Well, I have patched via SP and CU's on SQL2008 and SQL2008R2 clusters doing the passive node first, then failing SQL Server over and then patching the other node without any isues quite a bit.