I have a IP SLA being tracking a primary interface of a remote 2951 router. When the the pings to 22.214.171.124 from the primary interface fail - it sends a UP->Down to the syslog buffer and then it triggers the Event Manager script and the backup interface comes up. And the opposite happens when the primary interface starts pinging 126.96.36.199 successfully.
The issue I was having was that when the primary interface was pinging to 188.8.131.52, every so often( like once a day) the ping would drop and it would trigger the Event Manager script - which would fail over to the backup and then quickly fail back to the primary but, the users would go down briefly. I had the tracking delay up and delay down set to 10 but, the frequency of the IP SLA set to 30. Now I have the frequency still set to 30 but I increased the tracking delay up and delay down to 180 - the max. I believe that will take care of the occassional ping lose.
This recovery time of 180 is a little long. The primary interface is connected to the VPN via a 50Mbs pipe. If I increased the IP SLA frequency to - say every second and decreased the tracking delay up and delay down to 10, first would it be taxing on the bandwidth and second, would the deleay down 10 delay up 10 be enough time for the occasional ping lose?