That one was probably related to the recently discovered bug where deleting a users with thousands of posts will lock out the database, thus causing a self-DOS. And since the deletion will propagate to other instances, other federated instances might be down too depending on how much federated contents that must be deleted in that instance.