full tunnel

travel router
fix reverse dns
2026-07-16 16:59:37 -07:00 · 2026-07-16 11:44:08 -07:00 · 2026-07-16 01:14:20 -07:00 · 2026-07-16 00:54:33 -07:00 · 2026-07-15 21:24:42 -07:00 · 2026-07-15 21:22:15 -07:00
22 changed files with 458 additions and 81 deletions
@@ -32,6 +32,10 @@ RADIUS_SECRET=
 # Applied on FIRST start only; change later via the console itself.
 TECHNITIUM_ADMIN_PASSWORD=

+# ── DNS zone for VPN client registration (A + PTR via learn-address) ──
+# Leave empty to disable client DNS registration entirely.
+VPN_DNS_ZONE=roam.atlashorizon.net
+
 # ── Optional (defaults shown) ─────────────────────────────────────────
 # Duo Auth Proxy version; pin (e.g. 6.5.1) for deterministic updates
 #DUO_AUTHPROXY_VERSION=latest
@@ -26,6 +26,7 @@ Oswald↔Trochalakis tunnel — this box never carries it.
 | Oswald LAN           | `192.168.7.0/24`  | AH-OSWLDC                        |
 | WG transfer          | `172.31.255.0/24` | hub .1, Troch .2, Oswald .3      |
 | OpenVPN clients      | `172.31.8.0/24`   | routed (not NATed) to both sites |
+| WG travel routers    | `172.31.9.0/24`   | GL.iNet roaming peers, hub .1 ([docs](docs/TRAVEL-ROUTERS.md)) |

 Public endpoint: `vpn.atlashorizon.net` — only 22/tcp (SSH), 51820/udp
 (WireGuard) and 1194/udp (OpenVPN) are open. RADIUS binds loopback only;
@@ -3,9 +3,10 @@
 # later run uses — key-only auth, passwordless sudo — verify it actually
 # works, then lock sshd down to keys. Run once per fresh VPS:
 #
-#   ansible-playbook bootstrap.yml          # root already has your key
-#   ansible-playbook bootstrap.yml -k       # root still uses a password
-#                                           # (needs sshpass installed)
+#   ansible-playbook bootstrap.yml
+#
+# It prompts for the root SSH password (needs sshpass on the control
+# node). If root already authenticates with your key, just press Enter.
 #
 # Afterwards set `ansible_user: ansible` in inventory.yml and never
 # connect as root again. Overridable with -e:
@@ -22,10 +23,16 @@
 - name: Create the automation user (connects as root)
  hosts: vpnhub
  gather_facts: false
+  vars_prompt:
+    - name: bootstrap_root_password
+      prompt: "Root SSH password (Enter if root already uses your key)"
+      private: true
+      default: ""
  vars:
    ansible_user: root          # play var — outranks inventory on purpose
+    ansible_password: "{{ bootstrap_root_password }}"
    bootstrap_user: ansible
-    bootstrap_ssh_public_key_file: "~/.ssh/id_ed25519.pub"
+    bootstrap_ssh_public_key_file: "~/.ssh/togo-ansible.pub"
  tasks:
    - name: Create the user (no password hash set, account password locked)
      ansible.builtin.user:
@@ -34,6 +34,21 @@ radius_secret: ""
 # console (this var must still match for the init job's API login).
 technitium_admin_password: ""

+# DNS zone where VPN clients get registered as <username>.<zone> with a
+# matching PTR (Technitium primary zone, created by the init job). Set to
+# "" to disable registration.
+vpn_dns_zone: roam.atlashorizon.net
+
+# ── Travel routers (GL.iNet etc.) as roaming WireGuard peers ──
+# Each gets a /32 from 172.31.9.0/24 (hub is 172.31.9.1). Generate the
+# keypair on the router (or wg genkey) and paste the PUBLIC key here.
+# The init job also registers <name>.<vpn_dns_zone> A + PTR records.
+wg_travel_peers: []
+#wg_travel_peers:
+#  - { name: glinet-1, public_key: "xxxx=", ip: 172.31.9.11 }
+#  - { name: glinet-2, public_key: "yyyy=", ip: 172.31.9.12 }
+#  - { name: glinet-3, public_key: "zzzz=", ip: 172.31.9.13 }
+
 # ── Optional ──
 # Pin for deterministic Duo updates (e.g. "6.5.1"); bumping it forces a
 # refetch of the tarball on the next rebuild.
@@ -23,6 +23,7 @@
    vpnhub_base_dir: /opt
    vpnhub_stack_dir: "{{ vpnhub_base_dir }}/vpnhub"
    vpnhub_force_rebuild: false
+    vpnhub_host_dns: [1.1.1.1, 9.9.9.9]
    public_iface: "{{ ansible_default_ipv4.interface }}"
    wg_peers_configured: >-
      {{ (troch_wg_public_key | default('')) != ''
@@ -30,9 +31,9 @@

  handlers:
    - name: reload firewall
+      # the reload flushes the Docker-managed chains, so restart docker after
      ansible.builtin.shell: |
        netfilter-persistent reload
-        # the reload flushes Docker's chains — put them back
        if systemctl is-active --quiet docker; then systemctl restart docker; fi

    - name: restart wireguard
@@ -50,17 +51,18 @@
        chdir: "{{ vpnhub_stack_dir }}"

    - name: recreate duo-authproxy
-      ansible.builtin.command: docker compose up -d --force-recreate duo-authproxy
+      # --build: entrypoint/cfg-template changes are baked into the image
+      ansible.builtin.command: docker compose up -d --build --force-recreate duo-authproxy
      args:
        chdir: "{{ vpnhub_stack_dir }}"

    - name: recreate openvpn
-      ansible.builtin.command: docker compose up -d --force-recreate openvpn
+      ansible.builtin.command: docker compose up -d --build --force-recreate openvpn
      args:
        chdir: "{{ vpnhub_stack_dir }}"

    - name: recreate stack
-      ansible.builtin.command: docker compose up -d --force-recreate
+      ansible.builtin.command: docker compose up -d --build --force-recreate
      args:
        chdir: "{{ vpnhub_stack_dir }}"

@@ -76,9 +78,12 @@
            update_cache: true
            cache_valid_time: 3600

+        # Distro-agnostic: ansible_distribution is Ubuntu or Debian, and
+        # Docker publishes separate repos per distro (a Debian codename
+        # like trixie does not exist under linux/ubuntu and vice versa).
        - name: Docker apt signing key
          ansible.builtin.get_url:
-            url: https://download.docker.com/linux/ubuntu/gpg
+            url: "https://download.docker.com/linux/{{ ansible_distribution | lower }}/gpg"
            dest: /etc/apt/keyrings/docker.asc
            mode: "0644"

@@ -87,7 +92,7 @@
            repo: >-
              deb [arch={{ 'arm64' if ansible_architecture == 'aarch64' else 'amd64' }}
              signed-by=/etc/apt/keyrings/docker.asc]
-              https://download.docker.com/linux/ubuntu
+              https://download.docker.com/linux/{{ ansible_distribution | lower }}
              {{ ansible_distribution_release }} stable
            filename: docker

@@ -124,10 +129,19 @@
        sysctl_file: /etc/sysctl.d/99-vpnhub.conf
        reload: true

-    # Technitium binds 0.0.0.0:53 — systemd-resolved's stub listener on
-    # 127.0.0.53:53 would collide with it.
+    # Technitium binds 0.0.0.0:53 — free the port while keeping host DNS
+    # working. Ubuntu runs systemd-resolved (stub on 127.0.0.53:53 must be
+    # disabled); plain Debian has no resolved and just needs a sane static
+    # /etc/resolv.conf.
+    - name: Check whether systemd-resolved is in use
+      tags: host
+      ansible.builtin.stat:
+        path: /run/systemd/resolve/resolv.conf
+      register: resolved_upstream
+
    - name: resolved drop-in directory
      tags: host
+      when: resolved_upstream.stat.exists
      ansible.builtin.file:
        path: /etc/systemd/resolved.conf.d
        state: directory
@@ -135,6 +149,7 @@

    - name: Disable systemd-resolved stub listener (frees :53 for Technitium)
      tags: host
+      when: resolved_upstream.stat.exists
      ansible.builtin.copy:
        dest: /etc/systemd/resolved.conf.d/99-vpnhub.conf
        mode: "0644"
@@ -143,14 +158,29 @@
          DNSStubListener=no
      notify: restart systemd-resolved

-    - name: Point /etc/resolv.conf at resolved's upstream file (not the dead stub)
+    - name: Point /etc/resolv.conf at resolved upstream file (not the dead stub)
      tags: host
+      when: resolved_upstream.stat.exists
      ansible.builtin.file:
        src: /run/systemd/resolve/resolv.conf
        dest: /etc/resolv.conf
        state: link
        force: true

+    - name: Static resolv.conf (no systemd-resolved on this host)
+      tags: host
+      when: not resolved_upstream.stat.exists
+      ansible.builtin.copy:
+        dest: /etc/resolv.conf
+        follow: false     # replace a dangling symlink rather than write through it
+        mode: "0644"
+        content: |
+          # Managed by Ansible (vpnhub). Public resolvers on purpose: the
+          # host must resolve registries/repos even when the stack is down.
+          {% for ns in vpnhub_host_dns %}
+          nameserver {{ ns }}
+          {% endfor %}
+
    - name: Firewall rules (rendered from host/rules.v4/v6.template)
      tags: [host, firewall]
      ansible.builtin.copy:
@@ -189,15 +219,23 @@
      ansible.builtin.debug:
        msg: "RackNerd WG public key: {{ wg_pubkey.stdout }}"

-    - name: Deploy wg0.conf (rendered from host/wg0.conf.example)
+    - name: Deploy wg0.conf (rendered from host/wg0.conf.example + travel peers)
      tags: [host, wireguard]
      ansible.builtin.copy:
        dest: /etc/wireguard/wg0.conf
        mode: "0600"
-        content: "{{ lookup('file', playbook_dir ~ '/../host/wg0.conf.example')
-          | replace('<RACKNERD_WG_PRIVATE_KEY>', wg_key_raw.content | b64decode | trim)
-          | replace('<TROCHALAKIS_WG_PUBLIC_KEY>', troch_wg_public_key | default('<TROCHALAKIS_WG_PUBLIC_KEY>', true))
-          | replace('<OSWALD_WG_PUBLIC_KEY>', oswald_wg_public_key | default('<OSWALD_WG_PUBLIC_KEY>', true)) }}\n"
+        content: |
+          {{ lookup('file', playbook_dir ~ '/../host/wg0.conf.example')
+             | replace('<RACKNERD_WG_PRIVATE_KEY>', wg_key_raw.content | b64decode | trim)
+             | replace('<TROCHALAKIS_WG_PUBLIC_KEY>', troch_wg_public_key | default('<TROCHALAKIS_WG_PUBLIC_KEY>', true))
+             | replace('<OSWALD_WG_PUBLIC_KEY>', oswald_wg_public_key | default('<OSWALD_WG_PUBLIC_KEY>', true)) }}
+          {% for p in wg_travel_peers | default([]) %}
+
+          # ── Travel router: {{ p.name }} (roaming /32, appended by Ansible) ──
+          [Peer]
+          PublicKey = {{ p.public_key }}
+          AllowedIPs = {{ p.ip }}/32
+          {% endfor %}
      no_log: true
      notify: restart wireguard

@@ -18,6 +18,9 @@ RADIUS_SECRET={{ radius_secret }}

 TECHNITIUM_ADMIN_PASSWORD={{ technitium_admin_password }}

+VPN_DNS_ZONE={{ vpn_dns_zone | default('roam.atlashorizon.net') }}
+VPN_TRAVEL_PEERS={% for p in wg_travel_peers | default([]) %}{{ p.name }}:{{ p.ip }}{{ "," if not loop.last else "" }}{% endfor %}
+
 DUO_AUTHPROXY_VERSION={{ duo_authproxy_version | default('latest') }}

 # /opt layout — consumed by compose ${*_DIR} interpolation
@@ -0,0 +1,62 @@
+client
+dev tun
+proto udp
+remote connect.atlashorizon.net 1194
+resolv-retry infinite
+nobind
+persist-key
+persist-tun
+dhcp-option DNS 172.31.8.1
+
+remote-cert-tls server
+tls-version-min 1.2
+data-ciphers AES-256-GCM:CHACHA20-POLY1305
+auth SHA256
+
+auth-user-pass
+auth-retry interact
+hand-window 120
+verb 3
+
+<ca>
+-----BEGIN CERTIFICATE-----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+-----END CERTIFICATE-----
+</ca>
+<tls-crypt>
+-----BEGIN OpenVPN Static key V1-----
+5fedf4b825a32eb88722a14f801ef416
+2d015c4b5c41654c556108fc2fa5006e
+f83c97ea210113935dbaf47469ba9cd4
+752f928caf48926b47b6da2b4573f6fa
+ed909502d027ba02e509dc48c6542d74
+53c3013ce00e443b43faac4f8f87717f
+1fe942bd1f1412fb38566a88ba7e0d5a
+b722ea79cb4a1b5e172185d2644db81b
+33fdc631815e47f6ae5ce099beee5c37
+b0880c7ff8f9c4ff7a8302e7eb692ce9
+eab50f743de8aa91bff266b2dc4f504c
+775137ae64d71c79e35a40e8bedb8f67
+65bfa02d05bfc2a5e8824eff9be34a26
+a8c8b151a215ae26c67c006e431c8a57
+8985d2feec6787b629b2e30dfdc5ffea
+a0d26b0b0f1aca83ccbfc77b660cceed
+-----END OpenVPN Static key V1-----
+</tls-crypt>
@@ -1,7 +1,7 @@
 # AtlasHorizon VPN — client profile template.
-# Per user: paste ca.crt, the user's cert/key, and ta.key (from the offline
-# CA, see openvpn/pki/README.md) into the inline blocks below, save as
-# <username>.ovpn, import into OpenVPN Connect.
+# ONE profile for everyone: paste ca.crt and ta.key (from the offline CA,
+# see openvpn/pki/README.md) into the inline blocks below and distribute.
+# No per-user certificates — identity is AD user/pass + Duo.
 #
 # Login: AD username; password field takes either
 #   ADpassword              -> Duo push
@@ -33,12 +33,6 @@ verb 3
 <ca>
 # ----- paste ca.crt -----
 </ca>
-<cert>
-# ----- paste <username>.crt -----
-</cert>
-<key>
-# ----- paste <username>.key -----
-</key>
 <tls-crypt>
 # ----- paste ta.key -----
 </tls-crypt>
@@ -29,7 +29,9 @@ services:
      - DNS_SERVER_ADMIN_PASSWORD=${TECHNITIUM_ADMIN_PASSWORD}
      - DNS_SERVER_WEB_SERVICE_HTTP_PORT=5380
      - DNS_SERVER_RECURSION=UseSpecifiedNetworkACL
-      - DNS_SERVER_RECURSION_NETWORK_ACL=127.0.0.0/8,172.31.8.0/24,172.31.255.0/24,192.168.20.0/24,192.168.7.0/24
+      # NOTE: seeded on first start only — on an existing deploy, add new
+      # networks in the console instead (Settings -> Recursion).
+      - DNS_SERVER_RECURSION_NETWORK_ACL=127.0.0.0/8,172.31.8.0/24,172.31.9.0/24,172.31.255.0/24,192.168.20.0/24,192.168.7.0/24
    volumes:
      - ${TECHNITIUM_DIR:-./technitium}/config:/etc/dns
    restart: unless-stopped
@@ -69,6 +71,7 @@ services:
    env_file: .env
    volumes:
      - ${OPENVPN_DIR:-./openvpn}/server.conf:/etc/openvpn/server.conf:ro
+      - ${OPENVPN_DIR:-./openvpn}/learn-address.sh:/etc/openvpn/learn-address.sh:ro
      - ${OPENVPN_DIR:-./openvpn}/pki:/etc/openvpn/pki:ro
      - openvpn-state:/var/lib/openvpn
    depends_on:
@@ -13,7 +13,8 @@ Add the RackNerd peer to each site's existing WG config. The sites
 [Peer]                       # RackNerd hub
 PublicKey = <RACKNERD_WG_PUBLIC_KEY>       # printed by host/setup.sh
 Endpoint = vpn.atlashorizon.net:51820
-AllowedIPs = 172.31.255.1/32, 172.31.8.0/24
+# .8 = OpenVPN clients, .9 = travel routers (docs/TRAVEL-ROUTERS.md)
+AllowedIPs = 172.31.255.1/32, 172.31.8.0/24, 172.31.9.0/24
 PersistentKeepalive = 25
 ```

@@ -35,6 +36,7 @@ LAN router (or directly on the DCs):
 | Destination      | Via                    |
 |------------------|------------------------|
 | `172.31.8.0/24`  | site WireGuard box     |
+| `172.31.9.0/24`  | site WireGuard box     |
 | `172.31.255.1/32`| site WireGuard box     |

 Verify no asymmetry: `tracert 172.31.8.10` from a DC must leave via the
@@ -42,9 +44,10 @@ local WG box → RackNerd, never via the other site.

 ## 3. Windows Firewall / segmentation on the DCs

-Allow from **`172.31.8.0/24`** (roaming clients, real source IPs): standard
-AD client ports — DNS 53, Kerberos 88/464, LDAP 389/636, SMB 445, RPC 135 +
-49152-65535, NTP 123, GC 3268/3269.
+Allow from **`172.31.8.0/24`** (OpenVPN clients, real source IPs) and
+**`172.31.9.0/24`** (travel routers): standard AD client ports — DNS 53,
+Kerberos 88/464, LDAP 389/636, SMB 445, RPC 135 + 49152-65535, NTP 123,
+GC 3268/3269.

 Allow from **`172.31.255.1/32`** (the RackNerd host itself):
 - TCP/UDP 53 → AHDC01, AHDC03, AH-OSWLDC (Technitium forwarder queries)
@@ -0,0 +1,68 @@
+# Travel routers (GL.iNet) as roaming WireGuard peers
+
+The three GL.iNet routers connect to the hub as always-on WireGuard
+clients from `172.31.9.0/24` (hub = `172.31.9.1`). They are roaming
+*clients*, not sites: they initiate to the hub, carry no LAN routes of
+their own on the hub side, and NAT their attached devices behind their
+single WG address — so the DCs see one IP per router.
+
+Security model (be aware): unlike OpenVPN users there is **no Duo** here —
+possession of the router's private key is the whole credential. Each
+router has its own keypair; if one is lost, remove its entry from
+`wg_travel_peers` and re-run the playbook — that revokes it completely.
+
+## Hub side (all via Ansible)
+
+1. Generate a keypair per router — in the GL.iNet UI (it shows the public
+   key) or manually: `wg genkey | tee priv | wg pubkey`.
+2. Fill `wg_travel_peers` in `group_vars/vpnhub.yml`:
+   ```yaml
+   wg_travel_peers:
+     - { name: glinet-1, public_key: "xxxx=", ip: 172.31.9.11 }
+     - { name: glinet-2, public_key: "yyyy=", ip: 172.31.9.12 }
+     - { name: glinet-3, public_key: "zzzz=", ip: 172.31.9.13 }
+   ```
+3. `ansible-playbook site.yml` — appends the peers to wg0.conf (restarts
+   wg0: brief blip for all tunnels), adds the firewall forwards, and
+   registers `<name>.roam.atlashorizon.net` A + PTR records.
+4. **Existing deploys only**: add `172.31.9.0/24` to the Technitium
+   recursion ACL by hand (console → Settings → Recursion) — that setting
+   was seeded at first boot and isn't re-read from the compose env.
+
+## Router side (paste into the GL.iNet WireGuard client UI)
+
+```ini
+[Interface]
+PrivateKey = <this router's private key>
+Address = 172.31.9.11/24          # .12 / .13 for the others
+DNS = 172.31.9.1                  # Technitium — AD names resolve
+
+[Peer]
+PublicKey = <RACKNERD_WG_PUBLIC_KEY>      # printed by setup/playbook
+Endpoint = vpn.atlashorizon.net:51820
+# FULL TUNNEL (GL.iNet's default, and usually what you want on hotel
+# wifi): everything rides the VPN; internet egresses via the VPS
+# (masqueraded by the hub firewall).
+AllowedIPs = 0.0.0.0/0
+# Split-tunnel alternative — only AD/hub networks ride the VPN:
+#AllowedIPs = 192.168.20.0/24, 192.168.7.0/24, 172.31.9.0/24, 172.31.255.1/32
+PersistentKeepalive = 25
+```
+
+## Site side (once, both sites — same drill as the OpenVPN subnet)
+
+- Add `172.31.9.0/24` to the RackNerd peer's AllowedIPs in each site's
+  WG config.
+- Return route: `172.31.9.0/24` via the site WG box (LAN router or DCs).
+- DC/Windows firewall: allow AD ports from `172.31.9.0/24`.
+- Optional: AD Sites & Services subnet object for `172.31.9.0/24`.
+
+## Verify
+
+```bash
+wg show wg0                        # router peers: recent handshakes
+ping 172.31.9.11                   # from the hub
+dig @127.0.0.1 -x 172.31.9.11 +short   # glinet-1.roam.atlashorizon.net.
+# from a device behind the router:
+#   ping ahdc01.ad.atlashorizon.net  (resolves via 172.31.9.1, routes via hub)
+```
@@ -1,5 +1,11 @@
 # Duo publishes no official image; build from their source tarball.
 # Install flags match Duo's documented non-interactive install.
+#
+# Slimming option: switch BOTH stages to debian:bookworm-slim (~80MB less,
+# and the base layer is shared with the openvpn image). Both must move
+# together — the proxy links against the builder's glibc. Alpine/musl
+# will not work.
+#FROM debian:bookworm-slim AS build

 FROM ubuntu:24.04 AS build

@@ -20,6 +26,8 @@ RUN groupadd -r -g 35505 duo \
 && ./install --install-dir /opt/duoauthproxy --service-user duo \
      --log-group duo --create-init-script no --enable-selinux no

+# Slim variant (must match the build stage base — see note above):
+#FROM debian:bookworm-slim
 FROM ubuntu:24.04

 # gettext-base = envsubst for rendering authproxy.cfg from env vars
@@ -8,7 +8,7 @@ debug=false

 [ad_client]
 host=${TROCH_DC1_IP}
-host_2=${TROCH_DC2_IP}
+;host_2=${TROCH_DC2_IP}
 service_account_username=${DUO_SVC_USER}
 service_account_password=${DUO_SVC_PASSWORD}
 search_dn=${AD_SEARCH_DN}
@@ -22,7 +22,7 @@ ssl_ca_certs_file=/certs/ad-ldaps-ca.pem
 ssl_verify_hostname=false
 ; Restrict VPN access to one AD group (fill DUO_SECURITY_GROUP_DN in .env
 ; and uncomment):
-;security_group_dn=${DUO_SECURITY_GROUP_DN}
+security_group_dn=${DUO_SECURITY_GROUP_DN}

 [radius_server_auto]
 ikey=${DUO_IKEY}
@@ -13,7 +13,7 @@ if [ ! -f /certs/ad-ldaps-ca.pem ]; then
  exit 1
 fi

-envsubst '${TROCH_DC1_IP} ${TROCH_DC2_IP} ${DUO_IKEY} ${DUO_SKEY} ${DUO_API_HOST} ${DUO_SVC_USER} ${DUO_SVC_PASSWORD} ${AD_SEARCH_DN} ${RADIUS_SECRET}' \
+envsubst '${TROCH_DC1_IP} ${TROCH_DC2_IP} ${DUO_IKEY} ${DUO_SKEY} ${DUO_API_HOST} ${DUO_SVC_USER} ${DUO_SVC_PASSWORD} ${AD_SEARCH_DN} ${RADIUS_SECRET} ${DUO_SECURITY_GROUP_DN}' \
  < /opt/duoauthproxy/conf/authproxy.cfg.template \
  > /opt/duoauthproxy/conf/authproxy.cfg
 chmod 600 /opt/duoauthproxy/conf/authproxy.cfg
@@ -43,12 +43,28 @@
 -A FORWARD -i tun0 -o wg0 -s 172.31.8.0/24 -d 192.168.7.0/24 -j ACCEPT
 -A FORWARD -i tun0 -o wg0 -s 172.31.8.0/24 -d 172.31.255.0/24 -j ACCEPT

+# Travel routers (roaming WG peers, wg0 -> wg0) -> both sites
+-A FORWARD -i wg0 -o wg0 -s 172.31.9.0/24 -d 192.168.20.0/24 -j ACCEPT
+-A FORWARD -i wg0 -o wg0 -s 172.31.9.0/24 -d 192.168.7.0/24 -j ACCEPT
+
+# Travel routers in FULL-TUNNEL mode (AllowedIPs 0.0.0.0/0, the GL.iNet
+# default): internet egress via this box, masqueraded below in *nat.
+-A FORWARD -i wg0 -o ${PUBIF} -s 172.31.9.0/24 -j ACCEPT
+
 # Site -> client NEW connections (e.g. RDP/RSAT from a DC to a roaming
 # laptop). Off by default; uncomment if you need it.
 #-A FORWARD -i wg0 -o tun0 -d 172.31.8.0/24 -j ACCEPT

 COMMIT

+# Internet egress NAT for full-tunnel travel routers ONLY — site-bound
+# traffic (192.168.x, 172.31.x) never matches this because it leaves via
+# wg0, not ${PUBIF}. OpenVPN clients stay split-tunnel/unNATed.
+*nat
+:POSTROUTING ACCEPT [0:0]
+-A POSTROUTING -s 172.31.9.0/24 -o ${PUBIF} -j MASQUERADE
+COMMIT
+
 # ── Fallback per project brief ─────────────────────────────────────────
 # If site-side return routing for 172.31.8.0/24 becomes a pain, NAT the
 # clients behind the wg0 address instead (DCs then see 172.31.255.1):
@@ -3,7 +3,8 @@
 # paste in each site's public key afterwards.

 [Interface]
-Address = 172.31.255.1/24
+# .255 = site transfer net, .9 = roaming WG peers (travel routers)
+Address = 172.31.255.1/24, 172.31.9.1/24
 ListenPort = 51820
 PrivateKey = <RACKNERD_WG_PRIVATE_KEY>
 # wg-quick auto-adds routes for each peer's AllowedIPs — that is exactly
@@ -25,3 +26,11 @@ AllowedIPs = 172.31.255.3/32, 192.168.7.0/24
 # Deliberately NOT listing the other site's LAN under either peer:
 # Oswald<->Trochalakis traffic must stay on their direct tunnel, never
 # hairpin through this box.
+
+# ── Travel routers (roaming /32 peers from 172.31.9.0/24) ──
+# The Ansible deploy appends these from the wg_travel_peers list in
+# group_vars. For manual setups, copy this block per router:
+#[Peer]  # glinet-1
+#PublicKey = <ROUTER_WG_PUBLIC_KEY>
+#AllowedIPs = 172.31.9.11/32
+# (no Endpoint — routers initiate and keep alive from behind NAT)
@@ -3,7 +3,7 @@ FROM debian:bookworm-slim
 # libpam-radius-auth: PAM -> RADIUS bridge to the Duo Auth Proxy on 127.0.0.1
 RUN apt-get update \
 && apt-get install -y --no-install-recommends \
-      openvpn libpam-radius-auth iproute2 procps ca-certificates \
+      openvpn libpam-radius-auth iproute2 procps ca-certificates socat curl jq \
 && rm -rf /var/lib/apt/lists/*

 COPY pam-openvpn /etc/pam.d/openvpn
@@ -11,4 +11,19 @@ chmod 600 /etc/pam_radius_auth.conf
 mkdir -p /var/lib/openvpn
 [ -c /dev/net/tun ] || { mkdir -p /dev/net; mknod /dev/net/tun c 10 200; }

+# OpenVPN does NOT pass its inherited environment to hook scripts — it
+# builds a clean env for them. Hand learn-address.sh its settings via a
+# file it sources (the hook runs as nobody after the privilege drop).
+sq() { printf '%s' "$1" | sed "s/'/'\\\\''/g"; }
+if [ -n "${TECHNITIUM_ADMIN_PASSWORD:-}" ] && [ -n "${VPN_DNS_ZONE:-}" ]; then
+  {
+    printf "TECHNITIUM_ADMIN_PASSWORD='%s'\n" "$(sq "$TECHNITIUM_ADMIN_PASSWORD")"
+    printf "VPN_DNS_ZONE='%s'\n" "$(sq "$VPN_DNS_ZONE")"
+  } > /etc/openvpn/learn-address.env
+  chown nobody:nogroup /etc/openvpn/learn-address.env
+  chmod 400 /etc/openvpn/learn-address.env
+else
+  rm -f /etc/openvpn/learn-address.env
+fi
+
 exec openvpn --cd /etc/openvpn --config server.conf
@@ -0,0 +1,73 @@
+#!/bin/sh
+# OpenVPN learn-address hook: registers VPN clients in Technitium DNS.
+#   add/update <ip> <cn>  -> A <cn>.$VPN_DNS_ZONE + PTR, TTL 60
+#   delete <ip>           -> removes both records
+#
+# CRITICAL: this script must ALWAYS exit 0 — a non-zero exit makes
+# OpenVPN reject the client's address. DNS registration is best-effort;
+# it must never take the VPN down with it.
+
+OP="$1"; ADDR="$2"; CN="${3:-}"
+
+API="http://127.0.0.1:5380/api"
+
+log() { echo "learn-address: $*"; }
+log "invoked: op=$OP addr=$ADDR cn=${CN:-<none>}"
+
+# OpenVPN gives hooks a clean environment — settings come from the file
+# the entrypoint writes, not from the container env.
+[ -f /etc/openvpn/learn-address.env ] && . /etc/openvpn/learn-address.env
+
+[ -n "${TECHNITIUM_ADMIN_PASSWORD:-}" ] || { log "TECHNITIUM_ADMIN_PASSWORD not available; skipping"; exit 0; }
+[ -n "${VPN_DNS_ZONE:-}" ] || { log "VPN_DNS_ZONE not in env; skipping"; exit 0; }
+
+TOKEN=$(curl -sf -m 5 -G "$API/user/login" \
+  --data-urlencode "user=admin" \
+  --data-urlencode "pass=$TECHNITIUM_ADMIN_PASSWORD" 2>/dev/null \
+  | jq -r '.token // empty')
+[ -n "$TOKEN" ] || { log "Technitium API login failed; skipping $OP $ADDR"; exit 0; }
+AUTH="Authorization: Bearer $TOKEN"
+
+# 172.31.8.5 -> 5.8.31.172.in-addr.arpa
+PTR_HOST=$(printf '%s' "$ADDR" | awk -F. '{print $4"."$3"."$2"."$1".in-addr.arpa"}')
+
+deregister() {
+  # PTR tells us which A record this IP belonged to
+  TARGET=$(curl -sf -m 5 -H "$AUTH" -G "$API/zones/records/get" \
+    --data-urlencode "domain=$PTR_HOST" 2>/dev/null \
+    | jq -r '.response.records[]? | select(.type=="PTR") | .rData.ptrName' | head -n1)
+  [ -n "$TARGET" ] || return 0
+  curl -sf -m 5 -H "$AUTH" -G "$API/zones/records/delete" \
+    --data-urlencode "domain=$PTR_HOST" \
+    --data-urlencode "type=PTR" \
+    --data-urlencode "ptrName=$TARGET" >/dev/null 2>&1 || log "PTR delete failed for $ADDR"
+  curl -sf -m 5 -H "$AUTH" -G "$API/zones/records/delete" \
+    --data-urlencode "domain=$TARGET" \
+    --data-urlencode "zone=$VPN_DNS_ZONE" \
+    --data-urlencode "type=A" \
+    --data-urlencode "ipAddress=$ADDR" >/dev/null 2>&1 || log "A delete failed for $TARGET/$ADDR"
+}
+
+register() {
+  # sanitize CN (= AD username) into a DNS label
+  NAME=$(printf '%s' "$CN" | tr '[:upper:]' '[:lower:]' \
+    | tr -c 'a-z0-9-' '-' | sed 's/^-*//;s/-*$//' | cut -c1-63)
+  [ -n "$NAME" ] || return 0
+  curl -sf -m 5 -H "$AUTH" -G "$API/zones/records/add" \
+    --data-urlencode "domain=$NAME.$VPN_DNS_ZONE" \
+    --data-urlencode "zone=$VPN_DNS_ZONE" \
+    --data-urlencode "type=A" \
+    --data-urlencode "ttl=60" \
+    --data-urlencode "ipAddress=$ADDR" \
+    --data-urlencode "ptr=true" \
+    --data-urlencode "createPtrZone=true" >/dev/null 2>&1 \
+    || log "A/PTR add failed for $NAME/$ADDR"
+}
+
+case "$OP" in
+  add)    register ;;
+  update) deregister; register ;;   # IP reassigned to a different client
+  delete) deregister ;;
+esac
+
+exit 0
@@ -1,10 +1,14 @@
-# OpenVPN PKI — offline CA workflow
+# OpenVPN PKI — offline CA workflow (server cert only)
+
+There are **no individual client certificates** in this design — users
+authenticate with AD password + Duo. The PKI exists solely so clients can
+verify they're talking to the real server (`remote-cert-tls server`), and
+`ta.key` (tls-crypt) gates who can attempt a handshake at all.

 The CA private key **never touches the VPS**. Run easy-rsa on an offline
-machine (or at minimum a machine that isn't this server), then copy only the
-issued material here.
+machine, then copy only the issued material here.

-## One-time CA + server cert (on the offline machine)
+## One-time setup (on the offline machine)

 ```bash
 git clone https://github.com/OpenVPN/easy-rsa.git && cd easy-rsa/easyrsa3
@@ -12,39 +16,34 @@ git clone https://github.com/OpenVPN/easy-rsa.git && cd easy-rsa/easyrsa3
 ./easyrsa build-ca                       # sets the CA passphrase — keep offline
 ./easyrsa build-server-full racknerd-vpn nopass
 openvpn --genkey tls-crypt ta.key        # shared tls-crypt key
-./easyrsa gen-crl
 ```

 ## Copy to the VPS (this directory)

-| File on offline box                  | Name here     |
-|--------------------------------------|---------------|
-| `pki/ca.crt`                         | `ca.crt`      |
-| `pki/issued/racknerd-vpn.crt`        | `server.crt`  |
-| `pki/private/racknerd-vpn.key`       | `server.key`  |
-| `ta.key`                             | `ta.key`      |
-| `pki/crl.pem` (optional, see server.conf) | `crl.pem` |
+| File on offline box            | Name here    |
+|--------------------------------|--------------|
+| `pki/ca.crt`                   | `ca.crt`     |
+| `pki/issued/racknerd-vpn.crt`  | `server.crt` |
+| `pki/private/racknerd-vpn.key` | `server.key` |
+| `ta.key`                       | `ta.key`     |

 `chmod 600 server.key ta.key` after copying.

-## Per-user client certs (offline)
+## Client profile

-```bash
-./easyrsa build-client-full nick nopass
-```
+Build ONE `.ovpn` from `client/client.ovpn.template`, pasting `ca.crt` and
+`ta.key` into the inline blocks. The same profile goes to every user.

-Build the client's `.ovpn` from `client/client.ovpn.template`, pasting
-`ca.crt`, the client cert/key, and `ta.key` into the inline blocks.
-CN should match the AD username — the server logs use it
-(`username-as-common-name` keeps them consistent).
+## Revoking access

-## Revoking a client
+No client certs means no CRL — access control lives where the identity
+lives:

-```bash
-./easyrsa revoke nick && ./easyrsa gen-crl
-```
-
-Copy the fresh `crl.pem` here, uncomment `crl-verify` in `server.conf`, and
-`docker compose restart openvpn`.
+- **Disable the AD account** (or remove it from the VPN group if you use
+  `security_group_dn`) — kills new logins immediately.
+- Remove/disable the user in **Duo** as backup.
+- If a *profile* leaks (ta.key compromised): regenerate `ta.key`, replace
+  it here, restart openvpn, redistribute the profile — old profiles stop
+  working instantly.

 Everything in this directory except this README is gitignored.
@@ -1,6 +1,8 @@
 # OpenVPN road-warrior endpoint — RackNerd hub
-# Auth chain: client cert (offline CA) + user/pass -> PAM -> RADIUS
+# Auth chain: AD user/pass -> PAM -> RADIUS
 #             -> Duo Auth Proxy (127.0.0.1:1812) -> LDAPS to the DCs.
+# No individual client certs: identity is LDAP + Duo only. tls-crypt
+# (shared key in the profile) still gates who can attempt a handshake.

 port 1194
 proto udp4
@@ -8,44 +10,53 @@ dev tun0
 topology subnet

 server 172.31.8.0 255.255.255.0
-ifconfig-pool-persist /var/lib/openvpn/ipp.txt
+
+# Identity = AD username (username-as-common-name), and one user WILL
+# connect from several devices (laptop + phone). duplicate-cn allows the
+# concurrent sessions and gives each its own pool IP. Sticky per-user IPs
+# (ifconfig-pool-persist) are incompatible with that — do not re-enable.
+duplicate-cn

 # Split tunnel: only the AD sites + transfer net ride the VPN.
 push "route 192.168.20.0 255.255.255.0"
 push "route 192.168.7.0 255.255.255.0"
 push "route 172.31.255.0 255.255.255.0"

-# All client DNS goes to Technitium on this box (block-outside-dns stops
-# Windows from leaking AD lookups to the local LAN resolver).
+# ALL client DNS goes to Technitium. Deliberately NO "dhcp-option DOMAIN"
+# or DOMAIN-SEARCH here: in OpenVPN3 clients (Connect on macOS/iOS/
+# Android/Windows) those options MEAN split-DNS — pushing them scopes the
+# VPN DNS to just those domains. With only DNS pushed, 172.31.8.1 becomes
+# the default resolver for every query. ADAPTER_DOMAIN_SUFFIX sets the
+# Windows connection-specific suffix (short names like "ahdc01" still
+# resolve) without triggering split-DNS. block-outside-dns additionally
+# stops Windows from leaking queries to the local LAN resolver.
 push "dhcp-option DNS 172.31.8.1"
-push "dhcp-option DOMAIN ad.atlashorizon.net"
-push "dhcp-option DOMAIN-SEARCH ad.atlashorizon.net"
-push "dhcp-option DOMAIN-SEARCH ad.calliecorner.net"
-push "dhcp-option DOMAIN-SEARCH ad.oswaldtheotter.net"
-push "dhcp-option DOMAIN-SEARCH ad.trochalakis.net"
+push "dhcp-option ADAPTER_DOMAIN_SUFFIX ad.atlashorizon.net"
 push "block-outside-dns"

-# PKI — issued from the offline CA, see pki/README.md
+# PKI — server certificate only (offline CA, see pki/README.md);
+# clients present no certificate.
 ca   /etc/openvpn/pki/ca.crt
 cert /etc/openvpn/pki/server.crt
 key  /etc/openvpn/pki/server.key
 dh none
 tls-crypt /etc/openvpn/pki/ta.key
-# Enable once you publish a CRL from the offline CA (revoked laptops etc.):
-;crl-verify /etc/openvpn/pki/crl.pem

-remote-cert-tls client
 tls-version-min 1.2
 data-ciphers AES-256-GCM:CHACHA20-POLY1305
 auth SHA256

 # User auth via Duo (password only = push; "password,123456" = passcode)
 plugin /usr/lib/openvpn/openvpn-plugin-auth-pam.so openvpn
-verify-client-cert require
+verify-client-cert none
 username-as-common-name
 hand-window 120        # allow time for the Duo push round-trip
-reneg-sec 0            # no mid-session TLS renegotiation (would re-prompt Duo)
-auth-gen-token 43200   # reconnects within 12h reuse a token instead of a new push
+# Hourly rekey + 12h auth token: renegotiations are satisfied silently by
+# the pushed token (no mid-session Duo prompts); a fresh password + push
+# is only required once the token expires. (auth-gen-token requires a
+# non-zero reneg-sec — do not set reneg-sec 0 here.)
+reneg-sec 3600
+auth-gen-token 43200

 keepalive 10 60
 explicit-exit-notify 1
@@ -60,6 +71,16 @@ group nogroup
 persist-key
 persist-tun

+# Register each client as <username>.$VPN_DNS_ZONE (A + PTR) in Technitium
+# on connect, clean up on disconnect. Invoked via sh so the mounted script
+# needs no exec bit. NOTE: openvpn runs hooks with a clean environment, so
+# the script sources /etc/openvpn/learn-address.env (written by the
+# entrypoint from the container env) rather than reading env vars directly.
+script-security 2
+learn-address "/bin/sh /etc/openvpn/learn-address.sh"
+
 status /var/lib/openvpn/status.log 30
-management 127.0.0.1 7505
+# Unix socket (root-only) instead of a passwordless TCP port:
+#   docker compose exec openvpn sh -c "socat - UNIX:/var/lib/openvpn/mgmt.sock"
+management /var/lib/openvpn/mgmt.sock unix
 verb 3
@@ -77,4 +77,42 @@ ensure_fwd_zone ad.oswaldtheotter.net "$OSWALD_DC_IP"
 ensure_fwd_zone 20.168.192.in-addr.arpa "$TROCH_DC1_IP" "$TROCH_DC2_IP"
 ensure_fwd_zone 7.168.192.in-addr.arpa  "$OSWALD_DC_IP"

+# Primary zones for VPN client registration (openvpn learn-address hook
+# writes <username>.$VPN_DNS_ZONE A records + PTRs for 172.31.8.0/24)
+ensure_primary_zone() {
+  zone=$1
+  if zone_exists "$zone"; then
+    echo "zone $zone already exists — leaving as-is"
+    return 0
+  fi
+  echo "creating primary zone $zone"
+  curl -sf -H "$AUTH" -G "$API/zones/create" \
+    --data-urlencode "zone=$zone" \
+    --data-urlencode "type=Primary" | api_ok
+}
+
+if [ -n "${VPN_DNS_ZONE:-}" ]; then
+  ensure_primary_zone "$VPN_DNS_ZONE"
+  ensure_primary_zone "8.31.172.in-addr.arpa"    # OpenVPN clients
+  ensure_primary_zone "9.31.172.in-addr.arpa"    # travel routers
+fi
+
+# Static A + PTR for travel routers (VPN_TRAVEL_PEERS="name:ip,name:ip").
+# overwrite=true keeps this idempotent across re-runs.
+if [ -n "${VPN_TRAVEL_PEERS:-}" ] && [ -n "${VPN_DNS_ZONE:-}" ]; then
+  echo "$VPN_TRAVEL_PEERS" | tr ',' '\n' | while IFS=: read -r name ip; do
+    [ -n "$name" ] && [ -n "$ip" ] || continue
+    echo "registering travel router $name -> $ip"
+    curl -sf -H "$AUTH" -G "$API/zones/records/add" \
+      --data-urlencode "domain=$name.$VPN_DNS_ZONE" \
+      --data-urlencode "zone=$VPN_DNS_ZONE" \
+      --data-urlencode "type=A" \
+      --data-urlencode "ttl=3600" \
+      --data-urlencode "ipAddress=$ip" \
+      --data-urlencode "overwrite=true" \
+      --data-urlencode "ptr=true" \
+      --data-urlencode "createPtrZone=true" | api_ok
+  done
+fi
+
 echo "Technitium AD forwarder zones OK."
Author	SHA1	Message	Date
nick	8dedcf6101	full tunnel	2026-07-16 16:59:37 -07:00
nick	2658d7403d	travel router	2026-07-16 11:44:08 -07:00
nick	d1ea7aaff2	fix reverse dns	2026-07-16 01:14:20 -07:00
nick	c14be296a8	working ig	2026-07-16 00:54:33 -07:00
nick	b834cd06f3	sigh	2026-07-15 21:24:42 -07:00
nick	22ffe2f617	idk	2026-07-15 21:22:15 -07:00