Mailing-List: contact yarn-issues-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: yarn-issues@hadoop.apache.org
Date: Tue, 25 Nov 2014 00:49:12 +0000 (UTC)
From: "Sunil G (JIRA)" <jira@apache.org>
To: yarn-issues@hadoop.apache.org
Message-ID: <JIRA.12711498.1398875224000.11743.1416876552641@Atlassian.JIRA>
In-Reply-To: <JIRA.12711498.1398875224000@Atlassian.JIRA>
References: <JIRA.12711498.1398875224000@Atlassian.JIRA>
 <JIRA.12711498.1398875224477@arcas>
Subject: [jira] [Commented] (YARN-2009) Priority support for preemption in
 ProportionalCapacityPreemptionPolicy
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/YARN-2009?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14223860#comment-14223860 ] 

Sunil G commented on YARN-2009:
-------------------------------

Thank you [~curino] for the thoughts.
I understand the complexity for the user when they experience preemption of few containers and that itself may be tougher for them to understand why that container is preempted and reasons for it. In a simple way, if timestamp is only considered (to an extent user-limit factor also now), that itself will be tougher to express through logs.

Hence solving some of these small imbalances like what I mentioned may not help much the users in a big level. Based on use cases we can check whether these are needed later. Coming to the focus of this JIRA, within a queue if slow and low priority applications are running and consuming full resources, it will be good if we can make some space by preempting lower priority ones. This preemption can be done within a Queue. we have seen some lower priority applications are taking more cluster and higher priority applications are waiting to launch for long time. Please suggest your thoughts on this.

> Priority support for preemption in ProportionalCapacityPreemptionPolicy
> -----------------------------------------------------------------------
>
>                 Key: YARN-2009
>                 URL: https://issues.apache.org/jira/browse/YARN-2009
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: capacityscheduler
>            Reporter: Devaraj K
>            Assignee: Sunil G
>
> While preempting containers based on the queue ideal assignment, we may need to consider preempting the low priority application containers first.


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