Well that sounds like exactly the sort of thing that shouldn’t happen when there’s an issue given the usual response is to spin things up elsewhere, especially on lower priority services where instant failover isn’t needed.