activemq-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gary Tully <gary.tu...@gmail.com>
Subject Re: Perhaps I'm just approaching this the wrong way
Date Thu, 16 Dec 2010 10:41:08 GMT
What you describe seems very reasonable. Would it be possible to
codify what you are experiencing in the form of a test case?
There is some good starting point code in some of the existing unit
tests, eg: http://svn.apache.org/viewvc/activemq/trunk/activemq-core/src/test/java/org/apache/activemq/network/FailoverStaticNetworkTest.java?view=markup

If you can produce a test case that captures what you need we easily
figure out what the problem is and commit the test case so the use
case is protected.



On 15 December 2010 17:32, logosdev <logosdev@yahoo.com> wrote:
>
> I'm very grateful for Dejan's input on the thread
> http://activemq.2283324.n4.nabble.com/broker-config-if-destination-may-be-down-td3086163.html
>
> However I feel like perhaps I'm hitting this issue because of some
> fundamental problem with my approach.
>
> What I am trying to do is actually easy to describe, especially when you
> don't take security or failover into account.
>
> 1. Say there are 2 servers, A and B.
>
> 2. There is only one type of message, and each server both sends and
> receives it.
>
> 3. Sometimes A or B is down, and sometimes one server is started and
> expected to run without the other ever starting.  No server/broker is
> _always_ up.
>
> 4. The sending process shouldn't care if the remote server is up (messages
> should cue up).
>
> 5. The receiving process shouldn't care if the remote server is up, it just
> waits for data.
>
> 6. It's okay if we lose messages if too many would cue up, etc.  The
> important thing is that we re-establish the message flow between A and B
> when they are both available.
>
>
> Currently I have one bridged broker on each server, and our sending and
> receiving processes communicate with it using a vm connection.  This setup
> works for most cases including transient network outages and even bouncing
> one of the servers, but it does NOT work if the remote server is unavailable
> at startup time*.
> (* possibly because I use a failover transport, which is probably required)
>
> What I'd love to hear is how other's might approach this problem?  Am I over
> or under engineering this?  Should I have 2 brokers on each server?
>
> --
> View this message in context: http://activemq.2283324.n4.nabble.com/Perhaps-I-m-just-approaching-this-the-wrong-way-tp3089461p3089461.html
> Sent from the ActiveMQ - User mailing list archive at Nabble.com.
>



-- 
http://blog.garytully.com
http://fusesource.com

Mime
View raw message