Using RMAN for recovery of RAC database

We are using a shell script to run rman backup jobs nightly, full backups on Sat nite, incrementals the rest of the week.  This is done on instance 1 of the cluster.

Using grid control, I perform recovery on the cluster database to a prior point in time by specifying an SCN.  I don't open it in read only mode so the open resetlogs command is issued at the end.  The recovery was succcessful in this test.

If I want to do the exact thing over, using the same SCN, what steps do I need to do?  I've been looking at the 10g docs that deal with resetting the incarnation and restoring a previous control file that knows of that incarnation.  I'm having trouble following the examples, when to have the DB mounted and not.