More often than not Service Pack upgrades go without any problems but if you are unlucky things can get really bad.
My fellow Cluster MVP, Robert Smit (also from the Netherlands) came across a failed SP1 upgrade ending with 0x800705b4 and 0×80070643. The end of the game is that your server is lost and you have to forcibly test your recovery plan.
If your server is part of a cluster and you have not used up too many resources and fully consumed your cluster reserve you may be able to remove the troublesome node from the cluster.
The primary lesson is: “don’t upgrade all your servers simultaneously”
Just upgrade one and see how that one goes. If the SP1 upgrade was successful, upgrade the next.
For the full story read Roberts blog at: