), probably we should add some fields to parent queues (like
which leaf queue has reserved resources) and do allocation from top to bottom for reserved
containers.
- If we allows reserve resource beyond queue's max capacity, we should consider how to show
that to users as well. Because user could ask questions like (why my queue can use more than
total of cluster/queue-max resource).
- To solve above issue, a different approach is gradually reserve resource, IAW, adding a
"Reserving" state to container. For example, if a request needs 2G, and only 1G available
in the cluster, it will reserve 1G first, and after a while, another 1G resource available,
scheduler can finally set container state from reserving to reserved.


> CapacityScheduler reservations may not prevent indefinite postponement on a busy cluster
> ----------------------------------------------------------------------------------------
>
>                 Key: YARN-4280
>                 URL: https://issues.apache.org/jira/browse/YARN-4280
>             Project: Hadoop YARN
>          Issue Type: Bug
>          Components: capacity scheduler
>    Affects Versions: 2.6.1, 2.8.0, 2.7.1
>            Reporter: Kuhu Shukla
>            Assignee: Kuhu Shukla
>
> Consider the following scenario:
> There are 2 queues A(25% of the total capacity) and B(75%), both can run at total cluster
capacity. There are 2 applications, appX that runs on Queue A, always asking for 1G containers(non-AM)
and appY runs on Queue B asking for 2 GB containers.
> The user limit is high enough for the application to reach 100% of the cluster resource.

> appX is running at total cluster capacity, full with 1G containers releasing only one
container at a time. appY comes in with a request of 2GB container but only 1 GB is free.
Ideally, since appY is in the underserved queue, it has higher priority and should reserve
for its 2 GB request. Since this request puts the alloc+reserve above total capacity of the
cluster, reservation is not made. appX comes in with a 1GB request and since 1GB is still
available, the request is allocated. 
> This can continue indefinitely causing priority inversion.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
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