Hi Martin, I'm not using bin/grid, but a find the logs, here are of one container http://pastebin.com/JskNReSt it seems that the failure is caused by class not found exception for JsonSerdeFactory !!
I have a dependency on *samza-serializers_2.10* in the pom.xml <https://github.com/dzlab/bigdata-finance/blob/master/rule-engine/pom.xml> with a runtime scope, I think I should remove this scope when comparing to pom.xml in http://samza.incubator.apache.org/startup/download/ On Fri, Nov 21, 2014 at 11:55 AM, Martin Kleppmann <[email protected]> wrote: > Hi Arbi, > > The app master log unfortunately doesn't contain the root cause. The > exceptions only say that a container failed, but not why. Could you please > look in the container logs? > > If you're running YARN through hello-samza's bin/grid, you'll find the > logs in hello-samza/deploy/yarn/logs/userlogs/application_*/container_*/ > (the container ending in _00001 is the app master, those ending in _00002 > or higher should contain the exception we're looking for). > > Martin > > On 21 Nov 2014, at 11:00, Arbi Akhina <[email protected]> wrote: > > > Hi Fang, thanks for the hint, > > I've checked the src.xml file many times but never noticed this mistake!! > > now the samza jobs works (we gona soon have lot of fun) and it's able to > > connect to kafka, however the container running the samza tasks fail for > > some reason! > > > > here is the log file of the app master http://pastebin.com/iEkvCWKN > > > > bests, > > > > On Thu, Nov 20, 2014 at 8:55 PM, Yan Fang <[email protected]> wrote: > > > >> Hi Arbi, > >> > >> To fix it, you need to do the following two steps: > >> > >> 1) in the src.xml change <include>samza:samza-jobs</include> to > <include> > >> *dz.lab.finance*:samza-jobs</include> . (You put the wrong groupId, as a > >> result, the assembly can not include that project) > >> > >> 2) remove the > >> > >> <!--added--> > >> > >> <dependency> > >> > >> <groupId>org.apache.samza</groupId> > >> > >> <artifactId>samza-api</artifactId> > >> > >> </dependency> > >> in rule-engine/samza-job-package/pom.xml .(It seems that, if you put > the > >> dependency in the assembly's pom but not "include" in the src.xml, it > will > >> be ignored) > >> > >> > >> Then it will work. > >> > >> Thanks, > >> > >> Fang, Yan > >> [email protected] > >> +1 (206) 849-4108 > >> > >> On Thu, Nov 20, 2014 at 2:06 AM, Arbi Akhina <[email protected]> > >> wrote: > >> > >>> Hi Chris, > >>> I can see the following samza jars under the lib folder of the > generated > >>> tar.gz: > >>> > >>> samza-kv_2.10-0.7.0.jar > >>> samza-core_2.10-0.7.0.jar > >>> samza-serializers_2.10-0.7.0.jar > >>> samza-yarn_2.10-0.7.0.jar > >>> samza-kafka_2.10-0.7.0.jar > >>> > >>> for some reason the samaza-api.jar is not included even if it's > >> referenced > >>> in my pom.xml. > >>> Here is a link to the samza project > >>> https://github.com/dzlab/bigdata-finance/tree/master/rule-engine > >>> > >>> bests, > >>> > >>> On Thu, Nov 20, 2014 at 12:03 AM, Chris Riccomini < > >>> [email protected]> wrote: > >>> > >>>> Hey Arbi, > >>>> > >>>> If you're missing the samza-api JAR, you're probably missing the rest > >> as > >>>> well, which would cause the job to fail. I recommend taking a look at > >>>> what's in your job's .tgz file. In the lib directory, there should be > a > >>>> bunch of JARs, including samza-api, samza-core, etc. If there's not, > >> you > >>>> have a problem with your package assembly. This can be fixed by > >> updating > >>>> the assembly/src.xml file accordingly. > >>>> > >>>> Cheers, > >>>> Chris > >>>> > >>>> On 11/19/14 2:53 PM, "Arbi Akhina" <[email protected]> wrote: > >>>> > >>>>> Hi, > >>>>> I'm having trouble getting to work a modified version of the of > >>>>> hello-samza > >>>>> (i just replaced the 3 wikipedia streaming tasks with my own task). > >>>>> > >>>>> When I do *mvn clean package* and then extract the generated tar.gz > >>> file I > >>>>> can't find samza-api.jar & sl4j-api.jar under the lib folder which > >> cause > >>>>> the failure of the submission (i.e. run-job.sh) with a class not > found > >>>>> exception. > >>>>> > >>>>> I had to add these jars manually so that the submission works but > then > >>> on > >>>>> the ResourceManger's UI I see the job as failed (containers exiting > >> with > >>>>> something like -1000). I can't find the logs left by the app master > or > >>> any > >>>>> one of the containers. I'm enabling logs aggregation, on hdfs I can > >> see > >>> a > >>>>> log file with a name like appId_hostname_port.log but *yarn logs > >>>>> -applicarionId <appId>* prints just the content of the classpath. > >>>>> > >>>>> Any hint that could help investigating this issue? > >>>>> > >>>>> bests > >>>> > >>>> > >>> > >> > >
