Agreed, we've been running multiple aurora clusters in production for years now and have not encountered this issue with failovers.