I just tried to implement with core api the request-reply pattern:
I have a server, a callee (something like a service) and a caller (will send messages to the callee and wait for answer.
It looks like everything works but after some time I get outofmemory - the profiler shows me that the messages are not collected - see attached.
It's interesting that OOM comes in Server so I ran Server with -Xmx64m -XX:+HeapDumpOnOutOfMemoryError -XX:HeapDumpPath=/tmp to force the OOM earlier and to see the dump.
I guess I'm doing something wrong - could someone give me a hint what has to be changed here?
I know creating temp queue for every request is an antipattern and I'm doing this only in this example - the messages are still there also if I don't create a temp queue for every request.