You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Sandeep Katta (Jira)" <ji...@apache.org> on 2021/03/03 07:37:00 UTC
[jira] [Commented] (SPARK-21564) TaskDescription decoding failure
should fail the task
[ https://issues.apache.org/jira/browse/SPARK-21564?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17294347#comment-17294347 ]
Sandeep Katta commented on SPARK-21564:
---------------------------------------
Recently I have hit with decode error, irony is all the tasks in the same taskset were able to deserialize but only 1 task is failed .
Which says that the data is corrupted, most of the times it will be very difficult to analyze why the data is corrupted , so for these kind of intermittent issue exception handling should be in place to achieve fault tolerant
*21/02/11 07:53:39 ERROR Inbox: Ignoring errorjava.io.UTFDataFormatException: malformed input around byte 5 at* java.io.DataInputStream.readUTF(DataInputStream.java:656) at java.io.DataInputStream.readUTF(DataInputStream.java:564) at org.apache.spark.scheduler.TaskDescription$$anonfun$deserializeStringLongMap$1.apply$mcVI$sp(TaskDescription.scala:110) at scala.collection.immutable.Range.foreach$mVc$sp(Range.scala:160) at org.apache.spark.scheduler.TaskDescription$.deserializeStringLongMap(TaskDescription.scala:109) at org.apache.spark.scheduler.TaskDescription$.decode(TaskDescription.scala:125) at org.apache.spark.executor.CoarseGrainedExecutorBackend$$anonfun$receive$1.applyOrElse(CoarseGrainedExecutorBackend.scala:100) at org.apache.spark.rpc.netty.Inbox$$anonfun$process$1.apply$mcV$sp(Inbox.scala:117) at org.apache.spark.rpc.netty.Inbox.safelyCall(Inbox.scala:205) at org.apache.spark.rpc.netty.Inbox.process(Inbox.scala:101) at org.apache.spark.rpc.netty.Dispatcher$MessageLoop.run(Dispatcher.scala:226) at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624) at java.lang.Thread.run(Thread.java:748)
!image-2021-03-03-13-02-31-744.png!
So it's better to have fault tolerant in place, Spark Driver does not have any idea about this exception so it still waits for this task to complete, thus the job is in zombie stage
CC [~dongjoon] [~hyukjin.kwon] [~cloud_fan] tagging you guys for more traction
> TaskDescription decoding failure should fail the task
> -----------------------------------------------------
>
> Key: SPARK-21564
> URL: https://issues.apache.org/jira/browse/SPARK-21564
> Project: Spark
> Issue Type: Bug
> Components: Spark Core
> Affects Versions: 2.2.0
> Reporter: Andrew Ash
> Priority: Major
> Labels: bulk-closed
> Attachments: image-2021-03-03-13-02-06-669.png, image-2021-03-03-13-02-31-744.png
>
>
> cc [~robert3005]
> I was seeing an issue where Spark was throwing this exception:
> {noformat}
> 16:16:28.294 [dispatcher-event-loop-14] ERROR org.apache.spark.rpc.netty.Inbox - Ignoring error
> java.io.EOFException: null
> at java.io.DataInputStream.readFully(DataInputStream.java:197)
> at java.io.DataInputStream.readUTF(DataInputStream.java:609)
> at java.io.DataInputStream.readUTF(DataInputStream.java:564)
> at org.apache.spark.scheduler.TaskDescription$$anonfun$decode$1.apply(TaskDescription.scala:127)
> at org.apache.spark.scheduler.TaskDescription$$anonfun$decode$1.apply(TaskDescription.scala:126)
> at scala.collection.immutable.Range.foreach(Range.scala:160)
> at org.apache.spark.scheduler.TaskDescription$.decode(TaskDescription.scala:126)
> at org.apache.spark.executor.CoarseGrainedExecutorBackend$$anonfun$receive$1.applyOrElse(CoarseGrainedExecutorBackend.scala:95)
> at org.apache.spark.rpc.netty.Inbox$$anonfun$process$1.apply$mcV$sp(Inbox.scala:117)
> at org.apache.spark.rpc.netty.Inbox.safelyCall(Inbox.scala:205)
> at org.apache.spark.rpc.netty.Inbox.process(Inbox.scala:101)
> at org.apache.spark.rpc.netty.Dispatcher$MessageLoop.run(Dispatcher.scala:213)
> at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:748)
> {noformat}
> For details on the cause of that exception, see SPARK-21563
> We've since changed the application and have a proposed fix in Spark at the ticket above, but it was troubling that decoding the TaskDescription wasn't failing the tasks. So the Spark job ended up hanging and making no progress, permanently stuck, because the driver thinks the task is running but the thread has died in the executor.
> We should make a change around https://github.com/apache/spark/blob/v2.2.0/core/src/main/scala/org/apache/spark/executor/CoarseGrainedExecutorBackend.scala#L96 so that when that decode throws an exception, the task is marked as failed.
> cc [~kayousterhout] [~irashid]
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org