There is a failure pattern for SQL Server service pack and cumulative updates that I had somehow not encountered up until now. The SP or CU runs various upgrade scripts while it is doing its thing and, rarely, one of these will fail due to something that is different about that particular server.
