[
https://issues.apache.org/jira/browse/SPARK-15849?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15326391#comment-15326391
]
Steve Loughran commented on SPARK-15849:
----------------------------------------
This sill be an S3 problem, quite a fundamental one: S3 is not "a real
filesystem"
http://hadoop.apache.org/docs/current/hadoop-aws/tools/hadoop-aws/index.html
http://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-common/filesystem/introduction.html
What happen, even with speculation= false is that there can be a lag between a
GET on a path (e.g. getFileStatus()) and an s3 listObjects() operation, which
hits the index. There's also a possibility that a negative GET, e.g. GET->404
is cached even if a PUT stuck something up there. This more like a get/list
inconsistency here, given it's {{listStatus()}} at fault, which is listing a
directory
# It'd be useful to know which s3 infra you were working with (US-east?
ireland? etc); Hadoop version
# If you are using Hadoop 2.7+, switch to s3a. The consistency problem still
exists, but that's the maintained FS where future fixes would surface
# do a test run using S3a with {{org.apache.hadoop.mapreduce.lib.output}} and
{{org.apache.hadoop.fs.s3a}} logging set to DEBUG. That should give some more
insight into what went wrong, which could be useful to us.
> FileNotFoundException on _temporary while doing saveAsTable to S3
> -----------------------------------------------------------------
>
> Key: SPARK-15849
> URL: https://issues.apache.org/jira/browse/SPARK-15849
> Project: Spark
> Issue Type: Bug
> Components: Spark Core
> Affects Versions: 1.6.1
> Environment: AWS EC2 with spark on yarn and s3 storage
> Reporter: Sandeep
>
> When submitting spark jobs to yarn cluster, I occasionally see these error
> messages while doing saveAsTable. I have tried doing this with
> spark.speculation=false, and get the same error. These errors are similar to
> SPARK-2984, but my jobs are writing to S3(s3n) :
> Caused by: java.io.FileNotFoundException: File
> s3n://xxxxxxx/_temporary/0/task_201606080516_0004_m_000079 does not exist.
> at
> org.apache.hadoop.fs.s3native.NativeS3FileSystem.listStatus(NativeS3FileSystem.java:506)
> at
> org.apache.hadoop.mapreduce.lib.output.FileOutputCommitter.mergePaths(FileOutputCommitter.java:360)
> at
> org.apache.hadoop.mapreduce.lib.output.FileOutputCommitter.commitJob(FileOutputCommitter.java:310)
> at
> org.apache.parquet.hadoop.ParquetOutputCommitter.commitJob(ParquetOutputCommitter.java:46)
> at
> org.apache.spark.sql.execution.datasources.BaseWriterContainer.commitJob(WriterContainer.scala:230)
> at
> org.apache.spark.sql.execution.datasources.InsertIntoHadoopFsRelation$$anonfun$run$1.apply$mcV$sp(InsertIntoHadoopFsRelation.scala:151)
> ... 42 more
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
