[
https://issues.apache.org/jira/browse/SPARK-5333?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14284135#comment-14284135
]
Josh Rosen commented on SPARK-5333:
-----------------------------------
Good catch. I've created a link to the SPARK-4014 JIRA so that we don't forget
to backport this patch, too, when porting that patch to earlier branches.
> [Mesos] MesosTaskLaunchData occurs BufferUnderflowException
> -----------------------------------------------------------
>
> Key: SPARK-5333
> URL: https://issues.apache.org/jira/browse/SPARK-5333
> Project: Spark
> Issue Type: Bug
> Components: Mesos
> Reporter: Jongyoul Lee
> Priority: Blocker
>
> MesosTaskLaunchData occurs exception when MesosExecutorBackend launches task
> because serializedTask.remaining is 0.
> {code}
> Exception in thread "Thread-6" java.nio.BufferUnderflowException
> at java.nio.Buffer.nextGetIndex(Buffer.java:498)
> at java.nio.HeapByteBuffer.getInt(HeapByteBuffer.java:355)
> at
> org.apache.spark.scheduler.cluster.mesos.MesosTaskLaunchData$.fromByteString(MesosTaskLaunchData.scala:46)
> at
> org.apache.spark.executor.MesosExecutorBackend.launchTask(MesosExecutorBackend.scala:81)
> {code}
> I've checked this bug with fine-grained mode. This is because
> MesosTaskLaunchData.toByteString doesn't rewind byteBuffer after they put
> data.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
