[ 
https://issues.apache.org/jira/browse/FLINK-4228?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16245767#comment-16245767
 ] 

ASF GitHub Bot commented on FLINK-4228:
---------------------------------------

Github user tillrohrmann commented on a diff in the pull request:

    https://github.com/apache/flink/pull/4939#discussion_r149980302
  
    --- Diff: flink-yarn/pom.xml ---
    @@ -153,6 +159,63 @@ under the License.
                                </plugins>
                        </build>
                </profile>
    +
    +           <profile>
    +                   <!-- Hadoop >= 2.6 moved the S3 file systems from 
hadoop-common into hadoop-aws artifact
    +                           (see 
https://issues.apache.org/jira/browse/HADOOP-11074)
    +                           We can add the (test) dependency per default 
once 2.6 is the minimum required version.
    +                   -->
    +                   <id>include_hadoop_aws</id>
    +                   <activation>
    +                           <property>
    +                                   <name>include_hadoop_aws</name>
    +                           </property>
    +                   </activation>
    +                   <dependencies>
    +                           <!-- for the S3 tests of 
YarnFileStageTestS3ITCase -->
    +                           <dependency>
    +                                   <groupId>org.apache.hadoop</groupId>
    +                                   <artifactId>hadoop-aws</artifactId>
    +                                   <version>${hadoop.version}</version>
    +                                   <scope>test</scope>
    +                                   <exclusions>
    +                                           <exclusion>
    +                                                   
<groupId>org.apache.avro</groupId>
    +                                                   
<artifactId>avro</artifactId>
    +                                           </exclusion>
    +                                           <!-- The aws-java-sdk-core 
requires jackson 2.6, but
    +                                                   hadoop pulls in 2.3 -->
    +                                           <exclusion>
    +                                                   
<groupId>com.fasterxml.jackson.core</groupId>
    +                                                   
<artifactId>jackson-annotations</artifactId>
    +                                           </exclusion>
    +                                           <exclusion>
    +                                                   
<groupId>com.fasterxml.jackson.core</groupId>
    +                                                   
<artifactId>jackson-core</artifactId>
    +                                           </exclusion>
    +                                           <exclusion>
    +                                                   
<groupId>com.fasterxml.jackson.core</groupId>
    +                                                   
<artifactId>jackson-databind</artifactId>
    +                                           </exclusion>
    --- End diff --
    
    Can't we enforce jackson 2.6 via dependency management? I think this would 
be cleaner than excluding the dependencies here and assume that 
`aws-java-sdk-s3` pulls in the missing dependencies.


> YARN artifact upload does not work with S3AFileSystem
> -----------------------------------------------------
>
>                 Key: FLINK-4228
>                 URL: https://issues.apache.org/jira/browse/FLINK-4228
>             Project: Flink
>          Issue Type: Bug
>          Components: State Backends, Checkpointing
>            Reporter: Ufuk Celebi
>            Assignee: Nico Kruber
>            Priority: Blocker
>             Fix For: 1.4.0
>
>
> The issue now is exclusive to running on YARN with s3a:// as your configured 
> FileSystem. If so, the Flink session will fail on staging itself because it 
> tries to copy the flink/lib directory to S3 and the S3aFileSystem does not 
> support recursive copy.
> h2. Old Issue
> Using the {{RocksDBStateBackend}} with semi-async snapshots (current default) 
> leads to an Exception when uploading the snapshot to S3 when using the 
> {{S3AFileSystem}}.
> {code}
> AsynchronousException{com.amazonaws.AmazonClientException: Unable to 
> calculate MD5 hash: 
> /var/folders/_c/5tc5q5q55qjcjtqwlwvwd1m00000gn/T/flink-io-5640e9f1-3ea4-4a0f-b4d9-3ce9fbd98d8a/7c6e745df2dddc6eb70def1240779e44/StreamFlatMap_3_0/dummy_state/47daaf2a-150c-4208-aa4b-409927e9e5b7/local-chk-2886
>  (Is a directory)}
>       at 
> org.apache.flink.streaming.runtime.tasks.StreamTask$AsyncCheckpointThread.run(StreamTask.java:870)
> Caused by: com.amazonaws.AmazonClientException: Unable to calculate MD5 hash: 
> /var/folders/_c/5tc5q5q55qjcjtqwlwvwd1m00000gn/T/flink-io-5640e9f1-3ea4-4a0f-b4d9-3ce9fbd98d8a/7c6e745df2dddc6eb70def1240779e44/StreamFlatMap_3_0/dummy_state/47daaf2a-150c-4208-aa4b-409927e9e5b7/local-chk-2886
>  (Is a directory)
>       at 
> com.amazonaws.services.s3.AmazonS3Client.putObject(AmazonS3Client.java:1298)
>       at 
> com.amazonaws.services.s3.transfer.internal.UploadCallable.uploadInOneChunk(UploadCallable.java:108)
>       at 
> com.amazonaws.services.s3.transfer.internal.UploadCallable.call(UploadCallable.java:100)
>       at 
> com.amazonaws.services.s3.transfer.internal.UploadMonitor.upload(UploadMonitor.java:192)
>       at 
> com.amazonaws.services.s3.transfer.internal.UploadMonitor.call(UploadMonitor.java:150)
>       at 
> com.amazonaws.services.s3.transfer.internal.UploadMonitor.call(UploadMonitor.java:50)
>       at java.util.concurrent.FutureTask.run(FutureTask.java:266)
>       at 
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
>       at 
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
>       at java.lang.Thread.run(Thread.java:745)
> Caused by: java.io.FileNotFoundException: 
> /var/folders/_c/5tc5q5q55qjcjtqwlwvwd1m00000gn/T/flink-io-5640e9f1-3ea4-4a0f-b4d9-3ce9fbd98d8a/7c6e745df2dddc6eb70def1240779e44/StreamFlatMap_3_0/dummy_state/47daaf2a-150c-4208-aa4b-409927e9e5b7/local-chk-2886
>  (Is a directory)
>       at java.io.FileInputStream.open0(Native Method)
>       at java.io.FileInputStream.open(FileInputStream.java:195)
>       at java.io.FileInputStream.<init>(FileInputStream.java:138)
>       at 
> com.amazonaws.services.s3.AmazonS3Client.putObject(AmazonS3Client.java:1294)
>       ... 9 more
> {code}
> Running with S3NFileSystem, the error does not occur. The problem might be 
> due to {{HDFSCopyToLocal}} assuming that sub-folders are going to be created 
> automatically. We might need to manually create folders and copy only actual 
> files for {{S3AFileSystem}}. More investigation is required.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Reply via email to