We have been using Seam with jBPM and are very happy with the results, but have something going on which is causing a performance issue.
In the case where we are working a task that then gets transitioned to another task state we load the screen which is used to resolve that new task. These activities all take place in the same conversation.
When the request is complete Seam cleans up the jbpmContext via the ManagedJbpmContext class. When the jbpmContext is closed, we get hundreds of these messages:
2008-04-29 13:27:30-DEBUG [http-127.0.0.1-8180-1] org.jbpm.svc.Services(156) executing default save operations
2008-04-29 13:27:30-DEBUG [http-127.0.0.1-8180-1] org.jbpm.svc.save.HibernateSaveOperation(37) saving process instance
2008-04-29 13:27:30-DEBUG [http-127.0.0.1-8180-1] org.jbpm.svc.save.SaveLogsOperation(43) flushing logs to logging service.
2008-04-29 13:27:30-DEBUG [http-127.0.0.1-8180-1] org.jbpm.svc.save.CascadeSaveOperation(39) cascading save of 'org.jbpm.graph.exe.ProcessInstance@bd48ef'
This takes a lot of time to do this and it turns what should be a 1 second request into 10 seconds. It almost looks like it is running a bunch of visitors to collect logging information for the different parts of the workflow, but I'm not sure.
Any help would be appreciated.
By any chance have your turned on the DEBU in log4j. If so you can turn it off