I'm having trouble figuring out what you're asking for exactly -- probably because I've never done anything involving a clustered failover system.

Is it a problem with keeping exactly one instance of a service running, or just with checking at one-minute intervals (not multiple times per second) that the service is running?

If the latter, I would expect that there's a simple way of probing from outside the cluster to check that the service is running, and this could just be a plain old cron job (or if you really want a perl script running 24/7, it could just be:  while(1){ checkservice(); sleep 60 } (and write the "checkservice" sub to do whatever needs to be done).

If you want some sort of failover on the checking as well, have the primary check routine write something to a specific file on shared disk, and have some other machine check the modification date on that file at regular intervals. If both of these machines outside the cluster go down, you probably don't need a script or log file to tell you that there's a bigger problem...

Sorry if I have completely misunderstood the question.