Mailing-List: contact yarn-issues-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Date: Mon, 26 Sep 2016 05:14:20 +0000 (UTC)
From: "Rohith Sharma K S (JIRA)" <jira@apache.org>
To: yarn-issues@hadoop.apache.org
Message-ID: <JIRA.12896089.1443095858000.659974.1474866860695@Atlassian.JIRA>
In-Reply-To: <JIRA.12896089.1443095858000@Atlassian.JIRA>
References: <JIRA.12896089.1443095858000@Atlassian.JIRA> <JIRA.12896089.1443095858707@arcas>
Subject: [jira] [Commented] (YARN-4205) Add a service for monitoring
 application life time out
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit
archived-at: Mon, 26 Sep 2016 05:14:27 -0000


    [ https://issues.apache.org/jira/browse/YARN-4205?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15522077#comment-15522077 ] 

Rohith Sharma K S commented on YARN-4205:
-----------------------------------------

Thanks [~gsaha] for the suggestion. Some clarifications on this JIRA. Few points might be re-iterated from my earlier comments also.

# This JIRA mainly aims at providing overall execution timeout of application. And monitor start from app-submission-time to completing states.
# And the patch tries to maintain compatibility wrt client APIs. Any support for new timeouts(allocation 0r queue_timeout etc) in future, implementer need to add new set of API's in ApplicationTimeout object and corresponding implementation in server side. *To the user, there will no change in the ApplicationSubmissionContext APIs.*
# The timeout can be started from any of the states like SUBMISSION/LAUNCH/RUNNING for an application. But some of the constraints from YARN are
## Say, if there is any multiple attempt failure, then there will be multiple launch time. Does timeout for each attempt? monitor start time would be keep changing where each time attempt launch need to register and unregister when failure happens. Similar issue is for RUNNIG state also.
## Some timeouts can not be supported such as state_store because these operations are/can-be too frequent. In this cases, any issue with statestore, RM itself will go down. Concrete use case for supporting this timeout is not yet known.

And would you elaborate use case for having timeout from launch-time/running states? 

cc :/[~vinodkv] [~jianhe]

> Add a service for monitoring application life time out
> ------------------------------------------------------
>
>                 Key: YARN-4205
>                 URL: https://issues.apache.org/jira/browse/YARN-4205
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: scheduler
>            Reporter: nijel
>            Assignee: Rohith Sharma K S
>         Attachments: 0001-YARN-4205.patch, 0002-YARN-4205.patch, 0003-YARN-4205.patch, 0004-YARN-4205.patch, 0005-YARN-4205.patch, 0006-YARN-4205.patch, YARN-4205_01.patch, YARN-4205_02.patch, YARN-4205_03.patch
>
>
> This JIRA intend to provide a lifetime monitor service. 
> The service will monitor the applications where the life time is configured. If the application is running beyond the lifetime, it will be killed. 
> The lifetime will be considered from the submit time.
> The thread monitoring interval is configurable.


--
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