What are the steps to reproduce your problem?
- create a process definition producing an incident
- deploy it
- create some process instances of it
- fix the behavior and deploy again
- run a migration and check "asynchronously" to do it in batch mode
What is the problem?
- the job created above gets stuck
- if "asynchronously" is not checked the migration runs fine.
The behavior is reproducible for at least 7.9 and 7.10. In 7.11.0-alpha3 it seems to be fixed.
What would be the expected behavior:
The migration job finishes eventually or ends failing.