[ 
https://issues.apache.org/jira/browse/FLINK-6427?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15992671#comment-15992671
 ] 

ASF GitHub Bot commented on FLINK-6427:
---------------------------------------

GitHub user aljoscha opened a pull request:

    https://github.com/apache/flink/pull/3807

    [FLINK-6427] Ensure file length is flushed in StreamWriterBase

    StreamWriterBase is used with BucketingSink. With HDFS, it can happen that 
the
    NameNode does not update the file length if we flush but don't properly 
close
    the file. We now specify that we also want to flush the file length when
    flushing.

You can merge this pull request into a Git repository by running:

    $ git pull https://github.com/aljoscha/flink 
bucketing-sink-flush-file-length

Alternatively you can review and apply these changes as the patch at:

    https://github.com/apache/flink/pull/3807.patch

To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:

    This closes #3807
    
----
commit dbc3899d5dbda642fdce0f06e9fece43dbc9585f
Author: Aljoscha Krettek <[email protected]>
Date:   2017-05-02T09:31:40Z

    [FLINK-6427] Ensure file length is flushed in StreamWriterBase
    
    StreamWriterBase is used with BucketingSink. With HDFS, it can happen that 
the
    NameNode does not update the file length if we flush but don't properly 
close
    the file. We now specify that we also want to flush the file length when
    flushing.

----


> BucketingSink does not sync file length in case of cancel
> ---------------------------------------------------------
>
>                 Key: FLINK-6427
>                 URL: https://issues.apache.org/jira/browse/FLINK-6427
>             Project: Flink
>          Issue Type: Bug
>          Components: Streaming Connectors
>    Affects Versions: 1.3.0, 1.2.1
>            Reporter: Aljoscha Krettek
>            Assignee: Aljoscha Krettek
>
> This is from a discussion on the user mailing lists: 
> https://lists.apache.org/thread.html/917dbbbd7c8f48b33e7b470fa9ed382df61cf43cb5deb46becebf4b5@%3Cuser.flink.apache.org%3E
> A possible solution is to add this to {{StreamWriterBase.hflushOrSync()}}:
> {code}
> if (os.getWrappedStream() instanceof DFSOutputStream) {
>     ((DFSOutputStream) 
> os.getWrappedStream()).hsync(EnumSet.of(HdfsDataOutputStream.SyncFlag.UPDATE_LENGTH));
> }
> {code}



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Reply via email to