Whose Change Is It?

A cluster’s failures come back mostly too late, and a change request would fire every handover earlier. Read the stored failure records it rests on, one at a time, and say whose change each one is, then what happens next where it matters.