Hi Leo,

Akka is used to transfer the data back to the master, and there is a
setting in Akka for the max message size, which is default to 10 MB here,
you can find it at:
core/src/main/scala/org/apache/spark/util/AkkaUtils.scala

So just increase spark.akka.frameSize to a larger number.




On Wed, Dec 18, 2013 at 4:49 PM, [email protected]
<[email protected]>wrote:

>  Hi, everyone
>
> I have a problem when I run the WordCount example. I read 6G data from
> hdfs , when I run collect(), the executer had died .
> there is the exception :
>  13/12/18 13:19:39 INFO ClusterTaskSetManager: Lost TID 55 (task 0.0:3)
> 13/12/18 13:19:39 INFO ClusterTaskSetManager: Loss was due to task 55
> result exceeding Akka frame size; aborting job
> 13/12/18 13:19:39 INFO ClusterScheduler: Remove TaskSet 0.0 from pool
> 13/12/18 13:19:39 INFO DAGScheduler: Failed to run collect at
> JavaWordCount.java:60
> Exception in thread "main" org.apache.spark.SparkException: Job failed:
> Task 55 result exceeded Akka frame size
>         at
> org.apache.spark.scheduler.DAGScheduler$$anonfun$abortStage$1.apply(DAGScheduler.scala:760)
>         at
> org.apache.spark.scheduler.DAGScheduler$$anonfun$abortStage$1.apply(DAGScheduler.scala:758)
>         at
> scala.collection.mutable.ResizableArray$class.foreach(ResizableArray.scala:60)
>         at
> scala.collection.mutable.ArrayBuffer.foreach(ArrayBuffer.scala:47)
>         at
> org.apache.spark.scheduler.DAGScheduler.abortStage(DAGScheduler.scala:758)
>         at
> org.apache.spark.scheduler.DAGScheduler.processEvent(DAGScheduler.scala:379)
>         at org.apache.spark.scheduler.DAGScheduler.org
> $apache$spark$scheduler$DAGScheduler$$run(DAGScheduler.scala:441)
>         at
> org.apache.spark.scheduler.DAGScheduler$$anon$1.run(DAGScheduler.scala:149)
>
> I saw there are some issues about this question in the github , it seems
> that if the middle resultset is larger than Akka frame size , the job will
> fail .
> I want to know if I can change some params to solve the problem ?
>
> Thanks
>
> Leo
>
> ------------------------------
>  [email protected]
>

Reply via email to