The following link does not mirror your configuration, but shows carp in action as I use it at work for redundant caching-only DNS servers-
http://openbsd-wiki.org/index.php?ti...nt_caching-DNS
If anyone asks, "Why would CARP be needed for DNS resolver redundancy?" then they haven't waited while their computer times out on the first failed DNS server (the primary) before attempting the secondary server. With CARP, the primary is "always on" even when taken down for maintenance. And if users think they're clever by reversing the primary and secondary resolvers in their machine configurations... well, the secondary is CARP'd as well.