resticStep escalates a restic lock error to `unlock --remove-all` + one
retry (safe: single-writer repo — sub-account isolation + single-flight
mutex); plain `unlock` is stale-only and can't clear a crash lock across a
container-hostname change. Pre-run stale unlock hygiene on run+restore.
C1: NewManager flips a persisted LastStatus=running to a truthful error.
Both red-proofed (A reproduces the exact campaign backup failure).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PSK5g6qYLknKj8u3QAFEr6
The QuotaGB hash change triggered a live re-apply that demoted the
escrowed demo to pending and wiped its runtime status. ApplyOffsiteTarget
now carries over EscrowState (custody tracks the preserved repo password,
not the coords) + status fields; fresh guests still land pending.
Red-proofed.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PSK5g6qYLknKj8u3QAFEr6
QuotaGB rides the descriptor into OffboxTarget; RepoSizeBytes persisted
from restic stats. Pre-run gate: >=100% refuses NEW backups (Hungarian
notice + operator alert) but prune STILL runs (red-proofed) and restore is
never gated; >=80% warns. /backups usage bar (quota>0 only). The hub
report gains the non-secret offsite status object for the OffsiteChecker.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PSK5g6qYLknKj8u3QAFEr6
EscrowAutoConfirmer flips pending->escrowed ONLY when sha256(local repo
password) matches the ACK's restic_pw_sha256 (blob-presence alone never
confirms — red-proofed). Mismatch warns once per hash naming the ceremony;
never un-confirms; wipes the staged secret on flip. Pinned cross-repo hash
vector; manual confirm deprecated to a legacy-blob fallback.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PSK5g6qYLknKj8u3QAFEr6
Key-auth-first: a KeyAuthProber seam lets the bridge skip consume+install
when the already-installed key still authenticates (pinned to the freshly
verified host key) — descriptor changes on provisioned guests no longer
loop on consume-404. Fingerprint verify still precedes everything.
Wipe-on-escrowed: confirm-escrow now calls the agent's new
DELETE /escrow/stage-secret (v0.78.0) best-effort, closing the hygiene gap
where a ceremony-less confirm left the staged password file behind.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PSK5g6qYLknKj8u3QAFEr6
Live finding: SFTP-mode ssh-copy-id mktemp's under ~/.ssh; the container
image has no /root/.ssh, so the install died locally AFTER the one-time
password was consumed (fail-safe held; password never transmitted).
Probe confirmed the pinned known_hosts + StrictHostKeyChecking=yes chain
is sound against the real box once ~/.ssh exists.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PSK5g6qYLknKj8u3QAFEr6
The SSHCopyIDInstaller used StrictHostKeyChecking=accept-new on the ssh-copy-id
and sftp-verify connections, so even though the bridge verifies the box host-key
fingerprint against the hub descriptor BEFORE installing, the actual install
connection was not pinned to that verified key — a MITM could substitute a
different key in the gap between the scan and the install (TOCTOU).
Now the bridge threads the scanner-verified known_hosts line into KeyInstaller,
which writes it to a temp known_hosts and connects with StrictHostKeyChecking=yes
+ UserKnownHostsFile — the install/verify sessions refuse any key but the one the
bridge already matched. Empty known_hosts now refuses to install.
Test asserts the installer receives the pinned known_hosts; red-proofed by passing
an empty line (the pre-fix TOFU shape) → test fails. Addresses the security-review
"host-key TOFU after verify" finding on internal/offsiteapply/seams.go.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PSK5g6qYLknKj8u3QAFEr6
On startup reconcile the hub-served offsite: descriptor into a key-only offbox
target. internal/offsiteapply.Bridge: verify-pin box host key vs host_fingerprint
(NO blind TOFU) → consume the one-time password (single-use, never logged) →
sshpass ssh-copy-id -s -f install + verify → configure offbox → EscrowState=pending
(fork-4 via Manager.ApplyOffsiteTarget) → persist a descriptor-hash marker LAST.
Idempotent + fail-safe. Seams faked in tests; both red-proofs run+reverted.
Dockerfile + sshpass. NOT yet live-applied (supervised end-to-end next runbook).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PSK5g6qYLknKj8u3QAFEr6
Pairs with agent v0.77.0. StageEscrowSecret pushes the repo password to the
agent (POST /escrow/stage-secret) at offsite-enable → EscrowState="pending".
Atomicity gate: RunOffboxBackup (scheduler + handler) refuses until
EscrowState="escrowed" (operator POST /backup/offbox/confirm-escrow after the
escrow ceremony) — no un-recoverable offsite ciphertext can exist. DR:
POST /backup/offbox/inject-password pre-places a recovered 64-hex password 0600
(honored by WriteOffboxSecrets' IsNotExist guard; refuses clobber without
force). DR recipe gains non-secret offsite_restic coords (DRResticCoord); SFTP
key regenerated at DR, not escrowed. New settings.OffboxTarget.EscrowState.
Tests + atomicity & inject companion red-proofs green; UI gates pass. NOT yet
live-validated (supervised ceremony).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PSK5g6qYLknKj8u3QAFEr6
offbox located each toggled app's recovery unit via AppNamespaceRoot→GetAppDrivePath,
which reads the app's LIVE app.yaml HDD_PATH and silently falls back to systemDataPath
when the app isn't deployed → looked on the wrong drive, backed up nothing, reported
ok/0 (DIAG root cause). Now:
- discoverOffboxUnit/offboxCandidateNSRoots scan the durable storage registry
(schedulable non-decommissioned paths ∪ systemDataPath) for backups/primary/<app>,
independent of deploy state; newest-by-manifest-CreatedAt wins on drive churn.
- RunOffboxBackup: runOffboxInternal returns (backedUp, missing, err); 0-of-N toggled →
hard error + operator alert; partial → ok + new OffboxTarget.LastWarning (shown on
/backups, preserved across config edit).
- AppNamespaceRoot + primary WRITE paths unchanged.
- 6 non-hollow tests (A-E + edge) + both companion red-proofs run (reverted).
- NOT yet live-validated against the Storage Box (spike creds torn down).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PSK5g6qYLknKj8u3QAFEr6
loadAndParse + LoadFromBytes ran os.ExpandEnv over the whole YAML before parse;
a bcrypt hash ($2a$10$...) is full of $word sequences that get replaced with
empty env values, silently corrupting web.password_hash on load (a silent
auth-integrity bug: $2a$10$N9qo8uL... -> "a0"). Remove both ExpandEnv calls;
parse raw bytes. The typed applyEnvOverrides path (FELHOM_WEB_PASSWORD_HASH) is
the sanctioned env mechanism and is unchanged. Tests + red-proof.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PSK5g6qYLknKj8u3QAFEr6
Re-adjudicates F4: /backup/restore, /backup/tier2/restore, /backup/offbox/restore
blocked the HTTP request until completion, so through cloudflared's 100s cap a
customer got an error page while the restore succeeded (offbox worse — bounded
on r.Context(), canceling the SFTP restore mid-flight). Convert all three to the
offboxRun async shape: fast-path IsRunning refuse, background goroutine
(offbox ctx off r.Context() -> Background+30m), instant redirect. Add mutex-
guarded op-status (opstatus.go) + GET /api/backup/restore-status + a 3s-polling
backups.html banner (neutral running, red on failure). Restore single-flight
unchanged. Tests + red-proof (sync handler blocks indefinitely vs <500ms async).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PSK5g6qYLknKj8u3QAFEr6
EjectDisk/Decommission switched from c.post (drops non-2xx body) to
postWithStatus + shared refusalError, so the agent's informative 403 body
("…decommission refused (role: X)") reaches the operator instead of a bare
"HTTP 403" (campaign F2 evidence gap). Generic post + other callers untouched.
Tests T-D1/T-D2/T-D3 + ok:false case; T-D1 red-proof shows the pre-fix bare
"HTTP 403". Bundles the v0.101.0 CHANGELOG entry (this + the F3 sync deadline).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PSK5g6qYLknKj8u3QAFEr6
List-removal only deletes the registry entry — it's the undo of a
manual path add. On an enrolled drive (/mnt/felhom-drives/) it leaves a
de-registered-but-still-bound limbo nobody wants; the real lifecycle
there is Biztonságos leválasztás / Végleges leszerelés. New
StoragePathView.IsEnrolled gates the button; manual paths keep it; the
decommissioned-branch removal (final cleanup) is unchanged. Endpoint
untouched.
Two Leválasztás buttons rendered per USB drive (registry safe-disconnect
+ the agent eject the D1 merge carried over like-for-like) and four
near-synonymous labels covered very different operations. The agent
eject now hides when the card offers the registry safe-disconnect (one
detach per card; non-USB + unregistered drives keep eject). Labels are
self-documenting with explanatory title tooltips: Új telepítések
letiltása/engedélyezése, Biztonságos leválasztás, Eltávolítás a
listából, Végleges leszerelés, Formázás…. Endpoints and semantics
unchanged.
The .drive-agent-extra slot stacked the agent tag row and the agent
action row with no vertical gap (and sat flush against the registry
action row). The slot is now a flex column with a .6rem gap + top
margin, hidden when empty; enrichCard's inline margin removed.
Both wizards now source candidates from the agent's Impl-2a raw-device scan
(GET /disks/candidates, proxied) instead of the Observe-based /api/disks — so a
brand-new non-PVE-storage drive is finally discoverable + enrollable end-to-end.
agentapi.ListCandidates + a passthrough proxy (no controller-side filtering; the
agent's unclaimed filter is authoritative). storage_init renders `initialize`,
storage_attach renders `attach`; the enroll flow + Impl-1 guarded mkfs unchanged.
Tests + go build/vet/test clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
PushResponse.ConfigVersion from the report ACK; ConfigRefresher reconciles vs.
the last-applied version (settings.applied_config_version) and on a change calls
bootstrap.RefreshConfig (re-pull controller.yaml + re-merge local_api) then
GracefulSelfRestart. First-run records baseline (no restart); unchanged = no-op
(no storm); failed pull keeps config + retries. Companion to hub v0.26.0.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HxLA1mZurFq9kt8hneFeCs
Encrypted restic repo over SFTP for the app-data tier (the off-site 3-2-1 leg). A dead
NAS fails fast via -oConnectTimeout (spike Q8), never hangs the runner; secrets are 0600
files (ride DR via PBS whole-CT); init-if-absent, retention forget --prune, restore,
single-flight, per-app toggle + UI. restic re-added to the image.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HxLA1mZurFq9kt8hneFeCs
Controller-side of NAS network storage, proxying to agent A1 /netstorage/*. Distinct
'network' storage kind (no drive lifecycle); add/list/remove + per-share health UI;
unreachable NAS is a recoverable warning, never the drive missing/stop cascade; SMB
creds pass through to the agent, never persisted.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HxLA1mZurFq9kt8hneFeCs
A channel broken at startup/reseed (e.g. controller boots into pin_mismatch) was dashboard-only,
no operator email ever. New 'alerted' flag drives alerting instead of prev=='': born-down
non-transient alerts cycle 1; transient still N>=2; healthy first-obs silent; recovery re-arms.
Red-proof + companion included.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Pg8ANF97SEeKYSN5Jxw3qJ
New internal/channelhealth Checker: ~60s probe via the PRODUCTION memoized client
(Server.ProbeAgentChannel, GET /storage), classifies failures (spike Q1 map), debounces transient
reasons (N>=2; construction error latches distinctly), seeds first obs, alerts operator+dashboard on
transition. Notifier.NotifyAgentChannelDown/Recovered (English, operator-only), AlertManager dashboard
banner (Hungarian). No agent/hub change. Spike-proven.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EPZ4GJ8L5Jqf8UiPwbn1kt
Gap 1: third shim listener :2526, plaintext, does NOT advertise STARTTLS (TLSConfig
nil) — for opportunistic-STARTTLS clients with no cert-skip (cal.com, nextcloud).
Gap 2: SMTPMapping tls_mode (picks port 2525/2526/2465) + from_domain_var (split
local-part + domain for nextcloud's MAIL_FROM_ADDRESS/MAIL_DOMAIN). Default keeps
existing apps on 2525. Hub untouched.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The controller honors an operator-enforced minimum version (FLOOR) on the hub
report ACK and auto-updates to the floor when below it (managed default, no click),
reusing the Phase 1 in-guest-pull + agent-swap + rollback. Latest stays the opt-in
button; the floor is the auto-target, never latest.
- pusher.go: PushResponse += min_controller_version, latest_version (existing ACK seam)
- main.go: OnPushResponse → updater.SetFloor + MaybeAutoUpdate (rides report cycle)
- updater.go: SetFloor/GetFloor + MaybeAutoUpdate reusing performUpdate (auto-floor);
no-op at/above floor, floor>latest, dev/no-agent/backup; no flap (in-mem+persisted)
- settings UI (HU): floor display + auto restart-poll during an auto-update
- tests: below/at/floor>latest/no-flap/raised-floor; below-floor red-proof verified
- no agent change (reuses Phase 1 POST /controller/swap)
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FSZmmSFVzGwEzhYmxbkgBK
syncFileBrowserMounts no longer force-recreates FileBrowser unconditionally:
captures config.yaml+compose before writes, re-reads final content after, and
recreates only when they actually changed (new pure helper fbNeedsRecreate).
Controller restarts / no-op storage syncs now issue a plain up -d and do NOT
bounce the customer's file UI. Restore-mode DB reset still forces a recreate.
Dockerfile: removed the unused restic apt package (disk-tier restic moved to the
host agent; no controller code execs the binary). ResticSchedule/migrateResticToRsync
config+settings paths untouched (still live in the dashboard).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FpBYrZCt9sFDqLgbG5GRGD
In the intermediary-mount model an enrolled drive binds live into the running
guest (no reboot), so the "… meghajtó aktiválásra vár / Újraindítás most" banner
was an obsolete relic — also dead since v0.78 (pendingActivationDrives keyed on
the raw MountPath vs the now-stable sp.Path). Removed the banner block +
activatePendingDrives JS (settings.html), the PendingDrives feed (handlers.go),
and the dead pendingActivationDrives helper + its unused internal/system import.
Renamed handleStorageActivate -> HandleServerReboot (split out a testable
serverReboot core), removed the /api/storage/activate case, and mounted the
handler at the new non-storage route /api/server/reboot (RequireAuth+CsrfProtect).
The agent GuestReboot primitive is reused unchanged.
Added the standalone "Kiszolgáló újraindítása" settings card (sibling to the
controller-only "Vezérlő újraindítása"), reusing the pollRestart() loop.
Test: TestHandleServerReboot_CallsGuestReboot (fake diskAgent asserts GuestReboot
invoked once + 202). diskAgent/mockAgent gained GuestReboot.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017PsnU2ASocYrvzqE82YDYW
The storage card displayed each external drive's raw host PVE mount (/mnt/<name>,
which doesn't exist in the guest) instead of the stable in-guest path
(/mnt/felhom-drives/<name> = guest_path) the registry/HDD_PATH/FileBrowser use.
The eject/wipe buttons also posted the raw path, so they would unmount the drive
but leave the stable registry entry orphaned, and the impact warning found no apps.
Fix: card sub-line + eject/wipe buttons use the stable path (regKey); type-to-confirm
name uses the basename; register keeps the raw path. handleStorageWipe maps to raw
via agentWhere() for the agent eject (matching handleStorageEject). Agent ops
unchanged (same raw paths); display + registry bookkeeping corrected.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017PsnU2ASocYrvzqE82YDYW
Follow-up to v0.78.0. The disk-view JS (settings.html regBadge/actions) checked
registration by the raw mount_path, but the registry stores the stable path since
v0.78.0 — so enrolled drives showed a spurious "Nem regisztrált" badge + Register
button. Fix: regKey(d) = d.guest_path || d.mount_path (agent reports the stable
guest_path per disk); registerDrive still posts the raw path. Display-only.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017PsnU2ASocYrvzqE82YDYW
handleStorageRegister (the "Regisztrálás" action for an already-mounted,
unregistered drive) registered the raw /mnt/<name> path verbatim, unlike
runStorageInit/runStorageAttach which map to the stable /mnt/felhom-drives/<name>
path the agent actually binds the drive at. The controller then watched an empty
placeholder dir on the guest rootfs → "Rendszermeghajtón" + stuck "activation
pending" banner after a re-provision.
Fix: register stablePathForName(path.Base(req.Where)); attachIntoGuest still uses
the raw path (the agent operates on raw). Test + red-proof added.
Diagnosis: felhom.eu/documentation/audits/DIAGNOSE-drive-bind-after-reprovision-2026-06-23.md
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017PsnU2ASocYrvzqE82YDYW
Belt (ensureUserdataMounts) + FileBrowser sync skip ensure/mount when an external
drive root is not a live mountpoint -> no 'mkdir userdata: permission denied' + no
rootfs-shadow during a drive-absent window. System/local path never gated. Reuses
system.IsMountPoint; matches planDriveGates external-only rule. T1-T4 + red-proofs.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
agentClient() built a new agentapi.Client (new bare http.Transport, IdleConnTimeout:0)
per call and discarded it without closing idle conns -> one leaked idle ESTABLISHED
socket per call to the agent :8443, exhausting the ephemeral port range after ~5 days
(EADDRNOTAVAIL). Memoize one shared client via sync.Once; harden Transport
(MaxIdleConns/PerHost + IdleConnTimeout 90s). Agent/firewall untouched.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
DR recipe slice (controller half), grounded in SPIKE-dr-recipe-2026-06-16. The
controller emitter is the BOUNDARY enforcement point: v1 ships an explicit
allowlist {catalog_ref, enabled, storage_bindings} and reads NOTHING from
AppConfig.Env, so no ENC:/token/password can leak. storage_bindings parsed from
the compose (${HDD_PATH}/${USERDATA_PATH} volume binds -> {container_path,
drive, subpath}).
Load-bearing tests: TestBuildAppRecipe_NoSecrets (synthetic-secret app -> none
leak) + TestBuildAppRecipe_AllowlistIsLoadBearing (red-proof companion) +
TestAppStorageBindings + TestBuildDRRecipeAppHalf. Red-proofed live: forcing the
emitter to dump Env makes the boundary test fail. recipe_version=1.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
admin