When close or delete database, the active replicators will be stopped. However the stop process is asynchronous and could still also write something to the database such as checkpoint. After chatting with @snej, we should wait until the all replicators finish stopping (no more active replicators) with some timeout.
In 2.0, we will throw a runtime exception if there is still a replicator or a query listener registered. In 2.1, we could consider waiting for the replicator and live query to finish.
When close or delete database, the active replicators will be stopped. However the stop process is asynchronous and could still also write something to the database such as checkpoint. After chatting with @snej, we should wait until the all replicators finish stopping (no more active replicators) with some timeout.