I was informed about this issue last week by an user, but nobody has opened a JIRA yet.. do you want to have the honors? :)
When I first wrote the refactoring on AMQPMessages right before 2.0.0 release, I was expecting users to use qpid-dispatch to load balance, and not the Message Bridge from Artemis. That message bridge is meant for core messages, and right now an AMQPMessage would be converted to Core Message and you would receive it as core message on the other side. My plan is to fix it in about 2 weeks given my current workload. The plan for the fix is to embed the AMQPMessage in the body of the core message over the bridge. You can still use qpid-dispatch though for the load balancing options... this will be fixed on 2.3.0 (2.2.0 that's due soon won't have this fix I'm still working on). On Tue, Jul 18, 2017 at 7:00 PM, Dan Langford <danlangf...@gmail.com> wrote: > I could use some help. We set up an Artemis cluster (2.1.0) and at first > glance it was looking great using the command line tools to produce and > consumer messages verified that they were getting distributed through the > cluster just fine. When we switched over to using AMQP1.0 via the > qpid-jms-client library we started getting some problems. We think the > pattern we have observed is as soon as the cluster tries to send an AMQP > message to another node in the cluster an exception is thrown server side. > > I built on top of the example `clustered-static-discovery` to > demonstrate/reproduce this behavior. > https://github.com/danlangford/tq-artemis-example when you run `mvn`verify > it first executes the stock `clustered-static-discovery` example > successfully and then it moves into my test using QPID client libs (thus > AMQP) and you can see the exception that the server > throws. > [org.apache.activemq.artemis.protocol.amqp.proton.ProtonServerReceiverContext] > null: java.lang.UnsupportedOperationException > > Does anybody have any insight into producing messages into a cluster via > AMQP1.0? Maybe i need to tweak my broker configuration? Maybe I am > misunderstanding something? > > for the record when not clustered the QPID libs / AMQP1.0 is all working > great against the artemis broker. > > Thanks your your time and consideration. > > Dan Langford -- Clebert Suconic
