Noticed there does not seem to be an Interface reachable metric (unless i missed it) but I do see an interface available metric.
I wanted to understand what is counted in interface available metric? Listed a few items PM may take into consideration but not certain this is at all accurate. Could someone clarify?
Items taken into account for Interface availability?
Interface Operational Status being down?
Interface Admin Status down?
Device being down?
Device is up (has power) but is not reachable at all due to WAN circuit being down.
Times where interface is admin up/operationally up but not reachable via directly ICMP due to BGP problem as example but device is reachable through secondary WAN router.
This is the calculation, at least according to the interface vendor cert:
Availability = (ifOperStatus == 1) ? 100 : 0
For a given poll, if ifOperStatus = 1 (up), then it's 100% available. If it's not 1 (others are down, testing, dormant, etc.) then it's 0. If you have 5 minute polling and it's down for 2 polls then ((10 * 100) + (2 * 0))/12 = 83.33% availability. I don't think there is any backfill of interface availability based on the value of ifLastChange but I could be wrong. That would be an interesting enhancement. Also, the poll is to the device managed IP, not the IP of the individual interface so in your BGP example, the device itself would need to be unreachable from PM not just that particular interface.
Hope that helps.
To clarify if the device is down from PM standpoint the SNMP poll fails and so ifOperStatus will be 0 to count against availability, correct?
Did Rob answer your question? If so, can we mark this one as answered?