Hi all,
I need to migrate fromjBPM 4.1 to jBPM 5. I'm trying to undstand how process instance recovery, and especially timer recovery works, but I'm completely confused.
I do have a server application which starts and processes workflows in parallel. Workflow state and history are persisted in an Oracle database.
The workflow execution moves from one stable state to the next stable state, which can be a timer node.
When I'm shutting down my application and restarting it, i need to build up the knowledge base in jbpm 5 as one of the first actions.
My assumption now is that I have to do recover each process instance separately by calling
JPAKnowledgedSession.loadStatefulKnowledgeSession();
At least I have to do this for all process instances which wait for a timer event to happen.
Is this really correct ? I cannot believe that this is really the case. This means that I cannot parallize my applications anymore, but I have to distribute the differnt process instances among them.
Can someone help me out ?
Elisabeth