[
https://issues.apache.org/jira/browse/SPARK-30225?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Mala Chikka Kempanna updated SPARK-30225:
-----------------------------------------
Description:
There is issues with spark.unsafe.sorter.spill.read.ahead.enabled in spark
2.4.0, which is introduced by https://issues.apache.org/jira/browse/SPARK-23366
Workaround for this problem is to disable readahead of unsafe spill with
following.
--conf spark.unsafe.sorter.spill.read.ahead.enabled=false
This issue can be reproduced on Spark 2.4.0 by following the steps in this
comment of Jira SPARK-18105.
https://issues.apache.org/jira/browse/SPARK-18105?focusedCommentId=16981461&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-16981461
Exception looks like below:
<code>
19/12/10 01:51:31 INFO sort.ShuffleExternalSorter: Thread 142 spilling sort
data of 5.1 GB to disk (1 time so far)19/12/10 01:51:31 INFO
sort.ShuffleExternalSorter: Thread 142 spilling sort data of 5.1 GB to disk (1
time so far)19/12/10 01:52:48 INFO sort.ShuffleExternalSorter: Thread 142
spilling sort data of 5.1 GB to disk (2 times so far)19/12/10 01:53:53 ERROR
executor.Executor: Exception in task 6.0 in stage 0.0 (TID
6)java.io.IOException: Stream is corrupted at
net.jpountz.lz4.LZ4BlockInputStream.refill(LZ4BlockInputStream.java:202) at
net.jpountz.lz4.LZ4BlockInputStream.refill(LZ4BlockInputStream.java:228) at
net.jpountz.lz4.LZ4BlockInputStream.read(LZ4BlockInputStream.java:157) at
org.apache.spark.io.ReadAheadInputStream$1.run(ReadAheadInputStream.java:168)
at
java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
at
java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
at java.lang.Thread.run(Thread.java:748)19/12/10 01:53:53 INFO
executor.CoarseGrainedExecutorBackend: Got assigned task 3319/12/10 01:53:53
INFO executor.Executor: Running task 8.1 in stage 0.0 (TID 33)19/12/10 01:54:00
INFO sort.UnsafeExternalSorter: Thread 142 spilling sort data of 3.3 GB to disk
(0 time so far)19/12/10 01:54:30 INFO executor.Executor: Executor is trying to
kill task 8.1 in stage 0.0 (TID 33), reason: Stage cancelled19/12/10 01:54:30
INFO executor.Executor: Executor killed task 8.1 in stage 0.0 (TID 33), reason:
Stage cancelled19/12/10 01:54:52 INFO executor.CoarseGrainedExecutorBackend:
Driver commanded a shutdown
</code>
was:
There is issues with spark.unsafe.sorter.spill.read.ahead.enabled in spark
2.4.0, which is introduced by https://issues.apache.org/jira/browse/SPARK-23366
Workaround for this problem is to disable readahead of unsafe spill with
following.
--conf spark.unsafe.sorter.spill.read.ahead.enabled=false
This issue can be reproduced on Spark 2.4.0 by following the steps in this
comment of Jira SPARK-18105.
https://issues.apache.org/jira/browse/SPARK-18105?focusedCommentId=16981461&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-16981461
Exception looks like below:
<code>
19/12/10 01:51:31 INFO sort.ShuffleExternalSorter: Thread 142 spilling sort
data of 5.1 GB to disk (1 time so far)19/12/10 01:51:31 INFO
sort.ShuffleExternalSorter: Thread 142 spilling sort data of 5.1 GB to disk (1
time so far)19/12/10 01:52:48 INFO sort.ShuffleExternalSorter: Thread 142
spilling sort data of 5.1 GB to disk (2 times so far)19/12/10 01:53:53 ERROR
executor.Executor: Exception in task 6.0 in stage 0.0 (TID
6)java.io.IOException: Stream is corrupted at
net.jpountz.lz4.LZ4BlockInputStream.refill(LZ4BlockInputStream.java:202) at
net.jpountz.lz4.LZ4BlockInputStream.refill(LZ4BlockInputStream.java:228) at
net.jpountz.lz4.LZ4BlockInputStream.read(LZ4BlockInputStream.java:157) at
org.apache.spark.io.ReadAheadInputStream$1.run(ReadAheadInputStream.java:168)
at
java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
at
java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
at java.lang.Thread.run(Thread.java:748)19/12/10 01:53:53 INFO
executor.CoarseGrainedExecutorBackend: Got assigned task 3319/12/10 01:53:53
INFO executor.Executor: Running task 8.1 in stage 0.0 (TID 33)19/12/10 01:54:00
INFO sort.UnsafeExternalSorter: Thread 142 spilling sort data of 3.3 GB to disk
(0 time so far)19/12/10 01:54:30 INFO executor.Executor: Executor is trying to
kill task 8.1 in stage 0.0 (TID 33), reason: Stage cancelled19/12/10 01:54:30
INFO executor.Executor: Executor killed task 8.1 in stage 0.0 (TID 33), reason:
Stage cancelled19/12/10 01:54:52 INFO executor.CoarseGrainedExecutorBackend:
Driver commanded a shutdown
</code>
> "Stream is corrupt" exception on reading disk-spilled data of a shuffle
> operation
> ---------------------------------------------------------------------------------
>
> Key: SPARK-30225
> URL: https://issues.apache.org/jira/browse/SPARK-30225
> Project: Spark
> Issue Type: Bug
> Components: Input/Output
> Affects Versions: 2.4.0
> Reporter: Mala Chikka Kempanna
> Priority: Major
>
> There is issues with spark.unsafe.sorter.spill.read.ahead.enabled in spark
> 2.4.0, which is introduced by
> https://issues.apache.org/jira/browse/SPARK-23366
>
> Workaround for this problem is to disable readahead of unsafe spill with
> following.
> --conf spark.unsafe.sorter.spill.read.ahead.enabled=false
>
> This issue can be reproduced on Spark 2.4.0 by following the steps in this
> comment of Jira SPARK-18105.
> https://issues.apache.org/jira/browse/SPARK-18105?focusedCommentId=16981461&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-16981461
>
> Exception looks like below:
>
> <code>
> 19/12/10 01:51:31 INFO sort.ShuffleExternalSorter: Thread 142 spilling sort
> data of 5.1 GB to disk (1 time so far)19/12/10 01:51:31 INFO
> sort.ShuffleExternalSorter: Thread 142 spilling sort data of 5.1 GB to disk
> (1 time so far)19/12/10 01:52:48 INFO sort.ShuffleExternalSorter: Thread 142
> spilling sort data of 5.1 GB to disk (2 times so far)19/12/10 01:53:53 ERROR
> executor.Executor: Exception in task 6.0 in stage 0.0 (TID
> 6)java.io.IOException: Stream is corrupted at
> net.jpountz.lz4.LZ4BlockInputStream.refill(LZ4BlockInputStream.java:202) at
> net.jpountz.lz4.LZ4BlockInputStream.refill(LZ4BlockInputStream.java:228) at
> net.jpountz.lz4.LZ4BlockInputStream.read(LZ4BlockInputStream.java:157) at
> org.apache.spark.io.ReadAheadInputStream$1.run(ReadAheadInputStream.java:168)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
> at java.lang.Thread.run(Thread.java:748)19/12/10 01:53:53 INFO
> executor.CoarseGrainedExecutorBackend: Got assigned task 3319/12/10 01:53:53
> INFO executor.Executor: Running task 8.1 in stage 0.0 (TID 33)19/12/10
> 01:54:00 INFO sort.UnsafeExternalSorter: Thread 142 spilling sort data of 3.3
> GB to disk (0 time so far)19/12/10 01:54:30 INFO executor.Executor: Executor
> is trying to kill task 8.1 in stage 0.0 (TID 33), reason: Stage
> cancelled19/12/10 01:54:30 INFO executor.Executor: Executor killed task 8.1
> in stage 0.0 (TID 33), reason: Stage cancelled19/12/10 01:54:52 INFO
> executor.CoarseGrainedExecutorBackend: Driver commanded a shutdown
>
> </code>
>
>
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
