Well, I'm a server engineer myself, and we are required to have worst case scenario contingency plans in place. One of the things required in that respect is the ability to roll back to the way things were. In other words, abandon the upgrade and go back to the previous server and then look back and try to get the answers to why the upgrade failed.
I just I think putting things back the way they were should have been an option.
MM 8)