Files
ducoterra 3cdbb405ab add WireGuard wg0 watchdog units for always-on SSH
When SSH to a Pi targets a WireGuard tunnel address, reachability depends on
three things that each fail independently: the unit starting at boot, the NAT
mapping staying open, and the peer hostname still resolving. Rebooting onto a
network with no route to the endpoint previously meant no way back in.

- wg-quick@wg0.service.d/10-boot-resilience.conf: retry a failed boot-time up
  forever (Restart=on-failure, RestartSec=15, StartLimitIntervalSec=0) and cap
  endpoint resolution (WG_ENDPOINT_RESOLUTION_RETRIES=60, TimeoutStartSec=120)
  instead of hanging on the stock infinity.
- wg0-watchdog: POSIX sh healer, oldest symptom first - missing interface or
  peer restarts the unit, a missing PersistentKeepalive is set directly, a
  stale handshake re-points the peer at a freshly resolved endpoint, and a
  very stale one restarts. Restarts are gap-limited so a broken network cannot
  make the unit flap, endpoint resolution falls back to the kernel endpoint
  and then to a last-known-good cache (tunnel DNS may only be reachable
  through the tunnel), and nothing it did not create is ever removed.
  Supports --dry-run.
- wg0-watchdog.service/.timer: oneshot on a 60s timer, first check 30s after
  boot, Persistent=true to catch up after downtime, TimeoutStartSec=120 and
  Nice=10 so a restart cannot wedge the watchdog.
- wg0.conf.template: documents /etc/wireguard/wg0.conf with placeholders only.
  Deliberately omits the DNS key so name resolution does not depend on the
  tunnel (Debian wg-quick has no `DNS = none`), and sets PersistentKeepalive
  25 to hold the NAT mapping for inbound SSH.
- wireguard.md: runbook for the placeholder table, failure modes, deploy, and
  verification.

The real wg0.conf and key material never enter the repo.

LLM-Generated: true
2026-10-16 09:07:56 -04:00

21 lines
902 B
Plaintext

# wg0.conf template — /etc/wireguard/wg0.conf on the host, mode 0600 root:root.
# Never commit a real copy: this file holds the tunnel private key.
# Generate the key with `umask 077; wg genkey | tee privatekey | wg pubkey`.
[Interface]
PrivateKey = <tunnel private key>
Address = <TUNNEL_ADDR>
# Deliberately no `DNS` key: any resolver listed here is pushed as a system
# upstream while the tunnel is up, which makes all name resolution depend on the
# tunnel. Omitting it keeps the LAN's DHCP resolvers. Debian's wg-quick has no
# `DNS = none` - a non-IP value becomes a search domain.
[Peer]
PublicKey = <PEER_PUBKEY>
AllowedIPs = <PEER_NETWORK>
Endpoint = <WG_SERVER_HOST:PORT>
# Required when the client is behind NAT and must be reachable inbound (SSH):
# without it the mapping expires and the server cannot initiate. 25s survives
# typical 30-60s NAT UDP timeouts.
PersistentKeepalive = 25