Root cause of the .198 load spiral (2026-07-28): needs_repair() called
ensure_image_present() every 30s tick to render the expected unit, so
under IO pressure the image-existence check timed out, read as "image
missing", and a 900s podman build ran inside the PROBE while the
companion was up — each build pegging the disk that made probes fail.
- needs_repair() is now build/pull-free: unit file present → service
is-active (10s cap; a hung systemctl reads as "assume active", never
as dead) → unit matches one of the three image refs install_one could
have written → context-newer-than-image staleness only when the unit
uses the auto-built :latest.
- Per-companion 10-min repair cooldown after a failed install_one, so a
failing build retries at most every REPAIR_COOLDOWN instead of every
reconcile tick.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Bridge WebSocket patches into the resource layer: a /peer-health/<onion>
patch invalidates that peer's cloud.peer-browse entry and the federation
node list; /package-data patches invalidate the tor-services list.
invalidate() debounces 800ms and refetches only keys with mounted
subscribers, so patch storms cost one revalidation per key; the 30s
staleness reconciliation remains the backstop for unmapped data.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- Credentials: identity.list + identity.list-credentials become
useCachedResource entries; explicit reloads still toast on failure.
- OpenWrtGateway: openwrt.get-status caches in the shared store (revisits
paint the last router state instantly); the connect flow's
params/No-router-configured semantics are preserved on top of the entry.
- ContainerApps assessed and left as-is: its Pinia store already persists
across navigation, keeps last data on error, and gates the spinner on
empty — same class as Apps/Marketplace/Fleet.
This closes the B4 rollout list from docs/FIPS-UPTIME-AND-UI-STATE-PLAN.md.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The keeps-data-visible-while-refreshing tests for PeerFiles, Server, and
LightningChannels mounted without Pinia (the converted components now
pull the resources store in setup) — add createPinia to the mounts.
LightningChannelsPanel: refresh the main channel list before the closed
history so the primary entry gets the first response, and null-guard
both fetchers' response shapes.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
network summary (4-RPC allSettled aggregate), fips row, vpn peers,
interfaces, and tor services become useCachedResource entries — revisits
paint instantly, background refreshes keep content on screen. Mutations
write through the cache: DNS apply + the 15s vpn poll patch the network
aggregate via optimistic() instead of refetching all four RPCs; peer
removal filters the cached list. loading/refreshing flags derive from
entry loadState (drops the hand-rolled hasLoaded bookkeeping).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
federation nodes + dwn.status become useCachedResource entries: revisits
paint the node list instantly, `loading` fires on true first-load only
(the old showLoader semantics), the 5s poll refreshes silently like the
old surfaceErrors:false path, and explicit reloads after mutations still
surface failures in the error banner. Replaces the hand-rolled
loadNodesWithOptions SWR.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
lnd.listchannels (+summary) and lnd.closedchannels become separate
useCachedResource entries: reopening the panel paints the last channel
lists instantly and revalidates behind them; a closed-history failure
keeps its last list without touching the main view (same semantics as
the old nested try). Open/close mutations still force a refresh.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
monitoring.current/history/alerts/alert-rules become useCachedResource
entries: revisiting the page paints the last snapshot, chart, and alert
list instantly and the 5s poll revalidates behind them (refreshes dedup
in the store; errors keep last-known values instead of blanking).
Alert-rule toggles and acknowledgements refresh their entries after the
mutation as before.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- lnd.getinfo and wallet.networking-profits become useCachedResource
entries (web5.lnd-info / web5.networking-profits): revisits paint
instantly from cache, errors keep last-known values, refreshes dedup.
- walletConnected is now derived from the lnd-info entry (with a manual
disconnect override preserving the connect/disconnect toggle).
- Drop the eager wallet.ecash-balance + lnd.gettransactions loaders and
their 30s polling — they fed only the hidden wallet card; the lnd-info
poll remains for the connected pill until B5 moves it to WS-push.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- PeerFiles.vue reads the SAME `cloud.peer-browse:<onion>` entry Cloud.vue's
per-peer fan-in fills, so Cloud → peer files paints instantly from cache
and revalidates behind it; catalog/error/loading/transport are now
computed views over the store entry.
- preview-peer fan-out is capped at 3 concurrent with a queue (was one 30s
RPC per media item, all at once, unbounded) and aborts on unmount.
- browse + preview RPCs drop to maxRetries:1 — retry×3 turned one slow
peer into a 90s spinner.
- fix useCachedResource's interface types: `ReturnType<typeof computed<T>>`
resolves to the writable overload (WritableComputedRef), which broke
vue-tsc against the plain computed() returns; use ComputedRef<T>.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Phase A3 of docs/FIPS-UPTIME-AND-UI-STATE-PLAN.md (RC5), measured against
the A2 dial_stats baseline:
- 25s connectivity watcher in the fips supervisor: re-applies seed anchors
immediately on an anchor-link drop, on startup-disconnected, AND on
silent data-path death (connect_fails growing with zero fips_ok — the
live .198 failure where the daemon reported "connected" while every
dial blackholed and the 300s tick never healed it). Bounded to one
re-apply per 60s.
- anchors::apply is now concurrent with a 15s per-connect cap — the old
serial loop waited unbounded on each `sudo fipsctl connect`, so one
hung subprocess stalled the whole periodic tick.
- rebindable peer listener: the accept loop returns after persistent
accept errors (was: continue forever = inbound-dead until restart) and
peer_late_bind_loop rebinds — also on fips0 ULA change.
- is_service_active gets a 10s TTL cache (was up to 2 systemctl spawns
per dial attempt and per warm-tick peer).
- the warm tick now warms the union of federation peers + configured
seed anchors (direct anchor links used to go cold between 300s ticks),
skipping the redundant per-peer service check.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Part B3 of docs/FIPS-UPTIME-AND-UI-STATE-PLAN.md — the worst
fetch-on-every-navigation offender converted to the cached-resource layer:
- section counts / peer nodes / my files / paid items are cached resources:
revisits paint instantly, refresh happens behind the content
(sticky-ready), errors keep last-known data
- peer files: per-peer cached browse entries replace the all-or-nothing
Promise.allSettled — each peer's rows render the moment it answers, with
"still fetching from N peers" + unreachable counts; browse-peer runs
with maxRetries:1 so one dead peer costs its timeout once, not ×3
- peer cards get a live transport badge (FIPS green / Tor amber, with
measured latency) from the transport field the browse response already
carried — the per-peer FIPS-uptime view, for free
- cloud store: per-path listing cache with stale-while-revalidate
navigate() and a last-wins guard; CloudFolder no longer reset()s the
store on every folder entry (that wipe forced a spinner each time)
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Part B1+B2 of docs/FIPS-UPTIME-AND-UI-STATE-PLAN.md. Foundation for pages
that render instantly from cache on revisit and revalidate in the
background, instead of unmount-refetch-spinner on every navigation.
- stores/resources.ts: keyed {data, loadState, fetchedAt, error} entries
with sticky-ready (never regress ready→loading), keep-last-value on
error, per-key in-flight dedup, sessionStorage snapshot hydrate,
debounced invalidate() fan-out, optimistic-update-with-rollback
- composables/useCachedResource.ts: SWR hook over the store — synchronous
hydrate, TTL-gated background revalidate, revalidate-on-focus,
abort-on-unmount fetcher signal
- rpc-client: AbortSignal support (aborts pending retries too), opt-in
in-flight dedup keyed method+params, per-call maxRetries override
- 10 tests covering the SWR semantics
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Phase A2 of docs/FIPS-UPTIME-AND-UI-STATE-PLAN.md (RC6). Fallbacks to Tor
were debug!-only and uncounted, so "FIPS uptime" was unfalsifiable and
paths that were 100% Tor by construction went unnoticed for months.
- fips::telemetry: process-lifetime counters for FIPS successes and the
six fallback reasons (no_npub, service_inactive, dns_fail, connect_fail,
http_404, http_5xx), exposed as `dial_stats` in fips.status
- dial.rs: every fallback branch now counts + logs at info! with a
`reason` field (resolve/connect/status branches)
- PeerRequest::record_transport(data_dir): opt-in hook that writes the
transport actually used to federation storage off the hot path — wired
into the dial sites that never recorded (DWN sync ×3, mesh blob fetch,
federation deploy notify, onion-rotation notify, node messages via a
new send_to_peer data-dir param)
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Phase A1 of docs/FIPS-UPTIME-AND-UI-STATE-PLAN.md — the five changes that
made FIPS fall back to Tor even when a FIPS path existed:
- RC0: the fips.d drop-in now opens PEER_PORT 5679 (was 80+8443 only, so
every hardened node firewalled peers' FIPS dials; 28k drops on .198)
- RC4: /blob/ and /dwn/ added to the peer-path allowlist — mesh file
sharing and DWN sync were 404 → 100% Tor by construction
- RC2-G2: lan_fips_anchors dials PUBLISHED_UDP_PORT (2121) instead of the
dead 8668, with a drift-guard test against the rendered daemon config
- RC2-G1: direct LAN peering actually runs now — mDNS TXT advertises the
FIPS npub, discovery calls set_fips_npub, and the anchor tick hydrates
npubs from federation storage for peers on older builds
- RC3: FIPS attempt budget is a hard cap (retry no longer doubles it) and
the 12 hot call sites get explicit fips_timeout fast-fail so Tor keeps
its full budget (browse-peer, preview, /blob, DWN, node-message,
rotation notifies)
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Consolidated plan from the deep repo review: credential rotation, secret
scrub, repo restructure, registry parameterization, docs overhaul, deep
code cleanup, and fresh-history publish mechanics.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The dashboard's three-finger hold now opens the menu overlay in place
instead of jumping to the remote screen; the hub's Remote and Keyboard
cards do the navigating (Keyboard lands directly in keyboard mode via a
nav arg). The dashboard host carries the full Nodes page — add, edit,
remove and QR pairing — and Mesh Party.
Also: the panel scales to 92% of screen height instead of a fixed 560dp
cap so pages fit without scrolling; the FIPS sub-page no longer repeats
the FIPS Mesh header inside the detail card; the safe-area injection
fires an archy-insets event the web UI listens for (update-install
status-bar fix, other half in neode-ui).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The mobile nav sampled --safe-area-top once at mount, but the companion
WebView injects it asynchronously. An authenticated session mounts the
dashboard before the injection lands (fresh installs mount after login,
long after it), so the content padding baked in 0 while the fixed tab
bar grew by the real inset — content slid underneath by exactly the
status-bar height, the update-install-only overlap.
Re-read on the WebView's new archy-insets event, with a retry ladder as
fallback for APKs that predate the event.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Three-finger menu becomes a contained card hub: Dashboard, Remote,
Keyboard, Nodes, FIPS Mesh, Mesh Party — with Nodes and FIPS as
sub-pages behind a back header. The panel is a centred glass card that
scrolls within its own bounds instead of filling the screen.
The Dark/Classic style toggle leaves the menu and becomes a palette
button next to the settings gear on all three input surfaces (landscape
controller, portrait controller, keyboard mode).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
When neither the LAN origin nor the mesh ULA answers the probe window,
target the mesh URL (when paired) instead of the LAN IP: off-LAN the LAN
address can never answer, and loading it showed a confusing 'can't reach
192.168.x.x' error while the mesh session was still coming up.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Root cause (diagnosed live via CDP on a Pixel 9a): camera permission
granted and the native camera streamed (2m35s active) but ZXing returned
zero decodes for Lightning-invoice QRs — window.__archyQrResult received
0 calls. Dense BOLT11 codes need more pixels/module AND sharp focus; the
9a's main lens rests at a far focus and 1280x720 wasn't enough, so dense
invoices never resolved while sparse address QRs did. Fix: analysis at
1920x1080 + a repeating centre autofocus tick (a static hand-held QR
never retriggers continuous-AF on its own).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Slow multi-hop payments (>15s routing) surfaced as "Payment failed"
while LND settled them in the background: the shared LND REST client's
15s total timeout aborted the synchronous /v1/channels/transactions
wait, and every UI path treated that abort as a definitive failure. The
payment then succeeded anyway and only appeared in history on the next
background poll.
Backend: lnd.payinvoice now decodes the invoice up front for its payment
hash, pays on a dedicated 120s client, and answers status:"pending" with
the hash (never an error) when the wait elapses after the payment was
handed to LND — only a pre-connect failure is still a hard error. New
lnd.paymentstatus RPC reports succeeded/failed/in_flight (with humanized
failure reasons) from /v1/payments.
Frontend: new rpcClient.payLightningInvoice() pays then polls
lnd.paymentstatus to a real terminal state (3s interval, up to 2 min);
all five call sites (send modal, scan modal, web5 unified send, peer-file
purchase, app-launcher payments) migrated. Failure is only shown when LND
itself declares FAILED; a still-settling payment shows an in-flight state
and success fires the transaction refresh.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- In-app browser pages (node apps) now start below the status bar again:
the WebView stays edge-to-edge (page colour fills the bar) and a
body{padding-top:<statusbar>} injection pushes content down — the
pre-edge-to-edge look without the black bar.
- Add/Edit server in the menu now has a Use HTTPS toggle (was scheme-locked;
edit preserved the old scheme, add forced http).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- Transport handoff (Wi-Fi ⇄ 5G, BLE-ready): ArchyVpnService registers a
ConnectivityManager callback that re-pins the tunnel to the new default
network (setUnderlyingNetworks) and re-homes the mesh (fresh warmer pass
→ discovery/sessions rebuild on the new path in seconds). Previously the
tunnel stayed pinned to the network it launched on and roaming stranded
the mesh until an app restart.
- FIPS Mesh section in the NESMenu settings modal: status, mesh address
(fd… ULA), identity npub (both copyable), configured peer/anchor count,
and a one-tap Reconnect (manual re-home). Gives users oversight of what
the embedded mesh node is doing.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Bundles all post-115 fixes into an OTA release: boot READY-before-recovery
+ Restart=always (no more 'server starting up'), and the app-install
daemon-kill fix (v6 relay only bridges live app ports; port-cleanup
excludes our own PID).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The v6 app-port relay preemptively bound [::]:<port> for ALL catalog
ports, even apps not installed. Installing such an app (grafana:3000,
photoprism, uptime-kuma, jellyfin — framework-pt 2026-07-27) then hit
'address already in use' from the relay, which triggered
cleanup_stale_pasta_port -> -> killed archipelago
itself (it held the port) mid-install. The daemon crash-looped and the
half-created apps were rolled back and vanished.
Two fixes:
- relay only bridges a port that a running app already answers on over
IPv4 (probe 127.0.0.1:port first) — an uninstalled app's port is never
held, so its install sees a free port and never triggers the cleanup.
- cleanup_stale_pasta_port excludes our own PID from both the ss-based
kill and the fuser kill, so freeing a port can never terminate the
daemon even when the relay legitimately holds it (reinstall case).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Root cause of 'server starting up' forever / crash-on-install
(framework-pt, v1.7.114->115, 2026-07-26): on a node with many stacks,
the synchronous boot recovery (recover + start_stopped_containers) runs
BEFORE sd_notify(Ready), so the unit sits in 'activating' for minutes.
Anything touching the service in that window — a superseding
start/restart, an install-time reconcile churn — killed a half-started
instance; it exits 0 on SIGTERM and Restart=on-failure then never
restarts it. Node dead behind 'server starting up'.
Fixes:
- signal READY (+ start the watchdog keepalive) BEFORE boot recovery, so
the unit reaches 'active' in seconds; recovery/reconcile/listener
continue after. No more minutes-long activating window.
- Restart=always (was on-failure): a clean-exit SIGTERM must still bring
the daemon back. Manual is still honored.
- OTA restart via a PID1-owned transient timer (systemd-run --on-active=2)
instead of a tokio-sleep child of the process being stopped, whose
start-half was being lost.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- app webview draws behind the status bar again (the inset rework
painted an opaque black bar there); page background owns the top
- the in-app overlay eats EVERY touch its children don't handle —
a near-miss on Close was falling through to the kiosk AIUI tab
- loading screen never titles itself with a raw mesh IPv6/IP host
- re-land vc37 connect fixes vc38 shipped without: pairing restarts
the mesh immediately when consent exists; session warmer probes all
targets concurrently (5s) instead of 20s each in sequence
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Signed OTA manifest (release root verified). Also: create-release-manifest
tarball perms check made SIGPIPE-proof so releases cut on macOS too.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Rootless podman's wildcard publish claims [::] but BLACK-HOLES inbound
v6 (accepts, forwards nothing — vaultwarden 'empty response' from the
phone, 2026-07-26), while most apps got no v6 listener at all. Two
halves: quadlets now publish unbound ports on 0.0.0.0 explicitly
(frees the v6 side; the one-time drift/recreate at upgrade is the
deploy vehicle), and the daemon runs a self-selecting V6ONLY relay on
every catalog launch port forwarding raw TCP to the v4 loopback
listener. Rescans every 60s so new installs bridge without a restart.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The companion opens catalog apps by direct port over the mesh; the
fips0 default-deny baseline blocked every one of them (apps 'stuck'
from the phone, 2026-07-26). Core now writes a second fips.d drop-in
from the generated catalog launch-port list on every install/upgrade.
Service/RPC ports (bitcoind 8332, LND 10009, Tor) stay closed.
generate-app-catalog.py emits the Rust port list alongside the TS one.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
fips::config::install() now writes /etc/fips/fips.d/80-web-ui.nft
(tcp 80/8443 accept) and reloads the baseline on every install/upgrade —
the hardening firewall default-denies inbound on fips0 and the UI was
unreachable over the mesh without it (root-caused live 2026-07-26).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The mesh is IPv6-only; the main web listener bound 0.0.0.0 only, so a
phone reaching a node over its fips0 ULA got RST on :80
(ERR_CONNECTION_ABORTED) — UI over mesh was structurally impossible.
Confirmed 2026-07-26 on framework-pt: v4:80 = 200, v6:80 = refused.
Mirror an IPv4-any main listener with a V6ONLY [::] socket on the same
port; V6ONLY so it coexists with the v4 listener regardless of
net.ipv6.bindv6only.
Also: fips.yaml generator carries the fast-reconnect profile (validated
live on framework-pt since 2026-07-24; snapshot tests updated).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
