[ 
https://issues.apache.org/jira/browse/FLINK-8813?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16383476#comment-16383476
 ] 

Till Rohrmann commented on FLINK-8813:
--------------------------------------

Auto max parallelism does not make much sense anymore in the Flip-6 world. Yarn 
clusters, for example, will no longer be started with a set of preallocated 
resources. Therefore, it is not clear what max parallelism would mean in such a 
case. I think we should discourage people from using it and instead motivate 
them to always specify the parallelism with which they want to run the job.

As a fix for this concrete problem, we could say that in Flip-6 mode, the max 
parallelism is always {{1}}. What do you think?

> AutoParallellismITCase fails with Flip6
> ---------------------------------------
>
>                 Key: FLINK-8813
>                 URL: https://issues.apache.org/jira/browse/FLINK-8813
>             Project: Flink
>          Issue Type: Bug
>          Components: JobManager, Tests
>    Affects Versions: 1.5.0
>            Reporter: Chesnay Schepler
>            Priority: Blocker
>              Labels: flip-6
>             Fix For: 1.5.0
>
>
> The {{AutoParallelismITCase}} fails when running against flip6. 
> ([https://travis-ci.org/zentol/flink/jobs/347373854)]
> It appears that the {{JobMaster}} does not properly handle 
> {{ExecutionConfig#PARALLELISM_AUTO_MAX}}.
>  
> Exception:
> {code:java}
> Caused by: org.apache.flink.runtime.client.JobSubmissionException: Could not 
> start JobManager.
>       at 
> org.apache.flink.runtime.dispatcher.Dispatcher.submitJob(Dispatcher.java:287)
>       at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
>       at 
> sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
>       at 
> sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
>       at java.lang.reflect.Method.invoke(Method.java:498)
>       at 
> org.apache.flink.runtime.rpc.akka.AkkaRpcActor.handleRpcInvocation(AkkaRpcActor.java:210)
>       at 
> org.apache.flink.runtime.rpc.akka.AkkaRpcActor.handleMessage(AkkaRpcActor.java:154)
>       at 
> org.apache.flink.runtime.rpc.akka.FencedAkkaRpcActor.handleMessage(FencedAkkaRpcActor.java:66)
>       at 
> org.apache.flink.runtime.rpc.akka.AkkaRpcActor.lambda$onReceive$1(AkkaRpcActor.java:132)
>       at 
> akka.actor.ActorCell$$anonfun$become$1.applyOrElse(ActorCell.scala:544)
>       at akka.actor.Actor$class.aroundReceive(Actor.scala:502)
>       at akka.actor.UntypedActor.aroundReceive(UntypedActor.scala:95)
>       at akka.actor.ActorCell.receiveMessage(ActorCell.scala:526)
>       at akka.actor.ActorCell.invoke(ActorCell.scala:495)
>       at akka.dispatch.Mailbox.processMailbox(Mailbox.scala:257)
>       at akka.dispatch.Mailbox.run(Mailbox.scala:224)
>       at akka.dispatch.Mailbox.exec(Mailbox.scala:234)
>       at scala.concurrent.forkjoin.ForkJoinTask.doExec(ForkJoinTask.java:260)
>       at 
> scala.concurrent.forkjoin.ForkJoinPool$WorkQueue.runTask(ForkJoinPool.java:1339)
>       at 
> scala.concurrent.forkjoin.ForkJoinPool.runWorker(ForkJoinPool.java:1979)
>       at 
> scala.concurrent.forkjoin.ForkJoinWorkerThread.run(ForkJoinWorkerThread.java:107)
> Caused by: org.apache.flink.runtime.client.JobExecutionException: Could not 
> set up JobManager
>       at 
> org.apache.flink.runtime.jobmaster.JobManagerRunner.<init>(JobManagerRunner.java:181)
>       at 
> org.apache.flink.runtime.dispatcher.Dispatcher$DefaultJobManagerRunnerFactory.createJobManagerRunner(Dispatcher.java:747)
>       at 
> org.apache.flink.runtime.dispatcher.Dispatcher.submitJob(Dispatcher.java:243)
>       ... 20 more
> Caused by: java.lang.IllegalArgumentException: The parallelism must be at 
> least one.
>       at 
> org.apache.flink.runtime.jobgraph.JobVertex.setParallelism(JobVertex.java:290)
>       at 
> org.apache.flink.runtime.executiongraph.ExecutionGraphBuilder.buildGraph(ExecutionGraphBuilder.java:162)
>       at 
> org.apache.flink.runtime.jobmaster.JobMaster.<init>(JobMaster.java:295)
>       at 
> org.apache.flink.runtime.jobmaster.JobManagerRunner.<init>(JobManagerRunner.java:170)
>       ... 22 more{code}
>  
> The likely culprit is this call to {{ExecutionGraphBuilder#buildGraph}} in 
> the {{JobMaster}} constructor:
> {code:java}
> this.executionGraph = ExecutionGraphBuilder.buildGraph(
>    null,
>    jobGraph,
>    jobMasterConfiguration.getConfiguration(),
>    scheduledExecutorService,
>    scheduledExecutorService,
>    slotPool.getSlotProvider(),
>    userCodeLoader,
>    highAvailabilityServices.getCheckpointRecoveryFactory(),
>    rpcTimeout,
>    restartStrategy,
>    jobMetricGroup,
>    -1, // parallelismForAutoMax
>    blobServer,
>    jobMasterConfiguration.getSlotRequestTimeout(),
>    log);{code}



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to