Currently, during recovery, endpoints are re-activated and replace endpoints that were left (see PETALSESBCONT-417) in the previous execution of the container.
But if for some reasons some of the previously activated endpoints of the container are not re-activated during recovery, they should not be left as activated since it is 100% sure at that point that they won't be activated automatically anymore and that the only reasons that are activated again is by an administrative action.