[rules-users] Drools Flow 5.1.0.M1 - Human Task - How to get the WorkItem change state after restart of MinaTaskServer

[Kris Verlaenen]

Vijay,

The current implementation of the communication between the human task
service and the engine is indeed not error-safe.  We are currently
working on making a more stable communication foundation (where
different communication technologies could be used) to solve this issue.

Kris

Quoting Vijay K Pandey <VPandey at mdes.ms.gov>:

> Hi,
> 
> I have the following simple process having the WSHT 'HumanTask'
> (Drools Flow 5.1.0.M1)
> 
> Start Node--> Human Task --> End Node
> 
> Under the following scenario:
> 
> a)       MinaTaskServer service is started
> b)       Drools Flow Process is started - as it reaches the Human
> Task ( having 'Wait for completion' as true) - following 4 commands
> are executed on the mina service
>                                                                i.    
>  AddTaskRequest ---> This basically creates the task in the WSHT task
> tables
>                                                              ii.     
> RegisterForEventRequest --> This is called 3 times to register for
> following task events (TaskCompletedEvent, TaskFailedEvent,
> TaskSkippedEvent)
> c)       I guess the above 3 events (b ii) are registered for a
> callback to the Drools process so that the 'work item' (related to
> Human Task) state can be properly changed when the above 3 task event
> occurs - and the drools process can move forward.
> 
> Now my question is what if the Mina service has to be stopped in
> between (task has been created but not yet reached the completed
> status) - then we lose all the registered events (as these are not
> persistent events) - now when we restart the Mina service we query
> the task - and let's say we changed the state through "Reserved-->
> InProgress-->Completed".  Now since the registered events are lost,
> how does the work item will be marked as "completed" (as the
> associated WSHT task has completed) so that the drools process can
> move forward - basically end in this case?
> 
> Any insight will be highly appreciated.
> 
> Thanks
> Vijay
> 
> 




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