For each interface, for each of a set of target IPs specific to that interface, it sends a ping once per second, and notes how long it took to return.
eth1 -> 126.96.36.199: 217/217 (0.00% loss) min: 0.59ms mean: 1.97ms max: 5.82ms eth1 -> 188.8.131.52: 217/217 (0.00% loss) min: 0.14ms mean: 107.43ms max: 3001.71ms eth2 -> 184.108.40.206: 207/207 (0.00% loss) min: 0.19ms mean: 3911.12ms max: 19246.84ms eth2 -> 220.127.116.11: 207/207 (0.00% loss) min: 0.13ms mean: 3926.73ms max: 19248.81ms
The first IP for each interface is the first hop outside my network on a traceroute to 18.104.22.168. The second is google public DNS.
eth1 is the speakeasy connection, eth2 is comcast.
Basically, comcast will be fine for a few seconds, and then stop responding at all.. a dozen or more pings will go out, nothings comes back.. then they all come back at once. So I get, frex, 19 pings in a row, with ping times of 19, 18, 17, ... 1, 0 seconds each. Entirely unacceptable.
At this point, the problem /definitively/ lies somewhere on THEIR side of the cable modem, as /everything/ on this side, including the modem, has been replaced.
I need to call and yell at them next week.