fix(deploy): trust the Forgejo host key so auto-upgrade can fetch

system.autoUpgrade fetches the flake over ssh as root. A machine whose
root has never connected by hand has no known_hosts entry, so the run
dies at 'Host key verification failed' before it even reaches
authentication. batm3 did exactly that, silently, from its 2026-08-06
install until 09-22: six weeks on its install generation while a unit
nobody was watching reported failure every night. sintra only ever
worked because a human had ssh'd as root once and accepted the key.

Declaring the key means a freshly flashed ATM updates from first boot
with no manual step. Verified against the key sintra's root already
trusts.

Refs #98

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This commit is contained in:
Padreug 2026-09-22 20:14:25 +02:00
commit 1405ad224c

View file

@ -159,6 +159,18 @@
# Auto-updates (optional - disabled by default for stability)
# system.autoUpgrade.enable = false;
# Trust the Forgejo host key up front. system.autoUpgrade fetches the flake
# over ssh AS ROOT, and a machine whose root has never connected by hand has
# no known_hosts entry, so every nightly run dies at
# "Host key verification failed" before it reaches authentication. batm3 did
# exactly that, silently, from its 2026-08-06 install until 09-22 (#98): it
# sat on its install generation for six weeks while reporting a failed unit
# nobody was watching. sintra only ever worked because a human had ssh'd as
# root once and accepted the key. Declaring it means a freshly flashed ATM
# can update from first boot with no manual step.
programs.ssh.knownHosts."git.atitlan.io".publicKey =
"ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAIMlo3f05o4+bk0+8x2VG91o9GubshOb46HmBPvND9pJx";
# pragma: allowlist secret
# Ensure WireGuard private key directory exists with correct permissions
system.activationScripts.wireguard-key = ''
@ -169,6 +181,20 @@
fi
'';
# The tunnel is operator-provisioned: wg0.key is written per machine after
# flashing, and until it is, `wg set … private-key` exits 1 with
# "fopen: No such file or directory". One failed unit makes
# switch-to-configuration exit 4, which marks the entire nightly
# system.autoUpgrade run as failed — so an ATM that simply never had its
# tunnel provisioned reports a broken updater for the life of the machine
# (sintra, #98). Skip the unit when there is no key instead of failing
# activation over an interface that was never set up; a provisioned machine
# is unaffected. Guarded on wg0 still being declared so the live image,
# which mkForce's the interfaces away, doesn't get a unit with no ExecStart.
systemd.services = lib.mkIf (config.networking.wireguard.interfaces ? wg0) {
wireguard-wg0.unitConfig.ConditionPathExists = "/var/lib/wireguard/wg0.key";
};
# In-place rename migration: lamassu user → bitspire user.
# Runs after `users` activation so the bitspire user exists with its UID.
# Idempotent: re-running on an already-migrated system is a chown no-op.