Host down detection

[Bombe]

I understand that host down detection is being better integrated into cacti/threshold?

Something I'd like to see is a difference between host down and host unreachable.

In large networks a scheduled change may take a section of routing/networks/equipment out for a period of time. A host unreachable is significantly different from a host down. What we do currently is use a devices next hop in addition to ping.

ie if a host becomes unreachable, is the switch it's connected to reachable?

If the switch is unreachable, the host is not down, it's unreachable.