[GitHub] spark pull request #13077: [SPARK-10748] [Mesos] Log error instead of crashi...

[srowen <gi...@git.apache.org>]

Github user srowen commented on a diff in the pull request:

    https://github.com/apache/spark/pull/13077#discussion_r72010691
  
    --- Diff: core/src/main/scala/org/apache/spark/scheduler/cluster/mesos/MesosClusterScheduler.scala ---
    @@ -502,41 +502,53 @@ private[spark] class MesosClusterScheduler(
           } else {
             val offer = offerOption.get
             val taskId = TaskID.newBuilder().setValue(submission.submissionId).build()
    -        val (remainingResources, cpuResourcesToUse) =
    -          partitionResources(offer.resources, "cpus", driverCpu)
    -        val (finalResources, memResourcesToUse) =
    -          partitionResources(remainingResources.asJava, "mem", driverMem)
    -        val commandInfo = buildDriverCommand(submission)
    -        val appName = submission.schedulerProperties("spark.app.name")
    -        val taskInfo = TaskInfo.newBuilder()
    -          .setTaskId(taskId)
    -          .setName(s"Driver for $appName")
    -          .setSlaveId(offer.slaveId)
    -          .setCommand(commandInfo)
    -          .addAllResources(cpuResourcesToUse.asJava)
    -          .addAllResources(memResourcesToUse.asJava)
    -        offer.resources = finalResources.asJava
    -        submission.schedulerProperties.get("spark.mesos.executor.docker.image").foreach { image =>
    -          val container = taskInfo.getContainerBuilder()
    -          val volumes = submission.schedulerProperties
    -            .get("spark.mesos.executor.docker.volumes")
    -            .map(MesosSchedulerBackendUtil.parseVolumesSpec)
    -          val portmaps = submission.schedulerProperties
    -            .get("spark.mesos.executor.docker.portmaps")
    -            .map(MesosSchedulerBackendUtil.parsePortMappingsSpec)
    -          MesosSchedulerBackendUtil.addDockerInfo(
    -            container, image, volumes = volumes, portmaps = portmaps)
    -          taskInfo.setContainer(container.build())
    +        var commandInfo: CommandInfo = null
    +        try {
    +          commandInfo = buildDriverCommand(submission)
    +        } catch {
    +          case e: SparkException =>
    +            afterLaunchCallback(submission.submissionId)
    +            finishedDrivers += new MesosClusterSubmissionState(submission, taskId,
    +              SlaveID.newBuilder().setValue("").build(), None, null, None)
    +            logError(s"Failed to launch the driver with id: ${submission.submissionId}, " +
    +              s"cpu: $driverCpu, mem: $driverMem, reason: ${e.getMessage}")
    +        }
    +        if (commandInfo != null) {
    --- End diff --
    
    Rather than make a big long if statement, is it simpler and clearer to just return from the method in the catch block? then you can even ...
    
    ```
    val commandInfo = try { buildDriverCommand(submission) } ...
    ```


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---

---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org