This is only tangentially relevant, but you might come across it at some point, so I'll go ahead and point it out.

You can't make a subnet out of any random grouping of sequential IP addresses, even if it's a correct number of them (which will always be a power of two). The math involved means that 69.125.107.32/255.255.255.248 is a valid network of eight addresses, but 69.125.107.30/255.255.255.248 is not. That's the reason I claimed your .254 WAN network settings are screwed up -- because they can't make a valid network.

The reason for this is that an IP address is made up of two numbers, the network number and the host number. The netmask defines how many bits the network number takes up, and, therefore, how many are left over for the host number. Obviously, all of the IP addresses in a given network must have the same network number. For an easy example, assume a netmask of 255.255.255.0. We're all fairly familar with that. That means that your network could encompass, for example, 192.168.1.0 to 192.168.1.255, but not 192.168.1.16 to 192.168.2.15. That seems obvious, but the math becomes a little more complex when you're translating binary numbers into decimal numbers that aren't as tidy as that example.

Also, you may see netmasks described differently. For example, 192.168.1.0/255.255.255.0 might also be listed as 192.168.1.0/24. The old-style netmasks are really, in binary, a bunch of ones followed by a bunch of zeroes. The new way simply describes how many ones there are. There's a direct mapping: /25 is .128, /26 is .192, /27 is .224, /28 is .240, /29 is .248, /30 is .252, /31 is .254, and /32 is .255. Netmasks can be smaller, too, but you're not really going to encounter any of those.

I point that out so that you can use aggis, which is a handy-dandy utility for figuring out network addressing. It requires that its input be in new-style format.