Upon looking into this issue, it appears the problem actually stems from the way persistence is used. When running stateless inside Eclipse, the issue is when the kcontext.getKnowledgeRuntime() from within the action node is returning a StatefulKnowledgeSessionImpl the same as what is being used by the other process signaling the event. While in jbpm-console, the processes are started using an instance of CommandBasedStatefulKnowledgeSession and then while trying to fire from one process to the other is still using an instance of StatefulKnowledgeSessionImpl and they aren't finding each other.
It does seem like a bug that when the process is started using a CommandBasedStatefulKnowledgeSession that the kcontext should return the same instance of the StatefulKnowledgeSession implementation instead of returning a separate StatefulKnowledgeSessionImpl. Can anyone comment on that or give some insights?
Thanks,
- Darin