failed node criteria (heartbeat)

[Karl <karl.mueller@xxxxxxxxxxxxxx>]

I'm using LVS-NAT, and I'm wondering if there is a 'wrapper' for
heartbeat that would detect component failures on the active node (i.e.
losing network connectivity on one interface) and fail-over to the
secondary for something less then a total failure of the primary. 
Essentially, I'm looking for an 'environment sensor' of some sort w/
enough logic to keep the workload on whichever LVS node is 'healthiest'.

thanks
-karl