Good day Camunda,
Sometimes a process does not start that has a (global) message event as a start event.
Only when I have deleted the BPMN model and deployed it again does it work again.
The problem occurs irregularly.
Many thanks
Frank
Hey @camfomd ,
I assume you start the process via REST API? Do you get an error response if a process instance is not started?
This sounds like an odd problem.
Hallo Nele,
I suspect the problem is with the Tomcat ConnectionTimeout values and have now increased this to 600000 (600 s).
If the error occurs again, I will contact you again.
Many thanks
Frank
Hola Nele,
Crece la sospecha de que la causa hay que buscarla en el despliegue.
En un modelo BPMN, hay tres grupos que interactúan entre sà a través de mensajes.
Un flujo de trabajo C llama al flujo de trabajo A y luego el proceso en el flujo de trabajo A se detiene cuando se supone que debe iniciar el flujo de trabajo B a través de un evento de mensaje.
A continuaciĂłn, despliego de nuevo los tres grupos y el flujo de trabajo A inicia el flujo de trabajo B.
ÂżExisten dependencias con mensajes globales que se resuelven durante el despliegue?
Con saludos cordiales
Frank
Hello Nele,
The suspicion is growing that the cause must be sought in the deployment.
In a BPMN model, there are three pools that interact with each other via messages.
A workflow C calls workflow A and then the process in workflow A stops when it is supposed to start workflow B via a message event.
I then deploy all three pools again and workflow A starts workflow B.
Are there any dependencies with global messages that are resolved during deployment?
With kind regards
Frank
Hey Frank,
gracias por la traducciĂłn. A mi me gusta mucho practica mi español pero creo que aquĂ la mayorĂa de la gente entienden ingles 
I hope I understand it correctly. You have one BPMN process model with 3 pools interacting with messages? You want all three pools to be executable?
It is best to create three separate BPMN files for each workflow (A,B,C) and then use the message name and some correlation key to make sure the message triggers one of the other flows.
If that is not helping could you share your model?
Cheers
Nele
1 Like
Hello Nele,
Thank you very much for your answer.
All workflows are executable and startable.
I guess I will have to do it that way, but I liked the clear layout of the message flows.
Frank
Hey Frank,
I agree having the collaboration diagram (bpmn with multiple processes (pools) and message flows) is nice to understand the overall functionality and dependencies between processes. The engine takes care of single processes.
Still having the collaboration diagram somewhere saved and stored might be nice for documentation and explaining people the functionalities.
I don’t know your processes and how they are related but maybe you could also have a look into Call Activities. That way in Cockpit you can drill down from the parent process to the child processes and you would have the link of both processes visible in the engine as well. But as I said it might not work for your logic.
Kind regards
Nele
Hello Nele,
yes, I agree with that. I do the same elsewhere. But I am trying to implement the Process Driven Approach fully. It works, but sometimes it doesn’t and I would like to find out the cause.
Best regards
Frank
Hey Nele,
In a collaboration model, for example, there are three pools, each with at least one workflow.
Process A starts process B with a message and process B starts process C with a message.
The send message events are asynchronous before and use the http connector or a Java delegate implementation.
Process C does not start, but if the model is deployed again, then process C is started.
The behaviour is reproducible.
Unfortunately I cannot provide you with the models via the forum, but I consider this behaviour to be an actual failure.
With kind regards
Frank
Hey @camfomd ,
have you deployed all processes in single bpmn files to the workflow engine? Or are all three diagrams in a single bpmn file?
Kind regards
Nele
all three diagrams in a single bpmn file
Hey @camfomd ,
this will not work. You need to deploy the diagrams independently. So you need to create for every pool a single bpmn file, deploy them to the engine and then your correlation should work.
The engine does not execute collaboration diagrams. A collaboration diagram is not supposed to be executed. It can be used for documentation.
Kind regards
Nele
Hey Nele,
Many thanks for the information
With kind regards
Frank
1 Like