[
https://issues.apache.org/jira/browse/FLUME-1252?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13294218#comment-13294218
]
Hari Shreedharan edited comment on FLUME-1252 at 6/13/12 7:09 AM:
------------------------------------------------------------------
Juhani,
I looked at the log. The reason the test is failing is because of
ConnectException on the Zookeeper port that is causing the ZK minicluster to
fail to startup, not due to the Hadoop jars - seems like I need to make a fix
in the test(it is interesting that the test does not fail on my setup). I
actually prefer using Minicluster for ZK/Hbase/Hadoop tests because that
assures the functionality and is not time consuming to write up. Since the
number of tests is becoming larger, we could add profiles to support all tests,
or a subset of tests.
* This is the same behavior as the current Hbase and HDFS sinks. Since we do
not guarantee that there will be no duplication, I think it is ok. The
SimpleAsyncHbaseSink is only an example, not really something that I'd expect
to be used in production.
* I agree the configuration is not documented, I will do this in the next patch
- for one, this was because the config was almost exactly the same has
HbaseSink. Nevertheless, I will update it.
* Agreed, will update in the next patch.
If you have any more concerns, please let me know. I will submit an updated
patch if you dont have any other concerns.
was (Author: hshreedharan):
Juhani,
I looked at the log. The reason the test is failing is because of
ConnectException on the Zookeeper port that is causing the ZK minicluster to
fail to startup, not due to the Hadoop jars - seems like I need to make a fix
in the test(it is interesting that the test does not fail on my setup). I
actually prefer using Minicluster for ZK/Hbase/Hadoop tests because that
assures the functionality and is not time consuming to write up. Since the
number of tests is becoming larger, we could add profiles to support all tests,
or a subset of tests.
* This is the same behavior as the current Hbase and HDFS sinks. Since we do
not guarantee that there will be no duplication, I think it is ok. The
SimpleAsyncHbaseSink is only an example, not really something that I'd expect
to be used in production.
* I agree the configuration is not documented, I will do this in the next patch
- for one, this was because the config was almost exactly the same has
HbaseSink. Nevertheless, I will update it.
* Agreed, will update in the next patch.
> Asynchronous HBase Sink
> -----------------------
>
> Key: FLUME-1252
> URL: https://issues.apache.org/jira/browse/FLUME-1252
> Project: Flume
> Issue Type: New Feature
> Components: Sinks+Sources
> Reporter: Otis Gospodnetic
> Assignee: Hari Shreedharan
> Fix For: v1.2.0
>
> Attachments: FLUME-1252.patch, async-err.log
>
>
> The current HBase sink is synchronous/blocking. Asynchbase HBase client
> outperforms HTable. It would be nice to have an HBase sink that uses
> asynchbase.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
