Setting up a gateway on OpenBSD
TODO(Cross): Wordsmith this section
This page describes how to configure a router running OpenBSD to transfer traffic between a 44Net subnet and the rest of the world.
Assume a setup that looks substantially similar to the following:
- A dedicated static IP address to use as an endpoint for 44net traffic.
- An ISP-provided router that is just a router; no NATing, no firewalling.
- An edge router OpenBSD with three ethernet interfaces. For example, using a Ubiquiti EdgeRouter Lite:
- cnmac0 is the external interface connected to the ISP's network
- cnmac1 connects to your internal network
- cnmac2 is your AMPRNet internal gateway to your subnet (in this example, we use 220.127.116.11/24)
On OpenBSD, tunneling interfaces for IPENCAP are provided by 'gif' pseudo-devices. Unlike Linux, one creates a separate 'gif' interface for each tunnel, but one can create an essentially arbitrary number of such interfaces. If there is a limit it is well above the number required for supporting the full AMPRNet mesh, so don't worry about scalability to the number of interfaces.
On AMPRNet, the UCSD gateway *will not* pass traffic for an IP address that does not have a corresponding entry in the AMPR.ORG DNS domain. Also, 18.104.22.168 does not respond to 'ping' from 44/8 IP's. Caveat emptor as one tries to test: make sure you have DNS entries for your addresses and try pinging something other than 22.214.171.124 or you'll suffer contusions banging your head against a desk trying to figure out why nothing appears to work....
Once I had a tunnel up to UCSD, I found that I could ping my 126.96.36.199 machine from a host on my internal network, but not from arbitrary machines. This was interesting; it turns out that hosts on my internal network get NAT'ed to another IP address on the small subnet I got from Comcast (through another, completely separate router -- not comcast's router but another ERL). What was happening was that as I ping'ed 188.8.131.52 from e.g. my laptop, ICMP echo request packets got NAT'ed to this other address and routed over to amprgw.sysnet.ucsd.edu and tunneled back to the external interface of my AMPRNet gateway. The gateway accepted the encapsulated ICMP echo requests (I have a PF rule that explicitly allows ping) and forwarded them across the tunnel interface where they were unencapsulated; the IP stack saw that the result was addressed to an IP address on a local interface (i.e., they were for the router) and generated an ICMP echo response packet with a
- source* address of 184.108.40.206 and a *destination* address of the
external address of my other router (that is, the address the ICMP echo request was NAT'ed to). This matched the network route for my local Comcats subnet and so my AMPRNet router realized it could pass the packet back to my other router directly. It did so and the other router happily took the packet, matched it back through the NAT back to the original requesting machine (my laptop) and forwarded it: hence, I got my ping responses back. But note that the response was not going through the tunnel back to UCSD: it was being routed directly through the external interface.
Now consider what happens when I tried to ping 220.127.116.11 from a different machine on some other network. The ICMP echo request packet gets routed through the UCSD gateway and tunneled back to my gateway as before, but since responses don't go through back through the tunnel, the response packet matches the default route of my gateway and get's forwarded to comcast's router. Comcast would look at it, see that 18.104.22.168 wasn't on one of it's known networks that it would route floor, and discard the response. Oops.
The solution was to set up a separate routing table in a different routing domain specifically for AMPRNet traffic, and tie the two together using firewall rules. In the AMPRNet routing table, I could set my default route to point to the UCSD gateway, so any traffic sent from one of my 22.214.171.124/24 addresses that doesn't match a route to a known tunnel gets forwarded through amprgw.sysnet.ucsd.edu. With that in place, I could ping my gateway from random machines. This must seem obvious to a lot of folks here, but it took me a little while to figure out what was going on. Things are working now, however.
So far I have encountered two other caveats: I decided to configure two tunnel interfaces statically at boot time: 'gif0' goes to the UCSD tunnel, and 'gif1' sets up a tunnel to N1URO for his 44.88 net. Under OpenBSD, I assumed that the natural way to do this would be to add /etc/hostname.gif0 and /etc/hostname.gif1 files and this does in fact create the tunnels at boot time. However, traffic going out from my gateway doesn't seem to get sent through the tunnels; I did not bother to track down exactly why, but I believe it has to do with some kind of implicit ordering dependency when initializing PF. When I set up the separate routing domain, it struck me that the language accepted by /etc/netstart in an /etc/hostname.if file was not sufficiently rich to set up tunnels in a routing domain, so I capitulated and just set up the static interfaces from /etc/rc.local; imperfect but it works.
The second caveat is that I seem to have tickled a kernel error trying to set up an alias of a second IP address on my 126.96.36.199 NIC; I get a kernel panic due to an assertion failure. It looks a bug to me, but I haven't had the bandwidth to track it down. In the meanwhile, simply don't add aliases to interfaces in non-default routing domains.
The biggest piece missing was a daemon to handle receiving 44net RIP packets and use that data to maintain tunnels and routes. I thought about porting one, but decided of write my own instead. It has been running for a few weeks now on my node and while it's still not quite "done" it seems to work well enough that I decided it was time to cast a somewhat a wider net and push it up to GitHub for comment from others.
A couple of quick notes on implementation:
- The program maintains a copy of the AMPRNet routing table in a modified PATRICIA trie (really a compressed radix tree). Routes are expired after receiving a RIP packet.
- A similar table of tunnels is maintained.
- Tunnel interfaces are reference counted and garbage collected. A bitmap indicating which tunnels are in use is maintained.
- The program is completely self-contained in the sense that I do not fork/exec external commands to e.g. configure tunnels or manipulate routes. That is all done via ioctls or writing messages to a routing socket.
There is more to do; I'm sure there are a few bugs. I'd also like to query the system state at startup to initialize the routing and tunnel tables. Exporting and/or parsing an encap file would be nice. Logging and error checking can, I'm sure, be improved.
It's about 1200 lines of non-comment code, compiles down to a 28K MIPS64 executable (stripped). The code is at https://github.com/dancrossnyc/44ripd