Sorry for this act of thread necromancy.  And thread-jacking, I guess.  Yay for first posts!

Perry, if you are still out there, would you be able to point me to any documentation that confirms the downgrade will occur?  I'm trying to confirm that if we apply a service pack to an inactive node then failover to it, we can safely failback to an unpatched node if things go south.

I'm having trouble finding confirmation one way or the other online.  The link you provided mentions possible data corruption by having mixed versions as possible owners.  This makes me think I want to avoid the scenario?  I guess we could try to uninstall the service pack instead, although the Microsoft rep we had on site a while back told us uninstalling SQL Server service packs was something we really wanted to avoid.  Anecdotes among the team seem to support that view.