[rules-users] Determining the active nodeinstance for a process

[Kris Verlaenen]

There are several ways of doing this.

1. Using the audit log
You can log the execution events so they can be queried later.  The
drools-bam module contains a WorkingMemoryDbLogger that logs these
events to database, so you can easily query them.

2. Inspecting the process instance itself
You can get all active node instances from the process instance itself
by casting it to a WorkflowProcessInstance.  Note that this will only
contain info about the currently active node instances, no history
information, but it seems that's exactly what you're looking for.  The
Eclipse plugin and gwt-console also use this approach to show the active
nodes in a process instance during execution.

3. Custom event listener
If the above methods are insufficient, you can just implement your own
listener and extract the info you need.

Regarding reconnect, that simply reconnects the process instance to the
engine, nothing more.  The engine will then continue by sending the
right triggers to that restored process instance, for example an
incoming event, a work item that has been completed, a timer that has
been triggered, etc.  So the engine will restore and reconnect a process
instance if a trigger for that process instance comes in but the process
instance is no longer in memory.

Kris

Quoting Garrett Conaty <garrett at conaty.net>:

> 
> 
> I'm trying to determine 'where' a process is (not the
> active/completed/ 
> etc. state), but the wait state.
> 
> One way of doing this (and tracing the process execution) would be
> via  
> a RuleFlowEventListener; however is there a way to query a  
> ProcessInstance (or something?) for this information?
> 
> Effectively I want to know which EventBasedNodeInstances are
> triggered  
> for a process and waiting for a response.
> 
> Use case is to show in a user interface what the process is waiting 
> 
> for (i.e. loan approval, timeout, some confirmation).
> 
> 
> Furthermore, I am confused as to where a process resumes after being 
> 
> marshalled from db back into memory.  I can see that there is a  
> 'reconnect' method which gets called, but it looks like the  
> implementation of this calls the start Node again!?!
> 
> Thanks,
> Garrett




Disclaimer: http://www.kuleuven.be/cwis/email_disclaimer.htm