the problem happened a second time and i can rule out problems on the DB side.
other ideas:
- we have some plugins for cockpits. they are doing read only request to the DB. maybe it is possible to have some kind of locking here? seems very unlikely
- another forum posts mentions problems with long running tasks?! Optimistic Locking Exception
- another forum posts has smae problems too with duplicate executions, but got never an answer: Asynchronous before + Exclusive true in cluster problem
in the best practice article it is suggested to flag all service tasks with “async before”: https://camunda.com/best-practices/dealing-with-problems-and-exceptions/#additional-save-points
maybe i try this and hope that it helps…
the only other solution would be to write a own implementation: Every service task executed is stored in a separate DB and checked if a double execution happens. but this sounds like a lot of overhead for something that the workflow engine should do for me?!