MEASURED, not supposed. On 2026-08-03 nine per-app recovery_unit_capture_failed
events reached the hub and TWO operator emails went out. The hub's operator
cooldown key is customerID:eventType(+tier) and that event carries `app` but no
`tier`, so the key held no app identifier: the first refused app took the hour's
slot and every other app's failure was discarded BEFORE anything was written
down, leaving no row on any channel.
The obvious fix — put `app` in the key — was ruled against: on a full disk it
produces one email per app, the volume problem wearing the correctness problem's
clothes.
internal/backup/runsummary.go: a per-run collector with exactly admissionSet's
lifetime, fed by all three write legs, emitting backup_run_failures ONCE at the
end and only when something failed. A clean run emits nothing.
The per-app event stays and becomes the RECORD — the hub routes it record-only,
stored and logged every time, never competing for an email slot. The record and
the notification are now different things.
Deliberate skips (disconnected, decommissioned) are excluded: they have their
own alert, and a nightly email about an unplugged drive is one the operator
learns to ignore.
A manual run always reports: the digest carries a unique run_id the cooldown
cannot collapse. Someone pressing the button is actively trying to get a backup.
THE PERIODIC SWEEP GETS A DIGEST TOO. With the per-app event now record-only, a
capture failure found between runs would be recorded and never notified — a new
silence introduced while closing one. That path emits a digest with NO run_id,
so the ordinary 1-hour cooldown caps it exactly as before while the mail now
lists every failing app instead of whichever was first.
A refusal is recorded ONCE, where the verdict is taken, not at the three legs
that consult it — R-181's contract is one verdict per app per run. Noting it per
leg listed one refused app three times and produced "2 of 1 apps failed". Found
by the digest's own test, not in review.
Silence is safe because the hub's deadline check raises expected_backup_missed
from report freshness, independently of any mail this box sends
(monitor/deadline.go:396,417). Confirmed, not assumed.
7 new tests, 4 red-proofs. The main.go seam walk did NOT fail on its first
attempt — the AST test walked the backup package and not main.go; the test was
fixed and the mutation re-run rather than the pass recorded.
Found by v0.193.0's own live proof run. The estimate was printed fixed to two
decimal GiB, so every app under ~10 MB rendered as "estimated 0.00 GiB write" —
which reads as "no estimate was available" and is the opposite of what happened.
Observed live on demo-hp 08:59:46: opengist's real 178 KB estimate printed as
0.00 GiB.
Shipped in the same session because it is the same defect class R-181 is about:
a message an operator cannot rely on is worse than no message.
The arithmetic is unchanged and still in GiB — the reserve's own unit, so the
comparison against FloorFreeGiB reads directly. Only the rendering moved to
humanizeBytes. estimatedWriteGiB -> estimatedWriteBytes, with the GiB conversion
done once at the point of comparison.
B2's capture floor (v0.192.0) was consulted in exactly ONE place —
captureAllRecoveryUnits, which writes a few KB. The two legs that write the BULK
into the same backups/primary/<app> tree, the DB dump and the volume dump, ran
FIRST and unguarded. Measured live on demo-hp 2026-08-03 06:40:03: opengist's
volume dump wrote 2.0 GB with no check, free fell to 1.0 GB, and the floor then
refused the cheap write it had already lost the argument to. Its refusal message
claimed "the previous unit is untouched" — measured false: that app's tar had
gone 182,272 B -> 2,147,666,432 B under a stale manifest. Sixth entry in
CLAUDE.md's table of shipped guarantees the code did not provide.
Fix: ONE admission verdict per app per run (internal/backup/admission.go), taken
before that app's FIRST write and covering all three legs — they write under one
per-app root, which is why one verdict can honestly cover them.
- Lazy, at the app's first write, NOT once at run start: app A's dump can put app
B under the reserve, so a run-start verdict reads a disk that no longer exists.
- Remembered for the run, never re-decided between an app's own legs — that is
the split this closes. Reset per run.
- Placed ahead of DumpAppVolumesSafe, which stops the stack as its first act, so
a refused app is never bounced. After the volume-less check, which has no write.
- Exactly one operator alert per refused app per run.
- Leg order unchanged: volume dumps still precede the capture.
The floor is now SIZE-AWARE: it asks whether THIS app's write would cross the
reserve, not only whether the filesystem is already below it — which is how an
app was admitted at 96% and then allowed to write 2 GB. Estimate = the app's
previous .sql + .tar on disk. No history -> headroom-only, deliberately, and the
alert says so.
A container-based du per volume was MEASURED and rejected: 66 timed runs on
demo-hp guest 9201, median ~355 ms/volume (341-404) on volumes holding tens of
KB — container start-up, not the walk. Decisive on top: docker run needs the
writable layer, so it can fail under exactly the pressure the reserve handles.
The message was NOT weakened; the behaviour was moved so the wording became true.
It now also names which term bound. Every claim is checked against a sha256
fingerprint of the tree it describes, never against the log line.
Still refuses and never deletes: nothing here is generational.
11 new tests through the production functions. The DB leg cannot run without
Docker, so its gate is pinned by an AST walk of backup.go asserting admitApp
precedes DumpOne (strings.Contains is insufficient — a commented-out call still
contains the string). 4 red-proofs demonstrated failing then restored.
Ships BEFORE the disk-layout merge it exists for, and is harmless on a box
that never gets it. The mp1 partition was a BULKHEAD as well as a ceiling:
it kept a runaway capture from filling the space the container runtime
needs, because /var/lib/docker was a different filesystem. After the merge
it is the same one, and a full Docker data-root is a stopped box.
The floor sits in captureAllRecoveryUnits, checked BEFORE anything is
written: below the reserve, that ONE app's capture is refused, its previous
unit is left byte-identical, the R-158 alert fires with the space figures,
and the loop continues.
Two terms whichever binds first (97% used / 1 GiB free) in fillwatch's
shape, deliberately BEYOND its critical band (95% / 2 GiB) so the customer
is always warned before a refusal can happen — a floor that fires before
its own warning is a silent failure wearing a threshold.
Headroom, never unit size: a per-unit cap would be R-163 rebuilt inside one
volume. Refuses, never deletes: nothing here is generational, so pruning
could only destroy a different app's only local copy; pruneStalePrimaryDirs
is an orphan sweep, not retention, and must not be repurposed.
Tests 1184 -> 1191. One fixture strengthened mid-red-proof: the "old 20 G
ceiling is gone" test sat at exactly 20 GB and survived a literal
UsedGB > 20 cap — hollow. Now 120 GB, and the mutation fails it.
R-167: new internal/fillwatch warns the CUSTOMER before a filesystem fills.
It emits the PRE-EXISTING disk_warning/disk_critical pair, which was
allowlisted, copy'd, default-enabled and checkbox'd with no producer in any
repo — the sixth "built but never wired" instance here. Two threshold terms
(85% or 5 GiB free; critical 95%/2 GiB) because a percentage alone lies at
both ends of this fleet's size range. Edge-triggered on escalation only,
state persisted, hysteresis dead zone at 75%/7 GiB pinned by a test. A nil
usage read is never a warning and never clears one. Per filesystem, never
per app. Daily 03:30, before the nightly app-data legs.
R-158: new unitNotify seam fires per app when a Tier-1 recovery-unit capture
fails, loop continuing, carrying the target filesystem's used/free bytes.
Operator-tier (recovery_unit_capture_failed) — deliberately NOT backup_failed,
which is customer-enabled and would email the customer about a failure they
cannot act on. D-c overrides R-158's own proposal here.
R-174: the app-stop guard no longer starts apps onto MISSING drives — a
regression in v0.189.0 code, found by review and closed the same session.
SetStarter got the raw stack manager, whose StartStack has no drive gate,
and Recover runs at startup. R-171 one path over. bootDriveGate could not be
reused whole (its holder #2 is the guard's own marker, and holders #1/#2 read
vars assigned after Recover runs), so holder #3 is extracted into a shared
driveStartGate with a test pinning the delegation. ErrStartRefused splits a
refusal from a failure: both keep the marker, only Failed alarms, because
routing a deliberate hold into NotifyBackupFailed is the same false alarm.
Tests 1157 -> 1184. All red-proofs demonstrated failing and restored.
R-171 (a regression v0.189.0 introduced, CONFIRMED on hardware before any fix
was written). Replacing isBootOrphan's container-count term with recorded intent
made a drive-gate-stopped app read as a boot orphan: the gate stops apps with
`compose down` (zero containers) and never touches desired_state, because it is
not the customer. Observed on 9201 with the drive held unmounted — the sweep
found and started it, burned both attempts, and handed it to the dead-app alarm.
The write hazard did not materialise (the unbound mountpoint is host-root-owned
and the guest is unprivileged) but that protection is accidental and untested.
New consumer-side seam bootrecon.StartGate, fail-safe (cannot determine ⇒ do not
start), wired in main.go. The rule is not new: the API's startGatedByMissingDrive
already refuses this; the sweep bypassed it.
R-157 mechanism A. The sweep looked once at T+5s, deriving candidates from a
fleet docker was still restoring — three of six hard resets. Now a settle-then-
sweep window: sample every 5s, settled after 3 identical samples, sweep ONCE at
the end; ends on settled or a 50s budget, and the log says which. The budget is
50s because settle+budget+one retry must stay under the 90s dead-app grace — a
test rejected 60s at 95s. A window that overruns emits a LATE RECOVERY warn
rather than the grace being widened to hide it.
Widening the window made two more holders reachable, so the one gate covers all
three: an absent drive, a quiesce, and an in-flight app-data operation — reusing
quiesce.SuppressedStacks() and a new read-only AppStopGuard.HeldStacks().
R-170. shouldRecreateOnBoot now reads desired_state with the identical three-way
table; absent keeps the old hasContainers behaviour exactly. Its comment argued
for the container count and was rewritten. presentStable is untouched. The two
gates' agreement is pinned from both sides against one fixture table.
27/27 packages green; 6 red-proofs observed FAIL then restored.
The box stops inferring the customer's intent from a container count and reads
what they actually asked for.
Part 1 — desired state. AppConfig gains a tri-state `desired_state`
(""/running/stopped), written ONLY by the customer's own action: the API action
switch, DeployStack, UpdateOptionalConfig's redeploy branch, and the .fab
import. Intent is written BEFORE the act and a failed write REFUSES the act.
StartStack/StopStack are deliberately not writers — 14 callers, only 2 are the
customer. bootrecon.isBootOrphan now reads intent instead of len(Containers)>0,
which closes R-157 mechanism B (a power cut or interrupted deploy left an app
with zero containers, read as a deliberate stop, and stranded silently).
ABSENT MEANS UNKNOWN, NEVER "running": every pre-v0.189.0 app.yaml reads absent,
so the legacy fallback is byte-identical to the old rule. A running-only startup
backfill converges the unambiguous cases; `stopped` is never inferred.
Part 2 — backup.AppStopGuard, a persisted marker over every stop→work→start
window (volume dump, offbox reconstitute, .fab export). Its own file, never
quiesce's. Written before the stop, cleared only after a restart that succeeded,
kept when one fails. Recover() completes before the boot reconciler is launched
and returns its outcome, which main.go reports on the existing backup_failed
event once the notifier exists. A defer is not the mechanism — a SIGKILL runs
none (Campaign 8 fault 10).
Also: SaveAppConfig rebuilt AppConfig field-by-field (the R-100 shape) and would
have dropped desired_state on every save across nine call sites. Replaced with
copy-and-overlay. Measured: app.yaml does not round-trip unknown YAML keys.
No hub change, no agent coupling, no user-visible string. 27/27 packages green;
7 red-proofs observed FAIL then restored.
The recovery unit on the customer's drive now carries the PORTABLE secret
class, so Tier-1/Tier-2 restore no longer depends on the whole-guest tier.
A customer needs the drive and nothing else.
Part 0's rulings overturned the brief's recommendation, on evidence:
- the data_key flag is untrustworthy (4+ encryption keys the catalog itself
labels as such are unflagged) -> R-127
- a DB password is not resettable in practice: POSTGRES_PASSWORD is ignored
once PGDATA is non-empty, so a regenerated value leaves the app unable to
authenticate against its own restored rows while the dump replay still
reports success (proven on a throwaway postgres:16-alpine)
Ruling (operator): type:secret travels, type:password never does, minus the
nonPortableSecrets code register. Plaintext -- withholding the internet-
reachable class is what licenses that, and the two are coupled.
Precedence: the UNIT WINS over the guest -- the unit's secrets were captured
in the same run as the dumps beside them, so they match the data being
restored. The fail-closed data-key gate is unchanged.
Secret values are never logged; the manifest records NAMES only.
Both are the system reporting healthy while the customer is not, and both live in the same
status-derivation code. Neither is fixed by making the system quieter.
C9-F1 (HIGH) — Tier-2 writes recovery-unit/ on EVERY run and RestoreTier2Files has never read
it (tier2_restore.go:101-104 reads hdd/ + userdata/ only). Phase 0 enumerated all 53 catalog
templates against both demo boxes: 43 apps have NO readable subtree, so the button stopped the
app, restored 0 files, restarted it and said "Nincs hiányzó fájl — minden fájl megvan a helyén."
— at the moment the customer pressed it because files were missing, with 156 MB of BookStack's
data unread in the same copy. 9 apps have file legs but never their DB or volumes, so the same
sentence was also a clean bill of health over data never opened (immich: 1.3 GB Postgres unit).
Honesty half shipped: a pre-flight coverage check refuses UP FRONT without stopping the app and
NAMES the action that works; a run that proceeds claims only what it EXAMINED and discloses that
the database and volumes are not covered. Completeness is filed as C9-F1b — routing to the
Tier-1 unit restore puts a destructive operation behind a non-destructive button, so its confirm
copy has to carry that difference. C9-F4 filed: nothing reads the Tier-2 recovery-unit/ mirror,
so the second local copy that exists for drive loss is unreachable by any customer action.
C9-F2 (HIGH) — a crash loop was counted as working. StateRestarting is deliberately NOT added to
IsDownState (that alarms on every deploy fleet-wide, the over-correction F-A1 nearly cost us); a
sustained run becomes down after crashLoopAfter = 5m, set above the 120s deploy timeout, Mealie's
60s start_period and R-97b's 180s grace. The dashboard counter uses the same predicate, so it no
longer contradicts the alarm on the same screen. README's claim that faults "still surface as
restarting" was a wish with no test — corrected in place; it is the seventh such instance.
Six red-proofs observed, including the one that matters most: adding StateRestarting to
IsDownState fails the brief-restart test with "every deploy and update would page the operator".
go test ./... rc=0, 27 packages, run and read separately from this commit.
Tier2LastRun is the attempt clock and was rendered as 'Legutóbbi másolat' in the restore
confirm dialog. New LastSuccess + SuccessTracked anchor; tier2Update makes the three
rebuild sites safe by construction. F-DIAG: six distinct causes, target-aware redaction.
LastRun records an attempt, not a result. New OffboxTarget.LastSuccess, set only on the
success branch via the pure offboxAnchorAfterRun rule, carried to the hub as last_success.
Closes two silent-wipe sites (settings save, hub re-apply).
${IMPORT_PATH} = <system namespace root>/userdata/import — ONE drop-zone per box,
on the system drive, injected at BOTH compose-env builders with NO per-drive
fallback (unresolvable leaves it unset so compose fails loudly rather than
quietly building a second, dead drop-zone).
Third BindRoot (RootImport) + Import list in BackupSpec, extended through
ValidateBackupSpec/ClassifyBinds. Load-bearing: a stale `userdata: import/<app>`
entry against the moved bind would be a WHOLE-BLOCK reject, taking the app's
mandatory hdd classification with it.
Exhaustive-root audit: resolveAbs/structuralGuard/ComputeCaptureSet/
ComputeFabBuckets now take importRoot explicitly (an import bind resolved
against hddPath would name a directory on the wrong drive); unresolvable is
refused loudly into Skipped. GetImportRoot added to both provider interfaces.
Catalog-derived skeleton: UserdataSkeleton() -> UserdataSkeletonCarry() +
BuildUserdataSkeleton(), SORTED. The carry-list makes zero-removals true by
construction (`documents` is in no catalog app but on both boxes) and is the
fresh-box floor. The sort is not tidiness: the naive map-order derivation
measured 20 distinct outputs from 20 identical runs, which with fbNeedsRecreate
is a fleet-wide FileBrowser restart loop.
One authoritative compose parser: ParseComposeUserdataMounts now delegates to
ParseComposeClassifiableBinds. Import root excluded from per-app migration.
Surfaces: FileBrowser /srv/beolvasas source; app-page "Hova tegyem a fajlokat?"
with PathEscape deep links (never QueryEscape) and class-driven copy;
data_paths: annotation with the Fork-3 asymmetry; system-owned beolvasas SMB
share refused server-side at handler AND store, button omitted in template.
Caught on the way: the sharing template's row struct was function-local, so
adding {{if .System}} would have 500'd every share row. ShareRow is now
package-level and the render test uses the handler's own type.
Tests 915 -> 949, all green. MinAgent unchanged.
gofmt -w across the controller tree (46 files) so gofmt -l is empty — disarms the
formatting landmine where a targeted edit + accidental gofmt -w swept ~46 unrelated
files. Pure formatting: whitespace + gofmt's optional-semicolon removal in reflowed
inline closures. One doc comment reworded ('' -> 'the empty string') to avoid gofmt's
Go-1.19 doc-comment typographic substitition ('' -> curly quote) muddying its meaning.
No build/vet/test behavior change.
Closes R-47. No new agent coupling — MinAgent stays 0.90.0.
The replay needs a running DB container, so both restore paths started the
WHOLE stack first, giving the application a window to rebuild the very schema
objects the dump was about to create. Measured live on 2026-07-19 (H4,
DIAG-immich-restore-round2): immich-server rebuilt clip_index two seconds
before the dump's CREATE INDEX, the replay aborted "already exists" under
ON_ERROR_STOP=1, and immich reported schema drift. The data survived only
because pg_dump emits COPY before CREATE INDEX.
Both paths now open a DB-ONLY window: only the stack's database service(s)
come up, the dump is replayed with the app still down, and the full start
runs only after the replay exits 0. Fail-closed: a dump with no identifiable
DB service refuses BEFORE the first mutation. Every exit from the window
still does a best-effort full start, so a failed restore never leaves a box
with a database and no application.
New: appbackup.DBServiceNames (yaml.v3 services-map parse — never a line
scan; immich's top-level volume keys are the decoy) sharing dbTypeForImage
with DiscoverDatabases; stacks.Manager.StartStackServices (refuses an empty
list — argument-less `up -d` is a full start); RedeployFromEnv split into
PersistUnitRedeployConfig + its unchanged tail. StackDataProvider's
RecreateStackFromUnit becomes RecreateStackDefinitionFromUnit — the hidden
`up -d` inside the old name is what carried the defect on the local path.
19 new tests (ordering plus state-at-replay-time, zero-mutation fail-closed
effects, replay-failure bring-up, parser decoys, empty-list refusal); three
companion red-proofs run and reverted. 23/23 packages green.
Not yet live-validated: STOP-1 supervised reconstitute, golden 0.153.0.
F7/R-53: app_export.html built the app's public URL as '<sub>.{{$.CSRFToken}}',
so the "Megnyitás" link was wrong for every app with a subdomain and a session
CSRF token was written into a URL. Template now uses {{$.Domain}}, and
exportPageHandler supplies the key — it builds its own data map instead of
going through baseData, which is where every other page gets it. The page's
real CSRF path (csrfH() reading the meta tag) is correct and untouched.
The 7 red internal/backup tests are green again, with no behaviour change.
TestTier2V2_* / TestSharesTier2* all failed for one environmental reason:
Tier-2's off-drive guard asks system.SamePhysicalDevice (st_dev equality)
whether a target is really a second disk, and every t.TempDir() here shares one
filesystem — so the guard correctly refused the fixture's "two drives" and the
tests never reached their subject ("nincs másik fizikai meghajtó").
Seam in the package's existing style: a nil-defaulted Manager.samePhysicalDevice
field + sameDevice wrapper, seven call sites routed through it. Nil resolves to
system.SamePhysicalDevice, so production is byte-for-byte unchanged; only the two
fixtures inject a fake modelling one drive per directory subtree. No assertion
weakened, nothing skipped/renamed/deleted; all 7 mutation-proved.
Also: the ssh->pct-exec ASCII-grep and heredoc-credential traps are now in
CLAUDE.md's live-validation section.
Closes the two findings from DIAG-immich-restore-2026-07-19. Viktor deleted 11
immich photos to test offsite restore; both runs flashed success and the photos
stayed gone. Two independent defects.
R-43 — no offsite path could restore a database. All three buttons were
file-only: the two "visszaállítás" actions staged to a scratch folder and never
touched postgres, and place-to-live merged only MISSING files. For a DB-indexed
app the bytes returned and the app still could not see them. The dump was
carried INTO every snapshot and could never be replayed OUT of one.
New ReconstituteFromOffsite (/backup/offbox/reconstitute): safety dump → stop →
files overwritten to the snapshot version → start → the snapshot's own dump
replayed → health wait. Two invariants:
- nothing is ever deleted (-a, no --ignore-existing, no --delete): a file
created after the snapshot survives as an extra;
- the undo exists before the act — the pre-restore- dump is verified ON DISK
before anything is stopped, overwritten or replayed; if it cannot be taken
the operation refuses with zero changes.
The replay reads the SCRATCH unit: the live unit is never overwritten, so
replaying from it would replay the current DB over itself and restore nothing.
R-44 — a manual push shipped an unrefreshed dump (up to ~24h old). That day's
predated the customer's account by four hours and probed to asset:0/user:0/
album:0 inside 52MB whose bulk was immich's shipped geodata. Every run, manual
AND nightly, now refreshes dumps + units BEFORE capturing. Order is the
mechanism: the gap can only ADD files the DB does not reference yet, never
remove one it does. Manifests carry offsite_run_id + dumps_at, so coherence is
verifiable at restore time rather than assumed; the periodic refresh carries a
prior stamp forward and never invents one.
Honesty surfaces, all warn-level and none a gate: unstamped (pre-v0.148) pairs
report their skew, ValidateDump gained an EXACT-match accounts-table sniff for
customer-empty dumps, the completion flash states an outcome instead of a
mechanism, and the missing-only button now says what it does NOT do.
11 tests; 5 red-proofs run and reverted. Two of those found real test weaknesses
rather than confirming strength — the first undo mutation was caught by a second
guard, and the first table-matching test did not discriminate between the two
matchers at all. Both tests were rewritten to the cases that separate them.
NOT in scope: R-41's catalog invariant check, nightly cadence, retention, quota
math, tier-2, and v0.147.x progress semantics beyond one added phase line.
Live acceptance (§9) has NOT run: no capability-map flip, customer-restore row
stays MISSING, R-3 stays DRAFT.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01P9Nn14TWGzKoqAJAiVwC2s
Third real run, third thing only a live run could show. The per-app legs
finished in ~15 seconds; the remaining 40 of the 57-second run was the shares leg
and forget --prune, during which the card sat frozen on "calibre-web — 8 / 8
fájl". The same frozen-looking silence 4c exists to remove, relocated to the end.
Progress now carries a phase. The post-app stages announce themselves and the
app-scoped counters are cleared when a phase starts, so the last app's finished
numbers are never shown against work that is no longer about that app. Starting
the next app clears the phase again. Pinned by a test.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Nn3VgQk9iwEGgyx6QJ2NvE
The v0.147.1 file-count fallback fixed the incremental case but not the one the
demo box actually hits. Watching a second real run: bookstack reported clean byte
progress (100%, 154.0 MB, 7/7 files — the byte path works), while immich sat at
files_done 1 of 46, bytes_done 0, for 42 seconds.
restic 0.14 only counts a file into bytes_done/files_done when it COMPLETES, so
an app dominated by a single large archive (immich's ~430MB volume tar) freezes
both counters. No percentage can move in that window, so stop trying to fake one.
restic keeps reporting current_files and seconds_elapsed throughout. The card now
names the file being processed and the elapsed time: "1 / 46 fájl (430.2 MB) ·
feldolgozás alatt: immich_upload.tar · 42 mp". "Working on this file for 42
seconds" is a completely different message from "0%", and it is the honest one.
The last known current_files value persists across ticks that omit it (restic
does not send it every tick, and blanking the label every other second is its own
flicker); switching app clears it so one app's file is never shown against
another. Both pinned by tests, with the real 42-second status line shape.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Nn3VgQk9iwEGgyx6QJ2NvE
Found by watching the v0.147.0 card during a real manual run on the demo box,
which is the only way this was going to surface: a 430MB immich push reported 0%
for 40+ seconds and then completed.
The parser was not broken. restic was genuinely reporting no transferred bytes —
on an incremental run where nothing changed, bytes_done is omitempty on restic's
side so it is not even in the JSON, and percent_done stays 0 for the whole run.
Confirmed against the real schema by capturing backup --dry-run --json from
restic 0.14.0 in the controller image rather than guessing; those captured lines
are now quoted verbatim in the type's doc comment.
Why it mattered: a byte-only bar is indistinguishable from a hang in the COMMON
case, which is precisely the silence 4c set out to remove. Shipping it would have
traded "no feedback" for "feedback that says 0% and looks stuck".
files_done/total_files are now parsed and published alongside the bytes; the card
prefers bytes when bytes move, otherwise drives the bar from files and says
"N / M fájl ellenőrizve". parseResticStatus returns a struct instead of four
positional values, and a new test pins the real incremental line shape (bytes
absent, files climbing) so a refactor cannot quietly restore the stuck bar.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Nn3VgQk9iwEGgyx6QJ2NvE
The systemic complaint, twice in one evening: you press a button and nothing
happens. No progress, no ETA, no named result. Three worst offenders, fixed on
the two patterns already here (deploy 3-step panel, storage-init status poll).
No new framework — that is a ROADMAP item; three targeted cards ship tonight.
4a — a verification restore names its result. The flash said the app had been
restored "to a verification folder on the drive"; which folder, on which drive,
was invisible, so the customer could not go and look at what they had just asked
for. Full path now. The restore page gained a listing of existing verification
copies (app, size, date, path) — nothing anywhere showed these, so they piled up
and the only way to find them was SSH — each with a double-confirmed delete.
That delete is the only one this release adds, so it names a STACK, never a
path: the Manager resolves the name inside a backups/offsite-restore root it
computed itself and refuses anything landing outside. Red-proofed — neutralise
the name guard and stack:"" resolves to the offsite-restore ROOT and takes every
copy with it. Refusals are asserted as non-effects.
4b — Megosztás enable shows what it is waiting for. Enabling ran ReconcileSamba
synchronously inside the POST handler; on a golden without felhom-samba baked
that is compose pulling ~100MB, i.e. minutes of an apparently-hung form post
followed by "Beállítás mentve." whether or not anything came up. Detached +
polled now, distinguishing "képfájl letöltése" from "indítás" — decided BEFORE
the work starts, since afterwards the image is always present. Success is
probed, not inferred (compose up -d exits 0 on a crash-loop). The password form
starts the same job: with UserSet false reconcile deploys nothing, so on a fresh
box that is where the pull actually happens.
4c — "Távoli mentés most" streams real progress. restic was already reporting
bytes and percent; the runner seam used CombinedOutput() and discarded them. The
manual run now passes --json and scans stdout line-by-line: total bytes, percent,
current app. Manual only — the nightly stays silent, pinned by a test that fails
if it ever passes --json. The poll now arms unconditionally, closing a race the
manual trigger always ran: the redirect rendered before the goroutine wrote
LastStatus=running, so the poll never armed and the page sat static during the
very run just started. Red-proofed twice.
Also closes the golden/controller infra-image drift at the source: infra.Images()
derives from the existing pins and --print-infra-images exposes it, so the golden
bake can stop carrying its own copy. That copy had already drifted — felhom-samba
was never added, so the golden baked 3 of 4, which is why enabling Megosztás
pulled at runtime in the first place.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Nn3VgQk9iwEGgyx6QJ2NvE
Found by LIVE VALIDATION, not by a unit test: the first demo tier-2 run pushed
'Masodlagos mentes elkeszult: _shares' — the reserved key reached Hungarian
customer/operator copy.
Mapped at the SOURCE of the notification (RunSharesTier2's tier2Notify calls) so no
future notifier wiring can reintroduce it, plus DisplayStackName at the main.go
wiring as idempotent defense in depth. Regression test added with a red-proof.
Part 4 — restore: RestoreSharesScratch + PlaceSharesRestore as SIBLINGS of the
per-app scratch/place pair. Files merged missing-only (never overwriting), each
destination PREFIX-ASSERTED against registered LIVE storage roots; definitions
merged with existing-wins; ReconcileSamba via a seam (backup must not import
stacks); credential restored best-effort into the samba named volume.
New routes POST /backup/shares/{restore,place} + a restore-page entry that renders
'Megosztasok', never the raw reserved key.
Also adds scratchJoin: reconstructing an absolute captured path under a scratch
must strip the volume name rather than rely on filepath.Join.
Part 5 — liveness: EffectiveProtected gains a settings-backed dynamic extra so the
samba CONTAINER (not the stack name — they differ) is watched exactly while sharing
is on. FINDING: the issue -> health 'fail' -> existing health_critical event ->
alert -> Hungarian degradation e-mail path needs NO further change, and introduces
no new event type, so the allowlist gotcha does not apply.
Part 6 — UI: per-tier backup status lines on the Megosztas page (amber only on
deviation). Verified the two warning-prose sites (offbox_capture/tier2_capture)
only ever receive per-app stack names, so no mapping is needed there.
RED-PROOFS RUN AND REVERTED (both fired):
4. prefix-assert removed -> place-guard traversal test FAILS
5. dynamic samba extra removed -> Scenario E enabled-case FAILS
ONE additional restic call tagged [felhom-offbox, _shares] carrying the payload
staging dir + every mandatory share folder. Hooked into runOffboxInternal AFTER the
per-app loop and BEFORE retention, so forget --group-by host,tags covers the _shares
group with no flag change. Reuses resticStep, the caller's repo-ensure and
single-flight, and the SAME enlargement-gate arithmetic.
- quota gate degrades the push to MANIFEST-ONLY, never to nothing
- EnlargedBlocked keeps the RAW _shares key (templates index by it); the display
mapping applies only at the notification + Hungarian-prose boundaries
- OffboxTarget gains SharesLastRun/Status/Count for per-tier page truth
- zero-toggle notice suppressed when the shares leg provided coverage
- reserved-name defense: an app keyed _shares is excluded from the run loudly
RED-PROOFS RUN AND REVERTED (all fired):
1. shares leg appends into the app's argv -> isolation test FAILS
2. mandatory->offsite mapping inverted -> Scenario A + B FAIL
3. manifest-only degradation dropped -> Scenario C FAILS
Sibling shares source for the local cross-drive tier. Reuses the tier2Mirror seam,
selectTier2TargetFrom (narrow source-drive seam extracted from selectTier2Target),
tier2ReconcileRoots (pure extraction), tier2SafeRemove, the marker-LAST discipline
and the recordTier2* helpers. Per-app paths are untouched.
- shares_payload.go: deterministic _shares-manifest.json + best-effort passdb capture
- tier2_shares.go: per-source-drive legs -> cross-drive target, payload, marker LAST
- infra.SambaContainerName/SambaPassdbVolume/Mount: single source of truth for the
container identity (renderer, stacks execs, backup execs, monitor all read it)
- RESERVED-NAME finding: ValidateSMBShareName did NOT exclude a leading underscore,
so "_shares" was an accepted share name. Now refused; RunAllTier2 additionally
skips a "_shares" stack loudly as defense in depth.
- fix: shareSourceDrive returned a slash-normalised path, which made the target
selector's source-drive equality check miss (a group could target its own drive)
tier2_capture.go: classified apps get TierSecondary per-bind legs (paperless copy shrinks — export
drops); legacy apps keep the byte-identical resolver set. v2 relpath-mirroring layout
(backups/secondary/<stack>/{marker LAST, recovery-unit/, hdd/<rel>/, userdata/<rel>/}); N>1 native
(errTier2MultiDir/tier2AppDataName deleted). Migration=delete-and-rebuild + reconcile; all RemoveAll
via tier2SafeRemove (refuses outside backups/secondary/). SSD=state-only tier. selectTier2Target
never picks network storage (pinned+auto, F-6C-1). Restore reads v2 behind a marker gate.
Part 0: offbox_enlarge_blocked is a persisted one-time Load seed (opt-out sticks), not a getter
append. Part 0.5: offsite restore scratch prefers a local (non-network) path.
Full v2 test suite + all 10 §10 red-proofs verified. Destructive writes bounded to backups/secondary/.
PlaceOffsiteRestore: live target via raw GetStackHDDPath not AppNamespaceRoot (F-3a-1a: no SSD
merge; undeployed refused), placement headroom gate (F-3a-1b), stat pre-pass over all placements
before any copy (F-3a-4: no partial writes), scratch removed on success/kept on failure (F-3a-2).
mapOffsiteRestorePaths refuses the namespace root itself (F-3a-3).
Delivery chain: DefaultEnabledEvents + GetNotificationPrefs append-if-absent migration + settings
checkbox + handler slice; paired with hub v0.55.0 allowlist (no customerMessages entry — raw
dynamic message survives). +8 tests; all 6 controller §10 red-proofs verified.
Task 2 of the backup-classification-redesign arc. Ships the referential-
coupling classification as DATA + PARSER + PURE CLASSIFIER, deliberately
inert — no backup tier changes behavior. Task 3 (tier policy engine) and
Task 4 (manual .fab UI) consume it.
- appbackup/classify.go: BackupSpec/BindSpec/ComposeBind/ClassifiedBind;
ClassifyBinds (SQ5 two-level default — explicit beats :ro; unlisted
writable→mandatory, unlisted :ro→excluded; nil spec→legacy/false);
ValidateBackupSpec (whole-block-reject on any defect, first defect named).
- stacks/classify_binds.go: ParseComposeClassifiableBinds — ${VAR}-relative
binds + :ro flag (NOT ParseComposeHDDMounts/ExportDataMounts, the traps).
- Metadata.Backup + LoadMetadata as the single validation choke point (bad
catalog block → nil + one ERROR → legacy, within one sync cycle).
- Manager.ClassifiedBinds + StackDataProvider.GetStackClassifiedBinds seam
(delegated by stackAdapter, nil-stubbed in every fake) — wired + tested
now so Task 3 consumes a tested seam.
INERT: full pre-existing suite green with zero test-logic edits. +14 tests;
red-proofs RP-1..RP-4 confirmed. The 13 catalog backup: blocks ship in the
same app-catalog change (this controller deploys first).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01A45Qop8YY8tS94bz63LFne
The controller assumed an app's HDD appdata dir is always appdata/<stackName>.
paperless-ngx writes appdata/paperless (stack paperless-ngx), so every consumer
keying by stack name silently missed it via a stat-and-skip. One canonical
resolver appbackup.AppDataDirNames derives the real dir name(s) from the app's
compose ${HDD_PATH} binds; all consumers use it.
- F-S2 (tier-2): RunTier2 mirrors the resolved appdata/<name> (paperless docs
got NO tier-2 copy before). Tier2Info size + RestoreTier2Files live dir use it.
WARN when a declared appdata dir is absent. New tier2Mirror seam.
- F-S3 (migrate, NEW): all six per-app appdata legs (collision/size/copy/verify/
cleanup/skip-set) now loop resolved names. scope="app" migration of paperless
previously copied nothing and left an empty media dir (scope="all" was saved by
the merge walk). WARN on missing declared dir in the copy leg.
- Multi-dir (N>1) refusal: tier-2 backup/info/restore refuse loudly (Hungarian);
migrate supports N. No catalog app hits it today; lifted by Task 3.
- Display: storage page sums resolved dirs.
- Truth repair: the v0.130.0 "tier-2 copies the namespace wholesale" claim is
false; corrected in CHANGELOG + main.go export-adapter comment.
+9 tests; red-proofs RP-1..RP-5 all confirmed. Controller-only, no agent/hub
coupling. Task 1 of the backup-classification-redesign arc.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01A45Qop8YY8tS94bz63LFne
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
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
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
Closes the engine half of drill finding F2: user files under appdata/<stack>
had no customer recovery path (operator copy-back only).
- RestoreTier2Files(stack): single-flight with backup/restore; ALL refusals
before any stop (no Tier-2 record / LastRun empty / copy dir absent →
"nincs másodlagos fájlmásolat"; Tier-2 drive disconnected; live drive
disconnected/decommissioned — Hungarian, flash-ready); source is the
RECORDED CrossDriveBackup.DestinationPath (never a fresh selectTier2Target);
stop → copy → start → waitForHealthy; copy/restart errors surface (F17).
- rsyncRestoreMissing: rsyncMirror's exec shape with the OPPOSITE-direction
flags: -a --ignore-existing --itemize-changes — existing live files are
never overwritten, nothing is ever deleted (the --delete trap this task
exists to avoid). Count = ">f" itemize lines (pure countRestoredFiles).
- restoreFilesCopier seam so orchestration tests never shell out; the one
FS-level test of the real rsync is LookPath-guarded (runs on the Linux
build server + live validation).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PSK5g6qYLknKj8u3QAFEr6
The proceed-path for a missing RESETTABLE secret redeployed the app with the
secret blank (compose "Defaulting to a blank string" → exit 1, live-hit in the
2026-07-04 drill Phase 5). Now the restore generates a fresh credential instead:
- stacks.Manager.GenerateSecretForField: replacement value from the field's
catalog generate spec via the deploy flow's generateValue (no logic copied);
refuses data-keys (defense-in-depth), spec-less and non-secret fields.
- backup.Manager.SetSecretGenerator seam (wired in main.go), consulted in
RestoreFromRecoveryUnit AFTER the untouched fail-closed gate, for missing
names NOT in DataKeyEnvVars. The generated value rides fullEnv into
RecreateStackFromUnit → RedeployFromEnv → SaveAppConfig, so it persists
encrypted in the guest app.yaml and round-trips on the next backup/restore
(no second write path). reconcileRestoreSecrets stays pure and untouched.
- WARNs now discriminate: "generated replacement for X (credential was reset)"
vs "X unrecoverable and has no generator — app may fail to start". Values are
never logged (asserted in test).
- Residual case (documented, not pretended away): if a restored volume tar
carries the OLD internal credential hash, the app may still fail auth until a
manual in-DB reset — generation fully fixes only the fresh-init case.
Companion red-proof: pre-fix behaviour (generation skipped) fails
TestRestoreGeneratesMissingResettableSecret on the non-empty DB_PASSWORD
assertion (verified, reverted). Data-key gate proven unreachable by generation
in TestRestoreGenerationNeverReachesDataKeys.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PSK5g6qYLknKj8u3QAFEr6
DumpAppVolumesSafe had NO production caller: no trigger ever produced
volume-dumps/, so named-volume app data (e.g. nextcloud's html volume) was never
captured into the recovery unit and the granular restore silently restored
nothing for class-B data (drill finding F3).
- runVolumeDumps: per-stack loop in runDBDumpsInternal, BEFORE
captureAllRecoveryUnits (so manifests enumerate the fresh tars). Gate order is
load-bearing: protected-stack and volume-check gates precede DumpAppVolumesSafe
(which stops the stack before its own check — unconditional calls would bounce
every volume-less app nightly). Disconnected/decommissioned drives skip with
the same summary style as the DB loop.
- No silent partials: a per-stack failure lands as a FAIL summary entry, flips
Success, and fails the run ("some backup steps failed: ..."), without aborting
the other stacks.
- Zero-DB early return removed: volume-bearing apps without a database still get
their class-B dump + unit refresh.
- dumpVolumesSafe seam (same style as the F17 discoverDBs/importDBDump seams) so
the gating is unit-tested without Docker. Companion red-proof: neutering the
volume gate fails TestRunVolumeDumps_GatesPrecedeDump (dump fired for the
volume-less stack) and _VolumelessNeverStopped (verified, reverted).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PSK5g6qYLknKj8u3QAFEr6
The backups.html restore panel fetched /api/backup/snapshots (a restic-era route
that no longer existed), so the snapshot dropdown never populated and the
"Visszaállítás indítása" button could never enable — customers could not restore
anything from the UI (drill finding F1, DRILL-appdata-restore-2026-07-04).
- backup.Manager.ListRestorePoints: the keep-side restore has exactly ONE restore
point per app (the current recovery unit); time = newest artifact mtime among
manifest/db-dumps/volume-dumps; tier always 1 (Tier-2 copies are NOT restorable
via POST /backup/restore — never listed); drive_label from the storage registry,
empty for the SSD fallback.
- api: /backup/snapshots route + validStackParam guard (same semantics as
web.validStackName; traversal → 400, unknown stack → 404, no unit → ok+[]).
- Tests dispatch through Router.ServeHTTP (the bug WAS a missing route) + unit
tests for newest-mtime/label/empty semantics. Companion red-proof: hollow
always-[] implementation fails TestListRestorePoints_UnitOnDisk +
TestBackupSnapshots_UnitOnDisk (verified, reverted).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PSK5g6qYLknKj8u3QAFEr6
Live validation surfaced 'option sftp.args is not known' on restic 0.14.0; switch to the
portable sftp.command SSH invocation (ConnectTimeout/StrictHostKeyChecking preserved).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01HxLA1mZurFq9kt8hneFeCs
admin