Failover for the failover monitor
-
Has anybody tried to set up multiple monitoring IPs for a MultiWAN config? Had a case recently where the monitoring IP, (DNS address of a big national ISP, which is itself multi-homed and ridiculously redundant), stopped responding for awhile. This is a rare case given our choice of monitoring IPs but it did mean one of our lines got flagged as "down" even though it was actually just fine.
Was considering adding the interfaces more than once to the LB pools, using different monitoring IPs for each entry.
Without looking at the code I should think the round-robin balancing itself would handle this okay, but would the monitoring test results for each one be handled as a logical AND, OR, XOR, NAND, something else?
Or would they all be separate and distinct and we would simply see a reduction in utilization on the affected line proportionate to the ratio of "good" monitoring results vs "bad" results? I'm guessing the latter but hoping somebody has fiddled with this and has an answer before going and experimenting on a production system…