spark-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From 阎志涛 <tony....@tendcloud.com>
Subject 答复: 答复: Executor hang
Date Tue, 09 Oct 2018 01:16:10 GMT
Yeah, the problem was work around by adding config spark.sql.codegen.wholeStage=false.

Thanks and Regards,
Tony

发件人: kathleen li <kathleenli168@gmail.com>
发送时间: 2018年10月8日 12:15
收件人: 阎志涛 <tony.yan@tendcloud.com>
抄送: user@spark.apache.org
主题: Re: 答复: Executor hang



https://jaceklaskowski.gitbooks.io/mastering-spark-sql/spark-sql-whole-stage-codegen.html#spark.sql.codegen.wholeStage
Sent from my iPhone

On Oct 7, 2018, at 3:21 PM, 阎志涛 <tony.yan@tendcloud.com<mailto:tony.yan@tendcloud.com>>
wrote:
It is not data skewed issue. The executor running more than 8 hours and I have to kill the
executor. I doubt the problem is caused by Spark codegen but it sound spark.sql.codegen.wholeStage=false
can not work. I met another problem with codegen which cause my spark job failed with following
stack trace:
18/10/08 01:58:36 WARN TaskSetManager: Lost task 1.0 in stage 7.0 (TID 1912, data1.aidata360.com<http://data1.aidata360.com>,
executor 4): java.util.NoSuchElementException
       at java.util.LinkedList.removeFirst(LinkedList.java:270)
       at org.apache.spark.unsafe.map.BytesToBytesMap$MapIterator.hasNext(BytesToBytesMap.java:310)
       at org.apache.spark.sql.execution.UnsafeFixedWidthAggregationMap$1.next(UnsafeFixedWidthAggregationMap.java:177)
       at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIterator.processNext(Unknown
Source)
       at org.apache.spark.sql.execution.BufferedRowIterator.hasNext(BufferedRowIterator.java:43)
       at org.apache.spark.sql.execution.WholeStageCodegenExec$$anonfun$8$$anon$1.hasNext(WholeStageCodegenExec.scala:377)
       at org.apache.spark.sql.Dataset$$anonfun$56$$anon$1.hasNext(Dataset.scala:2712)
       at scala.collection.Iterator$$anon$11.hasNext(Iterator.scala:408)
       at scala.collection.Iterator$$anon$11.hasNext(Iterator.scala:408)
       at scala.collection.Iterator$$anon$11.hasNext(Iterator.scala:408)
       at scala.collection.Iterator$$anon$13.hasNext(Iterator.scala:461)
       at org.apache.spark.util.collection.ExternalSorter.insertAll(ExternalSorter.scala:191)
       at org.apache.spark.shuffle.sort.SortShuffleWriter.write(SortShuffleWriter.scala:63)
       at org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:96)
       at org.apache.spark.scheduler.ShuffleMapTask.runTask(ShuffleMapTask.scala:53)
       at org.apache.spark.scheduler.Task.run(Task.scala:99)
      at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:322)
       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)

How can I disable wholestage code gen?

Thanks and Regards,
Tony

发件人: kathleen li <kathleenli168@gmail.com<mailto:kathleenli168@gmail.com>>
发送时间: 2018年10月8日 1:04
收件人: 阎志涛 <tony.yan@tendcloud.com<mailto:tony.yan@tendcloud.com>>
抄送: user@spark.apache.org<mailto:user@spark.apache.org>
主题: Re: Executor hang

It seemed you had data skewed issue since shuffle read size for executor 4 is almost 2 times
than other executors and GC time 11s almost 15 to 20 times than others.

Kathleen
Sent from my iPhone

On Oct 7, 2018, at 5:24 AM, 阎志涛 <tony.yan@tendcloud.com<mailto:tony.yan@tendcloud.com>>
wrote:
Hi, All,
I am running Spark 2.1 on Hadoop 2.7.2 with yarn. While executing spark tasks, some executor
keep running forever without success. From the following screenshot:
<image002.jpg>
We can see that executor 4 keep running for 26 minutes and the shuffle read size/records keep
unchanged for 26mins too.  Threaddump for the thread is as following:
<image004.jpg>

<image009.jpg>

The linux version is: Linux version 4.14.62-70.117.amzn2.x86_64 (mockbuild@ip-10-0-1-79) and
jdk version is Oracle JDK 1.8.0_181. With jstack on the machine, I can see following thread
dump:

"Executor task launch worker for task 3806" #54 daemon prio=5 os_prio=0 tid=0x0000000001230800
nid=0x1fc runnable [0x00007fba0e600000]
   java.lang.Thread.State: RUNNABLE
       at java.lang.StringCoding.encode(StringCoding.java:364)
       at java.lang.String.getBytes(String.java:941)
       at org.apache.spark.unsafe.types.UTF8String.fromString(UTF8String.java:109)
       at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIterator.processNext(Unknown
Source)
       at org.apache.spark.sql.execution.BufferedRowIterator.hasNext(BufferedRowIterator.java:43)
       at org.apache.spark.sql.execution.WholeStageCodegenExec$$anonfun$8$$anon$1.hasNext(WholeStageCodegenExec.scala:377)
       at org.apache.spark.sql.execution.datasources.FileFormatWriter$SingleDirectoryWriteTask.execute(FileFormatWriter.scala:243)
       at org.apache.spark.sql.execution.datasources.FileFormatWriter$$anonfun$org$apache$spark$sql$execution$datasources$FileFormatWriter$$executeTask$3.apply(FileFormatWriter.scala:190)
       at org.apache.spark.sql.execution.datasources.FileFormatWriter$$anonfun$org$apache$spark$sql$execution$datasources$FileFormatWriter$$executeTask$3.apply(FileFormatWriter.scala:188)
       at org.apache.spark.util.Utils$.tryWithSafeFinallyAndFailureCallbacks(Utils.scala:1341)
       at org.apache.spark.sql.execution.datasources.FileFormatWriter$.org$apache$spark$sql$execution$datasources$FileFormatWriter$$executeTask(FileFormatWriter.scala:193)
       at org.apache.spark.sql.execution.datasources.FileFormatWriter$$anonfun$write$1$$anonfun$3.apply(FileFormatWriter.scala:129)
       at org.apache.spark.sql.execution.datasources.FileFormatWriter$$anonfun$write$1$$anonfun$3.apply(FileFormatWriter.scala:128)
       at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:87)
       at org.apache.spark.scheduler.Task.run(Task.scala:99)
      at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:322)
       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)

I wonder why this happened? Is it related to my environment of a bug of Spark?

Thanks and Regards,
Tony

阎志涛
研发副总裁

M  + 86-139 1181 5695
Wechat   zhitao_yan

北京腾云天下科技有限公司
北京市东直门外大街39号院2号楼608室,100027

TalkingData.com<http://TalkingData.com>

Mime
View raw message