Hi Chris Thanks a lot for you advice. I will try.
On Mon, Dec 8, 2014 at 10:32 AM, Chris Riccomini < [email protected]> wrote: > Hey Jae, > > I've opened: > > https://issues.apache.org/jira/browse/SAMZA-491 > > To track this. Could you please post your patch there? > > Cheers, > Chris > > From: Chris Riccomini <[email protected]<mailto: > [email protected]>> > Date: Monday, December 8, 2014 10:30 AM > To: "[email protected]<mailto:[email protected]>" > <[email protected]<mailto:[email protected]>> > Subject: Re: How to set up yarn queue name > > Hey Jae, > > I don't believe Samza supports this right now. We've always run in the > default queue. This is definitely a feature we should add. Would you be > willing to try to patch your Samza library to fix it? > > The place to fix it is in ClientHelper.scala class. The > submitApplicaiton() method has an appCtx class. I believe you'll need to > set the queue there using ApplicationSubmissionContext.setQueue. > > The steps are: > > 1) Add a `yarn.queue` field to YarnConfig. > 2) Update YarnJob.scala's submit() method to get the yarn queue, and use > it in the client.submitApplication() call. > 3) Update ClientHelper.scala's submitApplication() method to take an > optional queue, and if it's set, add it to the > ApplicationSubmissionContext, as I described above. > > Cheers, > Chris > > From: <Bae>, Jae Hyeon <[email protected]<mailto:[email protected]>> > Reply-To: "[email protected]<mailto: > [email protected]>" <[email protected]<mailto: > [email protected]>> > Date: Monday, December 8, 2014 10:15 AM > To: "[email protected]<mailto:[email protected]>" > <[email protected]<mailto:[email protected]>> > Subject: How to set up yarn queue name > > Hello > > I tried submit Samza job in our EMR hadoop cluster but I got the following > error: > > [Inline image 1] > > So, I asked our company hadoop admin and I got advised to set up the queue > name. I added the following line in Samza job property but it didn't work: > > mapreduce.job.queuename=root.adhoc > yarn.scheduler.fair.user-as-default-queue=false > > Do you know how to set up yarn queue name? > > FYI, I copied our hadoop cluster fair scheduler policy. We disabled > default queue. > > <?xml version="1.0"?> > <allocations> > <queue name="adhoc" type="parent"> > <weight>10</weight> > <maxRunningApps>300</maxRunningApps> > <queue name="nsharma"> > <weight>5</weight> > </queue> > </queue> > > <queue name="default"> > <weight>0.0</weight> > </queue> > > <queue name="emergencyuseonly"> > <weight>0.0</weight> > <minResources>0 mb, 0 vcores</minResources> > <minSharePreemptionTimeout>10000</minSharePreemptionTimeout> > </queue> > > <queuePlacementPolicy> > <rule name="nestedUserQueue"> > <rule name="specified" create="false" /> > </rule> > <rule name="specified" create="false" /> > <rule name="reject" /> > </queuePlacementPolicy> > > <userMaxAppsDefault>20</userMaxAppsDefault> > <fairSharePreemptionTimeout>60</fairSharePreemptionTimeout> > </allocations> > > Thank you > Best, Jae >
