Mailing-List: contact yarn-dev-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: yarn-dev@hadoop.apache.org
Received-SPF: neutral (nike.apache.org: 216.145.54.173 is neither permitted
 nor denied by domain of matthieu@yahoo-inc.com)
From: Matthieu Morel <matthieu@yahoo-inc.com>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: quoted-printable
Subject: Recommended way to kill an app?
Date: Thu, 22 Nov 2012 15:07:26 +0100
Message-Id: <E446FDC3-C0A3-4DA8-8281-A8CD15CC8D52@yahoo-inc.com>
To: yarn-dev@hadoop.apache.org
Mime-Version: 1.0 (Apple Message framework v1283)

Hello,

I am working with yarn 2.0.2-alpha.


I want to be able to properly terminate long running applications =
deployed on Yarn. However, I can't see any tool to do that from the =
command line (something like "hadoop job -kill #jobid". but for yarn). =
It's not clear to me how to do that programmatically either.=20

Is this left to the developer of the application master? Is there a =
recommended way?


Currently an option I see is to implement a callback from some =
notification in the application master (e.g. a signal from ZooKeeper), =
terminate the launched container properly, through the =
ContainerManager#stopContainer method. But there might be a more generic =
way?

I also noted that ClientRMProtocol#forceKillApplication does not =
properly kill spawned processes, apart from the application master, at =
least when using MiniYarnCluster in tests.


Thanks for any suggestion or recommendation !


Matthieu