Hi

It doesn't seem possible to configure the log4j appender to use sink groups
without running a flume agent on the node producing the log output. We need
to support high availability of the logging system and my understanding is
this is accomplished using sink groups. Am I missing something in how you
would configure this?

Thanks

Cameron Gandevia

That can also be accomplished with the FailoverRPCClient.

https://github.com/apache/flume/tree/trunk/flume-ng-sdk/src/main/java/org/apache/flume/api

It looks to me like we need extend the log4j appender to support
FailoverRPCClient.

There is also the log4j2 flume appender which is basically an embedded
agent:

http://logging.apache.org/log4j/2.x/manual/appenders.html#FlumeAppender

Brock

On Mon, Nov 26, 2012 at 9:16 PM, Cameron Gandevia <[EMAIL PROTECTED]>wrote:

> Hi
>
> It doesn't seem possible to configure the log4j appender to use sink
> groups without running a flume agent on the node producing the log output.
> We need to support high availability of the logging system and my
> understanding is this is accomplished using sink groups. Am I missing
> something in how you would configure this?
>
> Thanks
>
> Cameron Gandevia
>

--
Apache MRUnit - Unit testing MapReduce - http://incubator.apache.org/mrunit/