tags:

views:

127

answers:

1

I'm having trouble setting up multicast transport using ActiveMQ. I noticed with version 5.2 there is a bug that prevents it from even building the correct factory (fixed in 5.3). The fact this bug exists makes me wonder if anyone is using multicast at all. How can this be tested when it only recently has been enabled?

Just knowing that this is possible would help me a lot. Thanks.

EDIT:

It turns out i was hanging on building the connection. The setClientID is a synchronous call, so the client sits and waits indefinitely. This is a landmine for anyone using multicast, there are many calls in the AMQConnection object that call ensureConnectionInfoSent() that blocks until it receives a response. This is just incompatible with multicast.

+1  A: 

I have seen the multicast transport work in the past, in a development rather than production context. It seemed a bit flaky, though, and didn't give me the confidence to trust it where there's money on the line.

IMO, ActiveMQ is ridiculously buggy. The most basic bugs seem to creep into the codebase on a regular basis, and it's very hard to have confidence that they can do the difficult things when the basics go wrong so often.

skaffman
Well, I probably won't be able to move away from AMQ. We having been using it successfully in prod for a while. I'm leaning toward this being a network setup issue.
reccles
Seems likely, the mechanism has been in ActiveMQ for a while now, long before 5.2
skaffman
The bug was probably introduced around 5.2. Well, that's good enough for me. Narrows the problem down a bit. Thanks.
reccles