we are still on Flowable 5.23. When having heavy load in production, we see from time to time, that a Service-Task (lets say my-activity-1) is executed twice.
The workflow gets signalled and reaches another ReceiveTask with a new JobEntity. But the former JobEntity is still in the Database and gets executed exactly 10 Seconds later - after another JobEntity
Hard to say without more information about the process model. Is the service task async? If so, seeing it twice might mean you have an optimistic lock later on in the flow, and one of the two has been rolled back (but the logging has already happened).
That setting changes the polling time of looking for new jobs.
Can you share (a snippet) of your process model here? Also, have you tried using the latest version - many changes have gone in the async executor since.
After observing the behaviour with our productive bpmn, we could also reproduce it with a simple demo bpmn, having 2 service tasks and 1 receive task. The start listener of the receive task was called twice
no, we are still on 5.23, we will upgrade (hopefully) soon