What you want to do sounds similar to what we have done.

We use P2P but send all our updates to a single node, which is identified using a DNS vanity name. At failover time we update the vanity name and our apps then connect to the new master node.

There one problem we have had to overcome. P2P does not update the identity seed value on receiving nodes. This means that after failover the new master node will want to insert records using an identity value that is already in use.

We have overcome this by writing some code that keeps track of identity value high-water mark, and automatically applying appropriate CHECKIDENT RESEED commands to the new master node at failover.