-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/9052/
-----------------------------------------------------------

(Updated Jan. 22, 2013, 4:34 a.m.)
Review request for Flume.
Summary (updated)
-----------------

Allow hdfs idle callback to clean up closed bucket writers
Description
-------

Stops close() from removing the idle callback and skips close in the idle callback if the writer is already closed.
Diffs
-----

  flume-ng-sinks/flume-hdfs-sink/src/main/java/org/apache/flume/sink/hdfs/BucketWriter.java 3f31ef2

Diff: https://reviews.apache.org/r/9052/diff/
Testing
-------

Tests pass. Since sfWriters is protected it is hard to verify the changed behavior in unit tests. I however verified the behaviour in the java debugger.
Thanks,

Juhani Connolly

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/9052/#review15560
-----------------------------------------------------------
Absence of both transactionCapacity and batchsize information would make things difficult for the spillable transaction [flume 1227].

if batch size is exposed to the transaction i am ok with dropping this transaction capacity in file and memory channels.

- Roshan Naik
On Jan. 22, 2013, 4:34 a.m., Juhani Connolly wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/9052/
> -----------------------------------------------------------
>
> (Updated Jan. 22, 2013, 4:34 a.m.)
>
>
> Review request for Flume.
>
>
> Description
> -------
>
> Stops close() from removing the idle callback and skips close in the idle callback if the writer is already closed.
>
>
> Diffs
> -----
>
>   flume-ng-sinks/flume-hdfs-sink/src/main/java/org/apache/flume/sink/hdfs/BucketWriter.java 3f31ef2
>
> Diff: https://reviews.apache.org/r/9052/diff/
>
>
> Testing
> -------
>
> Tests pass. Since sfWriters is protected it is hard to verify the changed behavior in unit tests. I however verified the behaviour in the java debugger.
>
>
> Thanks,
>
> Juhani Connolly
>
>

> On Jan. 22, 2013, 4:55 a.m., Roshan Naik wrote:
> > Absence of both transactionCapacity and batchsize information would make things difficult for the spillable transaction [flume 1227].
> >
> > if batch size is exposed to the transaction i am ok with dropping this transaction capacity in file and memory channels.

I misnamed this initially by accident. This review is just for a change in the behavior of idle callback. At this time I don't plan to work on 1860
- Juhani
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/9052/#review15560
-----------------------------------------------------------
On Jan. 22, 2013, 4:34 a.m., Juhani Connolly wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/9052/
> -----------------------------------------------------------
>
> (Updated Jan. 22, 2013, 4:34 a.m.)
>
>
> Review request for Flume.
>
>
> Description
> -------
>
> Stops close() from removing the idle callback and skips close in the idle callback if the writer is already closed.
>
>
> Diffs
> -----
>
>   flume-ng-sinks/flume-hdfs-sink/src/main/java/org/apache/flume/sink/hdfs/BucketWriter.java 3f31ef2
>
> Diff: https://reviews.apache.org/r/9052/diff/
>
>
> Testing
> -------
>
> Tests pass. Since sfWriters is protected it is hard to verify the changed behavior in unit tests. I however verified the behaviour in the java debugger.
>
>
> Thanks,
>
> Juhani Connolly
>
>