I have the Netgear gateway running the 1.04 firmware that was installed today.
I have subscribed to 5 static IP addresses.
Four of these are attached to a single interface in my router/firewall and all of my servers are behind that (one "real" address; three aliases/secondaries).
Everything to do with these four addresses works brilliantly. All port forwards, all of my internal 1:1 NAT (where applicable), VPN, etc. work as designed.
My fifth address is assigned to a cheap little wireless router that I use for guest access to the internet. I have the router configured with the highest-numbered static IP in the /30 subnet (255.255.255.252) so it can reach only the gateway IP Comcast assigned to me.
When I have a system connected to this wireless router, it gets an internal IP (192.168.0.0/24) as expected and its connection to the internet is rock solid. I can browse to my heart's delight and run persistent pings, etc. with no packet loss and it is nice and quick.
Whenever I try to ping one of my own addresses, however (if I want to go to my own website, for example), I get between 20% - 40% packet loss. HTTP/HTTPS connections are painfully slow and it is overall unreliable.
From a true external address (tethering from my mobile phone), I can ping all of my public addresses with low latency and no packet loss.
Based on this, I can only assume that something isn't traversing properly inside the Business Gateway.
Does this scenario sound familiar to anyone, or can anyone provide any guidance/clues/guesses/etc.?
So I had a bit of a rethink and changed the netmask on my guest wireless router from the /30 subnet I had been using (2 hosts) to the /29 subnet assigned to me by Comcast. My assumption was that the gateway might not be routing supposedly-local addresses properly because they should, ostensibly, not need to be routed.
So far, so good and everything appears to be working normally.
But if anyone has the time, I would greatly appreciate feedback on the following two questions:
1) Does this make sense, or am I simply mad?
2) If it does make sense, then why was the business gateway able to route 60%-80% of the traffic when it shoudln't have routed anything at all?