services/router: init

2026-09-17 08:57:32 +07:00
parent b85d6637f1
commit 60aac6efb2
39 changed files with 1831 additions and 929 deletions
@@ -0,0 +1,91 @@
+# router
+
+Turns a machine with several NICs into a site gateway: PPPoE WAN (ISP
+credentials via vars prompts), a VLAN-filtering bridge over the LAN ports with
+one L3 interface per VLAN, Kea DHCP and Blocky DNS per VLAN, nftables
+firewall/NAT, DHCPv6-PD, CrowdSec, an iperf3 server and a WAN speed-test
+timer. Optional: a Wi-Fi access point on the router's own radios (hostapd),
+the TP-Link Omada controller (podman) and an internal Caddy reverse proxy
+with a real wildcard certificate (ACME DNS-01).
+
+Addressing convention: a site owns `10.<siteId>.0.0/16`; VLAN `<id>` defaults
+to `10.<siteId>.<id>.0/24`, router at `.1`, DHCP pool `.100-.199`. The `mgmt`
+and `lan` VLANs are mandatory. Trust model: mgmt reaches everything; other
+VLANs get router DNS/DHCP and (with `allowWan`) the internet, no inter-VLAN;
+WAN nothing inbound; the admin mesh (`mesh.subnet`) gets SSH, metrics, iperf3
+and the Omada UI.
+
+## Usage from another clan
+
+```nix
+# flake.nix
+inputs.cnx-network.url = "git+https://<host>/B4L/cnx-network-clan";
+
+# clan.nix
+inventory.instances.router = {
+  module = { name = "router"; input = "cnx-network"; };
+  roles.default.settings.mesh.subnet = "fd..::/88"; # your admin overlay
+  roles.default.machines.gw-1.settings = {
+    site = "ams";
+    siteId = 1;
+    wan.interface = "enp1s0";
+    wan.vlanId = 10;                 # or null for untagged PPPoE
+    trunkPorts = [ "enp2s0" ];
+    accessPorts.enp4s0 = "mgmt";     # untagged on-site recovery port
+    # stagingPort = "enp3s0";       # DHCP uplink into the old LAN until cutover
+    vlans = {
+      mgmt.id = 10;
+      lan.id = 20;
+      iot = { id = 40; allowWan = false; };
+    };
+  };
+};
+```
+
+Then `clan vars generate gw-1` prompts for the PPPoE username/password.
+
+### Wi-Fi access point
+
+If the box has wireless cards, the router can be the site's AP. An SSID is
+defined once and behaves like an untagged access port of its VLAN; radios
+choose what to broadcast, so a dual-band card serves the same SSID twice:
+
+```nix
+wifi = {
+  enable = true;
+  countryCode = "TH";
+  networks = {
+    home.vlan = "lan";                                   # WPA3 with WPA2 fallback
+    things = { vlan = "iot"; security = "wpa2"; };       # legacy IoT
+    guest = { vlan = "guest"; isolateClients = true; };
+  };
+  radios = {
+    wlp5s0 = { band = "2g"; channel = 6; macAddress = "…"; networks = [ "home" "things" ]; };
+    wlp6s0 = { band = "5g"; channel = 36; networks = [ "home" ]; };
+  };
+};
+```
+
+Passphrases are vars prompts (`wifi-<name>-passphrase`), asked once at `clan
+vars generate`. A radio broadcasting more than one SSID needs its hardware
+`macAddress`: hostapd wants a fixed BSSID per extra SSID, derived from it.
+`security = "wpa3-transition"` (the default) offers SAE and WPA2-PSK-SHA256;
+devices that only speak classic WPA2-PSK need `security = "wpa2"`.
+
+### Internal proxy
+
+`proxy.enable` serves `<name>.<site><siteId>.<proxy.domain>` under a wildcard
+certificate obtained via RFC 2136 DNS-01 against `proxy.acme.nameserver`. The
+gateway signs updates with TSIG key `acme_<hostname with _>`, whose secret is
+the shared `dns-acme-<hostname>-secret` generator declared by this service.
+The nameserver machine must declare the same generator so both sides hold one
+secret — import `acme-secret.nix` from this directory with the gateway's name:
+
+```nix
+imports = [ (import "${inputs.cnx-network}/modules/clan/router/acme-secret.nix" "gw-1") ];
+```
+
+and load the key with an acl scoped to `_acme-challenge.<site><siteId>`.
+
+The service does not open the WAN to anything; reach gateways over your mesh.
+One instance per machine.