Hello,

Maybe the user forum is not the recommended place for this discussion: but
besides the blueprint xml discussion; there's another "design" question that
I'm not sure of how it is seen by the AMQ devs at this moment:

I'll be making a few assumptions since I'm not very experienced with
ActiveMQ or Karaf or Blueprint:
1) The default contract of ConnectionFactory is that it can throw exceptions
if the JMS broker is not immediately available for some reason (that's at
least the behavior I'm seeing).
2) The creation of the instances of ActiveMQConnectionFactory &
PooledConnectionFactory is in my example done with Blueprint, but it could
as well be written POJO classes, in any OSGi or JEE container. So the design
is very all-purpose.

The solutions we were currently discussing are all based on the Broker and
the Client being in the same OSGI/Spring/Blueprint context, but that may not
always be true. Wouldn't it be useful to have an option in the
(Pooled)ActiveMQConnectionFactory that there's a blocking start-method that
blocks until the connection with the broker(url) is established. This would
allow the "default" behavior of OSGI to be used:
The (pooled)connectionfactory is offered as a Service, and giving it a
blocking start/init-method that retries until success or some configurable
limit is reached (time/attempts/...) would delay the start of all bundles
consuming the javax.jms.ConnectionFactory Service until the broker is really
available (whether this in in the same runtime or anywhere else on the
network).

I can implement this try/retry strategy in the init-method of all my
blueprint bundles; but I think it would be a good addition to have this
functionality in AMQ.

Kind regards,

Tom Mercelis



--
View this message in context: 
http://activemq.2283324.n4.nabble.com/Connection-Refused-from-ActiveMQConnectionFactory-tp4664195p4664390.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.

Reply via email to