Hi Chris, now I have the tar file in my RM machine, and the yarn path points to it. I changed the core-site.xml to use HttpFileSystem instead of HDFS now it is failing with
Application application_1407640485281_0001 failed 2 times due to AM Container for appattempt_1407640485281_0001_000002 exited with exitCode:-1000 due to: java.lang.ClassNotFoundException: Class org.apache.samza.util.hadoop.HttpFileSystem not found I think I can solve this just installing scala files from the samza tutorial, can you confirm that? On 09 Aug 2014, at 08:34, Telles Nobrega <[email protected]> wrote: > Hi Chris, > > I think the problem is that I forgot to update the yarn.job.package. > I will try again to see if it works now. > > I have one more question, how can I stop (command line) the jobs running in > my topology, for the experiment that I will run, I need to run the same job > in 4 minutes intervals. So I need to kill it, clean the kafka topics and > rerun. > > Thanks in advance. > > On 08 Aug 2014, at 12:41, Chris Riccomini <[email protected]> > wrote: > >> Hey Telles, >> >>>> Do I need to have the job folder on each machine in my cluster? >> >> No, you should not need to do this. There are two ways to deploy your >> tarball to the YARN grid. One is to put it in HDFS, and the other is to >> put it on an HTTP server. The link to running a Samza job in a multi-node >> YARN cluster describes how to do both (either HTTP server or HDFS). >> >> In both cases, once the tarball is put in on the HTTP/HDFS server(s), you >> must update yarn.package.path to point to it. From there, the YARN NM >> should download it for you automatically when you start your job. >> >> * Can you send along a paste of your job config? >> >> Cheers, >> Chris >> >> On 8/8/14 8:04 AM, "Claudio Martins" <[email protected]> wrote: >> >>> Hi Telles, it looks to me that you forgot to update the >>> "yarn.package.path" >>> attribute in your config file for the task. >>> >>> - Claudio Martins >>> Head of Engineering >>> MobileAware USA Inc. / www.mobileaware.com >>> office: +1 617 986 5060 / mobile: +1 617 480 5288 >>> linkedin: www.linkedin.com/in/martinsclaudio >>> >>> >>> On Fri, Aug 8, 2014 at 10:55 AM, Telles Nobrega <[email protected]> >>> wrote: >>> >>>> Hi, >>>> >>>> this is my first time trying to run a job on a multinode environment. I >>>> have the cluster set up, I can see in the GUI that all nodes are >>>> working. >>>> Do I need to have the job folder on each machine in my cluster? >>>> - The first time I tried running with the job on the namenode machine >>>> and >>>> it failed saying: >>>> >>>> Application application_1407509228798_0001 failed 2 times due to AM >>>> Container for appattempt_1407509228798_0001_000002 exited with exitCode: >>>> -1000 due to: File >>>> >>>> >>>> file:/home/ubuntu/alarm-samza/samza-job-package/target/samza-job-package- >>>> 0.7.0-dist.tar.gz >>>> does not exist >>>> >>>> So I copied the folder to each machine in my cluster and got this error: >>>> >>>> Application application_1407509228798_0002 failed 2 times due to AM >>>> Container for appattempt_1407509228798_0002_000002 exited with exitCode: >>>> -1000 due to: Resource >>>> >>>> >>>> file:/home/ubuntu/alarm-samza/samza-job-package/target/samza-job-package- >>>> 0.7.0-dist.tar.gz >>>> changed on src filesystem (expected 1407509168000, was 1407509434000 >>>> >>>> What am I missing? >>>> >>>> p.s.: I followed this >>>> <https://github.com/yahoo/samoa/wiki/Executing-SAMOA-with-Apache-Samza> >>>> tutorial >>>> and this >>>> < >>>> >>>> http://samza.incubator.apache.org/learn/tutorials/0.7.0/run-in-multi-node >>>> -yarn.html >>>>> >>>> to >>>> set up the cluster. >>>> >>>> Help is much appreciated. >>>> >>>> Thanks in advance. >>>> >>>> -- >>>> ------------------------------------------ >>>> Telles Mota Vidal Nobrega >>>> M.sc. Candidate at UFCG >>>> B.sc. in Computer Science at UFCG >>>> Software Engineer at OpenStack Project - HP/LSD-UFCG >>>> >> >
