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.
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
ListDumpFiles ran ValidateDump (line-by-line scan) for every dump on every ~5-min
RefreshCache cycle — wasted I/O+CPU on large customer dumps. ListDumpFiles now takes
an optional cached(name,size,mod) lookup; on a (size+modtime) match it reuses the
prior result and skips ValidateDump. settings.DBValidationCache gains Size+ModTime;
listAllDumpFiles builds the lookup from the persisted cache and writes back only fresh
validations (cache miss), so an unchanged dump triggers neither a re-validation nor a
settings.json write each cycle. nil cached = legacy validate-always (back-compat).
Tests: cache-hit skips validate (sentinel), cache-miss validates, nil validates.
deriveStackName pure-suffix-stripped on '-' (postgres/db/mariadb/.../cache), so a
stack whose slug ENDS in a role token (e.g. 'my-cache') was misattributed (stripped
to 'my') — filing its DB dump under the wrong/nonexistent stack. Now threads the set
of deployed stack names (m.knownStackNames() <- ListDeployedStacks) into
DiscoverDatabases and cross-references: candidate suffix-strip if known, else the
container name if it IS a known stack, else longest known stack that is a prefix
(handles <stack>_postgres / <stack>-1), else legacy strip. nil/empty known = legacy
behaviour (appexport passes nil). Table test incl. the my-cache case (fails pre-fix).
The restore paths (RestoreFromRecoveryUnit + the RestoreApp fallback) repopulated
Docker volume tars but NEVER replayed the captured <stack>-<dbtype>.sql dump, so
DB-resident data (e.g. rows in a DB whose data dir is a bind mount) did not come
back — the romm marker round-trip in the audit lost the row.
New appbackup.ImportDump (read-side counterpart to DumpOne) replays a .sql/.sql.gz
into the running DB using the live container's OWN discovered credentials (no env
threading; reuses DiscoveredDB + getMariaDBPassword). backup.reimportDBDumps
orchestrates it AFTER volume restore + stack bring-up, so the logical dump WINS
over any volume-tar copy of the DB (operator-chosen precedence). pg_dump
--clean --if-exists and mariadb-dump (default --add-drop-table) make replay
idempotent; psql ON_ERROR_STOP=1 surfaces real import errors.
Also: volume-restore per-volume failures and DB-import failures now SURFACE (the
restore returns an error) instead of a swallowed WARN, so a failed data restore
cannot read as success.
Tests (restore_db_test.go, injectable discover/import seams): imports when dump+DB
present, failure surfaces, no-dump skips discovery, dump-but-no-matching-DB is a
non-fatal skip. Live DB round-trip to be validated post-deploy.
Extract the stateless, keep-side app-data backup primitives out of
internal/backup/ into a new self-contained internal/appbackup/ package:
- dbdump.go: DB dump discovery/execution (DiscoverDatabases, DumpOne, ...)
- appdata.go: StackDataProvider + app-data/volume discovery, HumanizeBytes
- paths.go: keep-side path helpers (AppDBDumpPath, AppVolumeDumpPath, AppDataDir)
backup/ keeps every name available via type/const aliases + one-line function
forwarders (appbackup_bridge.go), so the still-present delete-side code
(restic, cross-drive, drive-mount) and the both-side consumers (web/api/report)
compile unchanged. The keep-only consumers appexport and storage are rewired to
import appbackup directly and no longer import backup.
This is the Part-2 prerequisite for the Proxmox port: appbackup has zero
references to restic/cross-drive/drive-mount and does not import backup, so the
delete-side can later be removed without breaking app-data backup or appexport.
Behaviour-preserving: pure move + import/qualifier rewrites, no logic edits.
The four Manager methods (RunDBDumps/DumpAppVolumes/DumpAppVolumesSafe share the
delete-side mutex/status state; RestoreAppFromTier2 reads the cross-drive mirror)
intentionally stay on Manager and delegate to appbackup — for the re-platform step.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
admin