No subprocessId in Node with subprocess

[appendix]

Hi,

I'm facing a problem with a process definition which contains a task-node with a sub-process.
When deploying both processes into a clean database the definitions and nodes are persisted currectly, but when the node with the sub-process is instantiated during the workflow, an exception is thrown stating that a process with a NULL definition cannot be instantiated.

I had a look at the database and realized that the table jbpm_node has no entry in the column subprocessdefinition_ in the row with the sub-process task despite both processes are correctly deployed to jbpm_processdefinition.

I looked at the sources and saw that the process definitions are persisted in org.jbpm.db.GraphSession.deployProcessDefinition(). Here both process definitions are saved to the session and the the session. When the session is closed later on, the underlying hibernate cannot refer to the sub-process-id when populating the table jbpm_node.

If I execute a manual session.flush() after the sub-process definition is saved to the session, then the table jbpm_node is populated correctly with a valid subprocessid_ after all.

I'm using jbpm-3.1.1 with a postgres 8.1.4 and wonder if this is a known issue. I had a look at the 3.1.2 sources and saw that code for process definition deployment didn't change, so I assume a version upgrade wouldn't solve my problem.

Any suggestions on using task-nodes with sub-processes? How can i make sure that all sub-processes are refered in their corresponding task nodes correctly? I don't get any warnings or errors at deployment time and the exception is thrown at runtime.
I haven't included the process definitions in this post, because I'm positive that they are valid and they would only blow up this post.

Thanks for any help in advance,

Kurt

[koen.aers]

Kurt,

What exactly do you mean with task node with a subprocess?

Regards,
Koen

[appendix]

Hi Koen,

thank you for your reply and please accept my apologies for mixing up a term. It is indeed not a task node, which cannot hold a sub-process, but a process-state-node as exemplified by the following snippet.

<process-state name="subProcessStateNode">
 <sub-process name="childProcessName"></sub-process>
 <transition to="nextNode"></transition>
</process-state>

The child as well as the parent process are persisted to the database, but the column _subprocessdefinition in the table jbpm_node doesn't hold a valid entry for the node "subProcessStateNode".

Regards, Kurt

Regards, Kurt

[jainer]

Kurt:

You must deploy first the subprocess and later deploy the process, i did it and work it fine.

jainer e.

[appendix]

Hi Jainner,

I thought about that as well - and it would make sense, but I'm doing that already.
I'm using the ant deploy task:

<target name="deploy-process-definition">
 <taskdef name="deploypar" classpathref="ant.classpath" classname="org.jbpm.ant.DeployProcessTask">
 </taskdef>
 <deploypar cfg="resources/hibernate_ant.cfg.xml">
 <fileset id="processDefinition" dir="resources/processDefinition">
 <include name="subProcess.par" /> <!-- is referenced by the later -->
 <include name="superProcess.par" />
 </fileset>
 </deploypar>
</target>

Additionally I was using Seam build-in process deployment mechanism, but it lead to the same issues.

Is there a better way of deploying process descriptions without using ant?

Regards, Kurt