Issue with some unnamed ISP's in northern north america, RFC 1918 Violation
One of the fibre providers, they are issuing static ips, which is good, but the way the statics are handled. The modem/router, if you want static ips, requires it to be in router mode, you can't bridge. it seems the router's static ip gateway does some bgp towards the isp to route the static ip's block assigned to the customer to it, via its dhcp established route??? idk. ------ THE big issue, which is, since it has to be in router mode. they end up with a dynamic ip, that is the gateway for dhcp RFC1918 addresses, that are on its lan, sounds reasonable, but as a tertiary use, the static ip WAN gw stack, LOADS IN ON the same L2 lan bridge, side by side with the RFC 1918 space, without even attempting to put it in a tagged vlan. don't do this.... without using a custom linux router, good luck using both the static ips block, and the double 'natted' dynamic ip.
Not sure if you meant Comcast but it sounds similar to what they do. I think they are using RIP not BGP, but it works the way you're describing. You have to statically configure the static IP and gateway - if you let it DHCP it will get one of those 10.x.x.x numbers. It's just two addresses on the same interface on the CPE: the 10.x.x.1 IP and also the first address from your static IP block. On 3/10/26 9:13 PM, wolf1098--- via NANOG wrote:
One of the fibre providers, they are issuing static ips, which is good, but the way the statics are handled. The modem/router, if you want static ips, requires it to be in router mode, you can't bridge. it seems the router's static ip gateway does some bgp towards the isp to route the static ip's block assigned to the customer to it, via its dhcp established route??? idk. ------ THE big issue, which is, since it has to be in router mode. they end up with a dynamic ip, that is the gateway for dhcp RFC1918 addresses, that are on its lan, sounds reasonable, but as a tertiary use, the static ip WAN gw stack, LOADS IN ON the same L2 lan bridge, side by side with the RFC 1918 space, without even attempting to put it in a tagged vlan. don't do this.... without using a custom linux router, good luck using both the static ips block, and the double 'natted' dynamic ip. _______________________________________________ NANOG mailing list https://lists.nanog.org/archives/list/nanog@lists.nanog.org/message/CWEOU5LH...
That's not at all uncommon. ATT fiber works that way as well. The static IP block comes out on the same L2 port as the dynamic NAT pool. Devices which ask for an address via DHCP get NAT'd v4 space, devices which are configured manually use the static IP address and gateway. Works just fine. Doesn't seem like much of a violation to me. Perhaps you could clarify why you think it's a violation of the RFC to have both a dynamic DHCP block and a static block existing on the same layer 2 segment? Thanks! Matt On Tue, Mar 10, 2026 at 2:19 PM wolf1098--- via NANOG <nanog@lists.nanog.org> wrote:
One of the fibre providers, they are issuing static ips, which is good, but the way the statics are handled. The modem/router, if you want static ips, requires it to be in router mode, you can't bridge. it seems the router's static ip gateway does some bgp towards the isp to route the static ip's block assigned to the customer to it, via its dhcp established route??? idk. ------ THE big issue, which is, since it has to be in router mode. they end up with a dynamic ip, that is the gateway for dhcp RFC1918 addresses, that are on its lan, sounds reasonable, but as a tertiary use, the static ip WAN gw stack, LOADS IN ON the same L2 lan bridge, side by side with the RFC 1918 space, without even attempting to put it in a tagged vlan. don't do this.... without using a custom linux router, good luck using both the static ips block, and the double 'natted' dynamic ip. _______________________________________________ NANOG mailing list
https://lists.nanog.org/archives/list/nanog@lists.nanog.org/message/CWEOU5LH...
participants (3)
-
Laszlo H -
Matthew Petach -
wolf1098＠pm.me