It happens all the time these days, but back then, I explained how we managed cluster nodes that didn't quite match.