[
https://issues.apache.org/jira/browse/HADOOP-17063?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17132746#comment-17132746
]
Dyno commented on HADOOP-17063:
-------------------------------
> important: you are using the classic FileOutputCommitter; this is slow and
> unsafe on s3. Can you move to the S3a zero rename committer?
{noformat}
fs.s3a.committer.name Committer
directory directory staging committer
partitioned partition staging committer (for use in Spark only)
magic the “magic” committer
file the original and unsafe File committer; (default)
{noformat}
our setup is kubernetes/ spark 2.4.4 /hadoop-3.2.1
according to
https://hadoop.apache.org/docs/r3.2.1/hadoop-aws/tools/hadoop-aws/committers.html
https://github.com/aws-samples/eks-spark-benchmark/blob/master/performance/s3.md
directory/partitioned needs shared storage, magic is only supported in spark
3.0
so i think the only option for us is file. when you say "S3a zero rename
committer" you mean the directory one?
> S3ABlockOutputStream.putObject looks stuck and never timeout
> ------------------------------------------------------------
>
> Key: HADOOP-17063
> URL: https://issues.apache.org/jira/browse/HADOOP-17063
> Project: Hadoop Common
> Issue Type: Sub-task
> Affects Versions: 3.2.1
> Environment: hadoop 3.2.1
> spark 2.4.4
>
> Reporter: Dyno
> Priority: Minor
> Attachments: jstack_exec-34.log, jstack_exec-40.log,
> jstack_exec-74.log
>
>
> {code}
> sun.misc.Unsafe.park(Native Method)
> java.util.concurrent.locks.LockSupport.park(LockSupport.java:175)
> com.google.common.util.concurrent.AbstractFuture.get(AbstractFuture.java:523)
> com.google.common.util.concurrent.FluentFuture$TrustedFuture.get(FluentFuture.java:82)
>
> org.apache.hadoop.fs.s3a.S3ABlockOutputStream.putObject(S3ABlockOutputStream.java:446)
>
> org.apache.hadoop.fs.s3a.S3ABlockOutputStream.close(S3ABlockOutputStream.java:365)
>
> org.apache.hadoop.fs.FSDataOutputStream$PositionCache.close(FSDataOutputStream.java:72)
> org.apache.hadoop.fs.FSDataOutputStream.close(FSDataOutputStream.java:101)
> org.apache.parquet.hadoop.util.HadoopPositionOutputStream.close(HadoopPositionOutputStream.java:64)
> org.apache.parquet.hadoop.ParquetFileWriter.end(ParquetFileWriter.java:685)
> org.apache.parquet.hadoop.InternalParquetRecordWriter.close(InternalParquetRecordWriter.java:122)
>
> org.apache.parquet.hadoop.ParquetRecordWriter.close(ParquetRecordWriter.java:165)
>
> org.apache.spark.sql.execution.datasources.parquet.ParquetOutputWriter.close(ParquetOutputWriter.scala:42)
>
> org.apache.spark.sql.execution.datasources.FileFormatDataWriter.releaseResources(FileFormatDataWriter.scala:57)
>
> org.apache.spark.sql.execution.datasources.FileFormatDataWriter.commit(FileFormatDataWriter.scala:74)
>
> org.apache.spark.sql.execution.datasources.FileFormatWriter$$anonfun$org$apache$spark$sql$execution$datasources$FileFormatWriter$$executeTask$3.apply(FileFormatWriter.scala:247)
>
> org.apache.spark.sql.execution.datasources.FileFormatWriter$$anonfun$org$apache$spark$sql$execution$datasources$FileFormatWriter$$executeTask$3.apply(FileFormatWriter.scala:242)
>
> org.apache.spark.util.Utils$.tryWithSafeFinallyAndFailureCallbacks(Utils.scala:1394)
>
> org.apache.spark.sql.execution.datasources.FileFormatWriter$.org$apache$spark$sql$execution$datasources$FileFormatWriter$$executeTask(FileFormatWriter.scala:248)
>
> org.apache.spark.sql.execution.datasources.FileFormatWriter$$anonfun$write$1.apply(FileFormatWriter.scala:170)
>
> org.apache.spark.sql.execution.datasources.FileFormatWriter$$anonfun$write$1.apply(FileFormatWriter.scala:169)
> org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:90)
> org.apache.spark.scheduler.Task.run(Task.scala:123)
> org.apache.spark.executor.Executor$TaskRunner$$anonfun$10.apply(Executor.scala:408)
> org.apache.spark.util.Utils$.tryWithSafeFinally(Utils.scala:1360)
> org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:414)
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
>
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
> java.lang.Thread.run(Thread.java:748)
> {code}
>
> we are using spark 2.4.4 with hadoop 3.2.1 on kubernetes/spark-operator,
> sometimes we see this hang with the stacktrace above. it looks like the
> putObject never return, we have to kill the executor to make the job move
> forward.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: common-issues-h...@hadoop.apache.org
