Package the PineVoice/Wyoming local voice assistant as ONE store app "Pine"
that runs three rootless containers:
- pine-whisper — Wyoming faster-whisper STT (docker.io/rhasspy/wyoming-whisper
3.4.1, model base-int8/en), publishes :10300
- pine-piper — Wyoming Piper TTS (2.2.2, voice en_GB-alba-medium),
publishes :10200
- pine — nginx launcher serving a self-contained setup/status page
(WiFi provisioning + Home Assistant wiring instructions), the Open target
The two engines publish their Wyoming ports so Home Assistant reaches them via
host.containers.internal. All rootless (cap-drop=ALL, no_new_privileges),
manifest-driven — no hardcoded podman/installer. The engines are internal stack
members (added to the drift-check INTERNAL_MANIFEST_IDS allowlist); only "pine"
carries a catalog entry.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Supersedes the unpublished v1.7.107. Covers the multi-anchor FIPS fix, WiFi
self-heal, mesh fast-rejoin, kiosk TV-scaling, kiosk dark theme / VT hints /
UTF-8 logo / power-button hardening, and the ISO build fixes.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Three fresh-install kiosk issues found on a Framework node:
- MOTD said 'Ctrl+Alt+F7 for kiosk' / 'F1 for terminal', but the kiosk's
Xorg runs on vt1 — so F7 was an empty black VT (the reported black screen
and the failed 'return to kiosk'). Corrected: kiosk=F1, terminal=F2.
- The UTF-8 block-glyph logo corrupted intermittently because returning from
the kiosk can leave the console VT out of UTF-8 mode; force it with ESC%G.
- A short power-button press shut the node down instantly. Ignore the short
press, keep long-press poweroff, so accidental brushes don't kill the node.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Nodes depended on a single public anchor (185.18.221.160) that is
unreachable from most real networks — confirmed on the thinkpad WiFi and a
fresh Framework node, both of which islanded (is_root=true, depth=0). Its
DNS even resolves IPv6-first while the daemon is IPv4-only.
Stand up an Archipelago-operated FIPS anchor on vps2 (146.59.87.168, the
OTA host every node already reaches) and ship it as an additional default
anchor. default_public_anchor() -> default_public_anchors() -> Vec; load()
returns the set; the apply loop already dials each, so whichever the node's
network can reach wins. Verified end-to-end: a fresh node joined the tree
via vps2 in 3s (depth 1). anchors tests: 7/7 pass.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
AIUI themes via @media (prefers-color-scheme) and falls back to its light
variant (white panels) when the browser reports no preference — which the
minimal kiosk X session does, so AIUI came up light instead of its designed
dark. The main UI hardcodes dark and is unaffected. Push two independent
dark signals on the kiosk Chromium: GTK_THEME=Adwaita:dark (version-
independent, can never force light) and --blink-settings=preferredColorScheme=0
(0 = kDark in modern Chromium). Needs on-device kiosk verification.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Two fleet fixes that OTA can deliver (unlike the ISO-only polkit rule):
- WiFi (issue #99): nodes drive NetworkManager from a system-level service
with no logind seat, so the stock polkit rule denies them and Wi-Fi setup
fails with 'Insufficient privileges'. Fresh ISOs since 2026-05 ship the
scoped rule, but OTA'd nodes never got it (OTA replaces binary + web UI,
not host system config). Add a startup self-heal that installs the rule if
missing and best-effort ensures polkitd — both wrapped so an offline/locked
apt can never fail startup; the rule is written regardless so it activates
once polkitd lands. Idempotent on the rule's unique subject.user marker.
- FIPS: regenerating fips.yaml (this build does, once, on first boot after
the OTA) restarts the fips daemon; for a few seconds /run/fips/control.sock
is gone and every seed-anchor dial fails, islanding the node until the next
5-min tick. Detect that exact failure (all dials fail on control.sock) and
retry in 15s instead — bounded to 8 fast retries/episode so a node with no
fips daemon falls back to the steady cadence rather than busy-looping.
FIPS unit suite: 33/33 pass.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The relay image was rebuilt to the standard nostr-rs-relay layout — WORKDIR
/usr/src/app, execs ./nostr-rs-relay — but the build still cp'd from the old
/usr/local/bin path, so extraction silently produced nothing. Once the
missing-binary guard was added, that silent miss became a hard build
failure. Verified the binary is a valid amd64 ELF (GLIBC 2.34, under Debian
13's 2.40) at /usr/src/app/nostr-rs-relay.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The ISO build hard-failed at 'Required binaries not extracted: nvpn'
because it still pulled 146.59.87.168:3000/lfg2025/nostr-vpn:v0.3.7 — an
image that was deleted from the registry when NostrVPN was removed from
the product. The native nvpn daemon is already masked here (ln -sf
/dev/null nostr-vpn.service) and is never spawned at runtime (core vpn.rs
manages WireGuard/Tailscale, only reading a legacy nvpn config as
fallback), so extracting its binary was dead weight that bricked every
build once the image went away.
Drop the nvpn extraction and remove it from the missing-binary guard.
nostr-rs-relay stays required — its nostr-relay unit is still enabled and
the image still pulls. Change is in STEP 3, so the cached rootfs is
unaffected.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The ISO built FIPS from an unpinned --depth 1 clone of upstream main, so
every build shipped whatever happened to be on main that day. Pin to
v0.4.1 via a FIPS_VERSION build arg — the version fips/config.rs renders
its typed config against, and the one validated in the field on .198 and
the thinkpad. The pin lands inside the STEP 1 recipe-hash range, so the
cached rootfs correctly invalidates on the next build.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Captures the three commits this release carries, the FIPS 0.4.1 upgrade
state (2 nodes done, fleet not rolled), the peer-files diagnosis including
what is explicitly NOT explained, the open decisions not taken, and the
exact release + ISO ritual.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The daemon config was built by format!-ing a YAML string literal. Upstream's
config structs are #[serde(deny_unknown_fields)], so a key we get wrong does
not degrade — the daemon refuses to start and the node drops off the mesh.
String-built config made that a runtime discovery on a live node.
Replace it with a typed struct tree serialised via serde_yaml, verified
field-by-field against jmcorgan/fips v0.4.1, plus an exact-output snapshot
test so schema drift fails in CI rather than at boot. Also adds tests for
determinism (server.rs compares the render against disk to detect drift, so
instability would cause a reinstall+restart loop) and for the mDNS key path.
Enables node.discovery.lan.enabled (mDNS/DNS-SD, added upstream in v0.4.0)
so co-located nodes peer directly instead of depending on the public anchor
being reachable — an anchor blackhole on one segment currently islands a node
completely. Emitted unconditionally rather than version-gated: v0.3.0's
DiscoveryConfig has no `lan` field and no deny_unknown_fields, so a v0.3.0
daemon ignores it harmlessly and it activates on upgrade with no second
config migration.
Note: the first startup after this lands renders a config that differs from
disk, so the existing drift check reinstalls it and restarts the daemon once.
That is the intended self-healing path and settles immediately.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
RPC failures logged only the top-level anyhow context, so a peer-files
failure produced exactly "RPC error on content.browse-peer: Failed to
connect to peer" with the real cause discarded. That made it impossible
to tell a dead peer from a slow Tor circuit from a FIPS resolve failure
without reproducing by hand.
Switch the log line to `{:#}` so the whole context chain is rendered.
The client-facing message still uses `{}` through sanitize_error_message,
so no internal detail is leaked to callers.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The peer-files header title block is `hidden md:block` (the global header
carries the peer name on mobile), which also hid the transport pill nested
inside it — so mobile users browsing a peer's files had no indication of
whether they were on FIPS or Tor.
Render a `md:hidden` pill alongside the peer icon so the transport is
visible at every width, without duplicating the peer name on desktop.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
vue-tsc in the release build (unlike the gate's type-check) rejects
indexing WG_RETRY_DELAYS_MS with a bare counter; hoist the delay into a
local and gate on undefined instead of the length check.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
vpn.peer-config returned the config exactly as stored at creation time, so
after a node moves networks the reused companion peer's QR encodes the old
location's address (seen on .116: Endpoint = 10.125.9.0 from the previous
LAN) and the tunnel can never connect. Rewrite the Endpoint line with the
node's current address before rendering the QR, and persist it back so the
downloadable .conf matches. Endpoint detection is shared with
vpn.create-peer via current_wg_endpoint_host().
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
On a fresh node the wgqr step is often reached while the backend is still
settling (services starting, backend restarting during orchestration): a
single 'Failed to fetch' left a permanently blank QR. Retry network-class
failures on a 2s/4s/8s ladder while the step is still on screen, then fall
back to the friendly error + Try again button.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
After an unclean shutdown, crash recovery walked the running-containers
snapshot and retried 'no such container' failures (2 attempts, 10s
backoff) for containers that no longer exist. Recovery runs before the
server binds :5678 and notifies systemd ready, so a stale snapshot
pushed startup past TimeoutStartSec=5min — systemd killed the daemon
mid-recovery, the next boot saw a crash again, and the node looped
forever (151 restarts on .116 after a hard poweroff).
- pre-filter the snapshot against 'podman ps -a' and skip vanished
containers outright (fail-open if the query fails)
- never retry a 'no such container' failure
- extend the systemd start timeout ahead of each container start so a
heavy node recovering dozens of real containers isn't killed while
making progress
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Backend: tor.list-services now hides services whose name maps to a
known-but-uninstalled catalog app (aliases covered: bitcoin/knots/core,
electrumx/electrs, btcpay, mempool). The node's own service, the relay,
and custom user services always show.
ISO: first-boot tor setup no longer pre-creates hidden services for six
apps that may never be installed — only the node's own onion is baked.
It also only seeds services.json/torrc on FIRST boot instead of
clobbering backend-managed services on every boot.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
v1.7.104 shipped ISOs whose units crash-looped (nostr-relay 3s loop,
archipelago-diag 203/EXEC) because build-time extraction failures were mere
warnings and the cached rootfs never tracked its recipe:
- hash the rootfs-defining region of the build script and rebuild when it
changes (stale cache shipped ISOs without wpasupplicant/iw/rfkill)
- refuse to build without nvpn / nostr-rs-relay unless
ALLOW_MISSING_VPN_BINARIES=1
- ConditionPathExists on nostr-relay/nostr-vpn/diag units so a stripped
image skips them instead of crash-looping
- post-install test: every enabled archipelago*/nostr* unit must have an
existing ExecStart payload
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The sweep podman-execs into every running container; doing that on each 30s
reconcile tick was a permanent conmon 'Failed to create container' storm on
hosts where exec from the backend's cgroup context fails (Debian 13 first
boot). Ownership drift is an install/OTA-time event — sweep on first pass
after a container appears, then at most hourly.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
With :global(html.kiosk-mode) .bg-layer the SFC compiler drops the
descendant part and emits bare html.kiosk-mode rules — including
display: none !important — blanking the whole document on kiosk.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- Never show the get-the-app pitch inside the companion WebView itself
- Don't guess peer-exists when list-peers is unreachable: try create,
fall back to peer-config on already-exists errors
- Translate raw 'Failed to fetch' into an actionable network hint and
add a Try again button instead of a dead-end error
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
neode_intro_seen / neode_onboarding_complete are per-origin browser state:
after a reinstall (or another node on a DHCP-recycled IP) they describe the
previous node and muted a genuinely fresh install's intro. Root boots now
always ask the backend; a confirmed-fresh answer plays the intro and clears
the stale flag.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
A stray `archipelago --version` used to start a full second instance next
to the systemd one (issue #74). Handle -V/--version, -h/--help, and reject
unknown dashed options before any tracing/state init, mirroring the
ceremony subcommand's clean-stdout precedent. Version output matches the
health RPC format: <pkg-version>-<git-hash|dev>.
Fixes#74
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Both tests take the global single-flight UPDATE_OP_LOCK; run concurrently
by the test harness, one saw the other's lock and failed with 'another
update operation is already running' instead of its expected refusal.
A shared test mutex makes them mutually exclusive deterministically.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The pre-UI sync overlay used a hardcoded orange box-glyph SVG; it now renders
the same resolved app icon the loading screen uses (with the shared image-error
fallback), so the wait page reads as ElectrumX at a glance.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The manifest baked bitcoin-knots:8332 into DAEMON_URL. Nodes whose
backend runs under the bitcoin-core name (and any future variant
without a knots DNS alias) left electrumx permanently disconnected —
'connection problem' forever and a block index stuck at 0. The
startup script now picks the first backend name that resolves on
archy-net and falls back to bitcoin-knots.
Catalog regenerated (catalog manifests override disk ones fleet-wide);
this regen also embeds the recently-merged barkd/Ark manifest for the
first time.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
.198 crash-looped 236x on a truncated OTA binary with a good backup
sitting in update-backup/ — verify_pending_update() can't help when
the new binary never runs. New scripts/ota-crash-guard.sh runs as
root from ExecStartPre: while the post-OTA pending-verify marker
exists it counts start attempts, and after 5 failures restores the
backup binary (atomic rename), replaces the marker with an
update-rolled-back.json tombstone, and lets the service come back
on the previous version.
Wiring: fresh ISOs get the ExecStartPre line in the unit file;
existing nodes get a drop-in installed by apply_update's runtime
component step on their next OTA. '+-' prefix so a missing or
failing guard can never block the service.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Root cause of .198 bricking on the 1.7.103 OTA: two concurrent
update.download RPCs shared one staging file, cancel_download wiped
staging mid-flight, a third download began re-filling it, and
apply_update mv'd the 3-second-old 17MB partial of the 49MB binary
into /usr/local/bin -> SEGV boot loop (236 restarts, no rollback).
- Single-flight UPDATE_OP_LOCK across download/apply; concurrent
callers get an explicit 'already running' error.
- apply_update now requires the .download-complete marker AND
re-verifies every staged component (size + SHA-256 + BLAKE3)
against the manifest before touching the system.
- cancel_download only wipes staging when no operation holds the
lock; otherwise it just flags the in-flight loop to bail.
- Fixed the 'file already complete' path in
download_component_resumable, which skipped verification and fell
through to the bogus 'download failed without a captured error'.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The SW-update modal is noise on the demo (nothing to update to) and both
accept-paths end in a reload that replays the intro from scratch. With
skipWaiting/clientsClaim off, ignoring the waiting worker is safe — the
new build activates on the visitor's next session.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Dispatch the Enter keydown two frames after the input event so the login
button is enabled when it arrives. Keeps auto-login working against nodes
running older web builds where controller-nav's Enter-in-input pattern
would otherwise click Replay Intro (see the matching web-ui fix).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The companion's auto-login fills the password and dispatches Enter in the
same tick, before Vue re-enables the disabled submit button. Controller-nav's
'Enter in input clicks the next enabled button' pattern then hit the Replay
Intro button — clearing neode_intro_seen and hard-navigating to /, so every
app connect replayed the intro cinematic in a loop. The auth inputs all have
their own Enter handlers, so they opt out via data-controller-no-submit.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
