See steps to reproduce the issue.
This is only happening when the persistence is enabled. Basically, when we start, cancel a node instance and trigger a new one, we're doing this atm:

1.- Start node instance
– Internally, the data store is updated
2.- Cancel node instance
– Internally, the data store is updated
3.- Trigger node instance

We should update the data store as done in the steps 1 and 2 here, at the moment this is not being done. Without persistence, this is working because we update directly the instances in memory.

Possible solution:

I tried to move this line to be done at the end of the triggerNode method and this change made the test worked when persistence and without persistence. However, I'm not sure if this is the valid solution.