Commit Graph
260 Commits
Author SHA1 Message Date
admin c7e4ec9b5e Review fix 3: every install deletes the off-site end record — one point, SaveAppConfig (D72)
A `.fab` import installed the app and its end record survived; the restore
that installs from a unit had the same hole. Only DeployStack deleted it.

- stacks.SaveAppConfig, the only writer of an installed app's app.yaml, calls
  noteAppInstalled when the record says deployed (a hook per stacks directory,
  registered by SetMigrationDeps). A source pin lists every place that sets
  Deployed or writes an app.yaml; a new writer fails it.
- The living check is stacks.Manager.IsLivingApp (wired in main.go): the
  protected list, the manager's list (deployed, being deployed, protected) and
  the app's record on disk — an installed app the manager does not list is
  living. Tested by behaviour.

Tests seen red before the fix; 8 mutations seen red (felhom.eu
documentation/audits/build-release-4-2026-11-15/A/redproof-review-3.txt).

Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
2026-11-15 14:42:07 +01:00
admin f628cb5082 The off-site copies of a removed app end 30 days after „delete backup data” — the box's half (felhom.eu R-1374 part 1; D72, D84)
gates / gates (push) Successful in 41s
Built, not released; no box has run it. felhom.eu `07-backup-architecture.md` §8.2f.
It does nothing without a hub that knows it: an older hub agrees to no id.

- The record: settings `offsite_ends` {app, removed_at, ends_at}. One writer
  (backup.StartOffsiteEnd), called by a removal only with „delete backup data”
  and by „Törlés 30 nap múlva” on the restore page. A second press changes
  nothing. No record, no deletion. Only on the hub-provisioned tier.
- The date is on the remove dialog before the press; the restore page shows
  „Törlésre kijelölve …” with „Mégis megtartom”, and one history line per
  carried-out end. An install of the same app deletes the record.
- The weekly window: after the guard has passed the policy's plan under the cap
  as before, it adds the snapshots whose app tags are ALL due records, that the
  hub agreed to and the box named in the same request; the result names them
  apart (removed_for_ended_apps). `_shares`, the box's own stacks and an
  installed app are never due.
- The report carries the pending records; the household gets one mail per
  press (offsite_end_scheduled), in both languages.

MinAgent unchanged (0.158.1). 143 parity fixtures gained the dialog's one new
script line (patched, not regenerated); one new parity case.
Tests TestD72_*, TestD84_*; 56 mutations seen red
(felhom.eu documentation/audits/build-release-4-2026-11-15/A/redproof-ctrl-*.txt).

Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
2026-11-15 14:09:47 +01:00
admin 844f19f11e fill watch: one warning per filesystem (R-1371) — committed, not released
gates / gates (push) Successful in 31s
On a box whose guest has one data volume the two built-in fill targets, "/"
(the controller's overlay on the Docker data volume) and /mnt/sys_drive, are one
filesystem. The target list was de-duplicated by path only, so one fill would
send two warnings under two names.

fillTargets now keeps one target per filesystem; the first stays (app area,
system area, drives). The identity is system.FilesystemKey, not st_dev alone:
inside the container "/" is an overlay with its own device number, so an
overlay path takes the device of a Docker-kept file on the data-root
(/etc/hostname, /etc/hosts, /etc/resolv.conf). Measured read-only on demo-hp
9201/9202 and the N100 on 2026-11-10.

Tests: TestR1371_* (3), TestFilesystemKey_* (2). Red seen three times.
No version, no image, no floor: it waits for a release session.

Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
2026-11-10 14:18:36 +01:00
adminandClaude Opus 5.5 27360d6abe v0.351.0 (unreleased): old -rc controller images and old infra images are deleted (decision 56 amended, decision 285)
gates / gates (push) Successful in 31s
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
2026-11-09 10:23:06 +01:00
admin 37644458c7 R-1348: the mail of an interrupted put-back names a restore, not an off-site run (seen on the live proof); changelog: R-1347, R-1348 proven live on 9202. Commit only.
gates / gates (push) Successful in 32s
2026-11-03 01:12:19 +01:00
admin c4b6844d34 Traefik v3.7.17 (GHSA-xv5x-242w-7hf7); R-1347 a restore forgets the update-held flag; R-1348 a failed put-back at start-up mails; R-1339, R-1351, R-1323 truthful start-up lines and one lock over FileBrowser's compose. Commit only.
gates / gates (push) Successful in 29s
2026-11-02 23:14:16 +01:00
admin b69e90bdad D65 = A: the busy file the agent reads before an operator reboot, and the household's 'Felhom support restarted your box' line; R-1350: a hold clear records the wanted stop so no app_start_failed follows it
gates / gates (push) Successful in 32s
2026-10-29 16:24:40 +01:00
admin 99fac06aa7 R-1340 second producer: a restore_mixed hold's Why() names the cause from its record (found live on rc.1: the hub's Held apps line said off-site/database for a local restore with no database) — test red first
gates / gates (push) Successful in 30s
2026-10-29 14:35:07 +01:00
admin 6077ac0083 R-1343 fallback: the box command's own settings.json edit (no controller running) goes through the same clear door, so a waiting put-back is dropped from the marker there too — test red first
gates / gates (push) Successful in 30s
2026-10-29 12:59:41 +01:00
admin 5b3b541602 clear_hold (D64 = B) + the box command through the running controller (R-1343), R-1340, R-1341 — tests red first; commit only
gates / gates (push) Successful in 31s
- clear_hold: a fifth verb in the closed operator-action list; lifts a hold of any reason in the RUNNING
  controller (memory + settings.json together), forgets the update's held flag, drops a waiting put-back
  from the app-stop marker (copy kept). Refused on a bad name or no hold. No new event type.
- --clear-restore-hold asks the running controller over <data dir>/ctl/hold.sock (unix, 0600, peer uid
  checked, no TCP); edits settings.json only when no controller answers, and says so (in-guest restart).
- R-1340: the operator's HELD STOPPED mail is built from the hold record (restore kind, mixed cause,
  failed put-back step, database replay or not, kept copy) — never the option-C database story by default.
- R-1341: the OOM WARN line once per new kill (container, start time, kill count); event unchanged.
2026-10-29 12:49:46 +01:00
admin 652a690e5d R-1342: a restore copy under the stable-start watch is not deletable (D62 gap); an app removed during the watch raises no alarm — tests red first; commit only
gates / gates (push) Successful in 31s
2026-10-29 11:16:13 +01:00
admin e3fce09d96 D62: restore copies kept 30 days, listed with their date, deletable by hand (typed name, never a needed copy, operator event); D63: a test-build-only fault switch for the R-1256 H4/H5 live proofs — tests red first
gates / gates (push) Successful in 33s
2026-10-29 08:50:59 +01:00
admin eed23e7594 R-528: the 30-s OOM scan also reads the host's per-container kernel counter (agent GET /guest/oom-kills)
gates / gates (push) Successful in 32s
A count above 0 in a running container's current run fires the existing app_oom line (source cgroup/docker in the
details), whatever Docker's OOMKilled says. Old agent: 404 back-off; no agent: Docker's flag only.
Tests TestR528_* red first / red-proved. Commit only.
2026-10-28 23:22:46 +01:00
admin 342a394bff R-1256: four restore edges (review round 2 HUNT-1 H4/H5/H6/H8) — each test seen failing first
gates / gates (push) Successful in 31s
H4: a failed definition write in the off-site restore writes the live definition back before the start;
    a failed write-back holds the app (restore_mixed).
H5: a put-back whose stop fails pours nothing, keeps the copy and holds the app (off-site, local, power-cut);
    a failed definition write-back keeps the copy (D20); a failed 'volume rm -f' fails that volume (no merge).
H6: TryBeginRestoreOp/RestoreBusy: check every operation and record the restore in one step under m.mu;
    every web handler and the kept loads begin through it; install-time kept load asks about the mirror too.
H8: SweepUnusedRestoreCopies every 6 h: removes a copy only with no hold, no marker naming it, no restore
    running, and older than 14 days.

Commit only; rides the next controller release.
2026-10-28 22:24:16 +01:00
admin b156e4526a fix(R-1298): held-app reports only for installed apps (a restore hold always)
Red-proved. Commit only.
2026-10-26 20:26:44 +01:00
admin 5bd03ad152 feat(D53): the hub learns which apps are held — app_held / app_hold_cleared, daily repeat
gates / gates (push) Successful in 54s
internal/holdreport: one observer of the hold store, ticked every minute; app_held for a new or changed
hold (stack_name, reason, since, why), app_hold_cleared when gone (also while the controller was down:
settings.reported_holds), every hold again daily and at start. Severity info. RestoreHold.Why/WireReason
(holdWhy calls it). 8 tests, red-proved (inert Tick; unwired scheduler line). Commit only (felhom.eu 08 §10).
2026-10-26 18:23:54 +01:00
admin 838c834169 R-1286: a held app says why it is held — each hold reason gets its own sentence in the start gate, the boot sweep and --restore-holds (an unhealthy stop read as a failed restore). Red-proved. Commit only.
gates / gates (push) Successful in 1m28s
2026-10-25 22:05:23 +01:00
admin b5863a96eb R-1288: a restore of a set-aside unit sets the replaced install's unit aside at once, with no alarm. R-1282: the later media restart time is stored and resumed at start, once. Red-proved. Commit only.
gates / gates (push) Successful in 1m26s
2026-10-25 21:37:55 +01:00
admin a79a292d72 R-1261: the line hold holds every running seeder, drops the record of an app removed while held, and puts back the qBittorrent slot it held (alternative speed). Red-proved. Commit only.
gates / gates (push) Successful in 1m29s
2026-10-25 20:38:28 +01:00
admin 935e624f10 D47 follow-up: a set-aside tells the operator once (backup_run_failures, leg unit-set-aside, both identities, own run id). Red-proved. Commit only.
gates / gates (push) Successful in 1m33s
2026-10-25 17:19:08 +01:00
admin 3a1f09d29f Fix A (found live on 9202): the container sampler reads docker inspect JSON — a -f template stopped at 'no entry for key Health' for every app without a healthcheck, so a healthy restore read 'not started'. Commit only.
gates / gates (push) Successful in 1m27s
2026-10-25 17:02:32 +01:00
admin 1ee0353dbd Fix A: the container sampler keeps Docker's stderr in its error (a live run on 9202 logged only 'exit status 1'). Commit only.
gates / gates (push) Successful in 1m24s
2026-10-25 16:56:04 +01:00
admin 7720439440 D49 = A follow-ups: (A) the undo copy is released only after a stable start — 3 min of running, healthy, no restart, one rule for the first verdict and the late-start watch; a watch ends only for a restore begun after it; (B, R-1279) the .fab import downloads first with the restores' D48 helper, before it removes anything. Red-proved. Commit only.
gates / gates (push) Successful in 1m26s
2026-10-25 12:45:33 +01:00
admin 2d71270858 D48: a restore downloads first — every restore path resolves and pulls every image of the definition that will run before the stop, under the image-work lock held until the restore returns; a failed pull refuses with the app untouched (a deleted version and a busy registry read differently), one operator line. Red-proved. Commit only.
gates / gates (push) Successful in 1m26s
2026-10-25 09:38:28 +01:00
admin b680db85cb R-1270 (P2): a restore whose app does not reach running in time is never a success — every restore path returns ErrRestoreNotStarted, the household reads that the data is back but the app has not started yet, one operator alarm; the data stays, the undo copy is kept, a late start turns the record into a success and releases the copy. Red-proved. Commit only.
gates / gates (push) Successful in 1m32s
2026-10-25 09:33:26 +01:00
admin 02daebe9c5 D47 / R-1269 (P2): a new install never writes into an earlier install's backup — every install has an identity (install_id), the unit records it, every unit writer asks first; another install's unit is set aside whole (backups/set-aside/<app>/<stamp>, a D25 kept copy, „Visszaállítás” + „Eltávolítás” on the restore page); a move that fails refuses the write. Red-proved. Commit only.
gates / gates (push) Successful in 1m33s
2026-10-25 09:27:48 +01:00
admin c9b4b82dc7 R-1253 (P2): an export that stops its app refuses while another operation (a restore) holds the app-stop marker — its Undo is never replaced. Commit only.
gates / gates (push) Successful in 1m28s
2026-10-24 20:29:59 +02:00
admin 518a13592d R-1252 (P2): a rejected or absent backup block never deletes second-drive copies — reconcile only on a valid classification; a rejected block is said on the app's status and once a day to the operator (backup_run_failures); kept copies stay kept (H2); Remove drops the kept records it deleted (H9). Commit only.
gates / gates (push) Successful in 1m18s
2026-10-24 20:07:16 +02:00
admin 062f9b41d9 R-1251: sync the restore copy's folders before the marker names it (a cut left undo.json as .tmp on Tester 1); the interrupted-restore record says the put-back waits, not 'run it again'. Commit only.
gates / gates (push) Successful in 1m24s
2026-10-24 19:36:32 +02:00
admin bb2809dc68 R-1251 (P1): a pending put-back blocks every start — a cut restore's copy on an unbound drive waits (a restore hold), is poured back when the drive binds before any start; a failed put-back holds the app (H2); Stop during a restore still puts back (H7). Commit only, unreleased.
gates / gates (push) Successful in 1m18s
2026-10-24 19:07:02 +02:00
admin 046b808792 upwatch: hold a seeding torrent app for one line test, then put its limit back exactly (R-1250)
gates / gates (push) Successful in 1m22s
When the agent waits for quiet and qBittorrent or Transmission runs: record the limit first,
hold at 1 KiB/s, ask the agent for a test on request, put the old limit back (deferred; a
household change during the test stays). Start-up puts back a limit a killed controller left;
the 50 % rule skips while a hold record exists. Commit only.
2026-10-24 13:19:07 +02:00
admin 42cda07b42 v0.342.0 (unreleased): the watch's start line no longer says the line is learnt only from the off-site push
gates / gates (push) Successful in 1m22s
2026-10-24 09:58:13 +02:00
admin ed1011b088 v0.342.0 (unreleased): the upload watch reads the host's weekly line test (GET /host/line-test, agent >= 0.164.0); qBittorrent's limit = 50 % of the TESTED line only (D41), the watch uses the higher of test and push floor, the system page names the source; red-proved
gates / gates (push) Successful in 1m15s
2026-10-24 08:59:57 +02:00
admin a9c1997fbc v0.342.0 (unreleased): the upload and CPU watch — an app that sends or burns far more than its own numbers explain is one operator-only alarm; the line learnt from the off-site push; qBittorrent's default upload limit (never over the household's)
gates / gates (push) Successful in 1m17s
internal/upwatch: exact per-container counters from the Docker Engine API, per-app judging (upload >= max(5 Mbit/s,
25% of the line) for 15 min; CPU >= max(150%, 50% x cores) for 30 min), explainers qBittorrent / Transmission / Plex,
app_unexplained_upload / app_unexplained_cpu + _cleared (hub 0.169.0). The off-site push is sampled once a second
for the line (a lower bound). qBittorrent limit = 50% of the line, the household's limit always wins. 20 red-proofs.
Spike + design: felhom.eu audits/SPIKE-upload-watch-2026-10-23.md, architecture/08 section 9. Not released.
2026-10-23 21:26:09 +02:00
admin 943c517550 Review fixes for 0.340.0: C5 a catch-up releases the whole-guest backup when its backup legs end (red-proved); C3 the R-1187 c refusal text fits an unhealthy app; C6 two comments; C8 README backup.volumes wording; unreleased
gates / gates (push) Successful in 1m40s
2026-10-22 21:25:48 +02:00
admin 66a69d4954 Stray-copy fix: the capped second-drive copy container never outlives the copy — fixed name + label, removed by name on any error (the timeout included), every leftover removed at start before the scheduler and web server; a restore is refused while a copy runs or may still run; --mount (refuses a path the daemon does not see) and the image ID with --pull never (review); red-proved; unreleased
gates / gates (push) Successful in 1m53s
2026-10-22 20:20:35 +02:00
admin d5f88a2182 R-1187 (c): refuse a .fab import over an installed app that is not running (it merged old and bundle data); R-1189 (d): the off-site proof + integrity check run after a catch-up; unreleased
gates / gates (push) Successful in 1m54s
2026-10-22 17:27:56 +02:00
admin 6a17c95815 R-49 (D21 = B): a template may exclude a named volume from every backup (backup.volumes, class excluded) — Immich's model cache; guarded, red-proved; unreleased
gates / gates (push) Successful in 1m50s
2026-10-22 16:21:38 +02:00
admin add310c80a R-1199 reopened: a box with no list of its own reads the household's from the hub and adopts it, pushes its default once only when no household saved one; every push names its source; an empty saved list stays empty; R-1216 (D28 = A): the install page's swap sentence only when host swap is really short (shared case table); unreleased
gates / gates (push) Successful in 1m55s
2026-10-22 12:13:26 +02:00
admin ba22f9b2d9 R-1199: a box whose household never saved its notification choice pushes the default list its page shows at start, so the hub mails by it; unreleased
gates / gates (push) Successful in 1m37s
2026-10-21 23:40:21 +02:00
admin d582feb245 D22: every restore puts the data back exactly on failure and after a power cut (R-1177, R-1195); whole restore asks the unit first (R-1178); no mirror during a restore (R-1180); W22 restore refusals in Hungarian; unreleased
gates / gates (push) Successful in 1m40s
2026-10-21 21:19:46 +02:00
admin 5929b8e345 kept second-drive copies (D25: R-1175, R-1181, R-1179); .fab import keeps the family gate (R-1191); stale restore temp (R-1188 c); the overcommit sentence made true (R-1192); escrow before the floor (R-1176); unreleased
gates / gates (push) Successful in 1m20s
2026-10-21 15:46:10 +02:00
admin 8f5bea0148 R-1182: a finished off-site abandonment closes on the hub's abandon_purge_done (hub v0.163.0 keeps other copies' packages); unreleased
gates / gates (push) Successful in 1m18s
2026-10-21 13:36:12 +02:00
admin 93fb251ea9 R-893 option A (operator D20): a failed off-site restore puts the app's live volumes and definition back exactly and starts it; fit check before the stop; the held sentence made true (R-1169)
gates / gates (push) Successful in 1m11s
2026-10-20 09:17:00 +02:00
admin 15ea002458 v0.335.0: no app stop for a tier the agent will refuse because its host is busy (host_busy pre-check, R-921)
gates / gates (push) Successful in 53s
2026-10-19 20:05:52 +02:00
admin 546eef37bd v0.334.0: a live box refreshes its config with its own hub key (GET /box-config; R-1159, operator D19)
gates / gates (push) Successful in 49s
2026-10-19 15:55:10 +02:00
admin 6d90374f63 v0.328.0: the catalog allow-list — the box refuses container rights a template asks for and the box does not give (R-971, D2 = A)
gates / gates (push) Successful in 45s
internal/privpolicy checks the RENDERED compose (privileged, cap_add, devices, host namespaces, security_opt,
sysctls, volumes_from, build/include/extends, env_file/config files outside the stack dir, aliasing named volumes,
binds outside the box's data roots, the Docker socket, propagation, ${VAR} in a privileged key) against an
allow-list compiled into the controller (one entry: homepage's read-only socket). Asked before every compose
write: the 15-minute sync, the guarded update/ladder step, the install, both settings changes, both restores
(before the stop). A refusal writes nothing, keeps the app on its last good compose, raises the operator-only
event app_definition_refused once per definition and puts one line on the app page. The sync never copies a
template named like a controller infra stack. Structural guard TestAllowList_IsWiredIntoEveryComposeWriter.

Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
2026-10-17 16:34:12 +02:00
admin 6ed223886d R-1112: open the family store when the controller starts
gates / gates (push) Successful in 46s
Found on demo-hp with 0.325.0-rc.1: the store (and so the conversion of an old family.json) opened only on the
first family request; a box nobody visits through a family gate kept the session values. OpenFamilyStore from
main, one INFO line with the count. TestR1112_StartConvertsTheFamilyFile, TestR1112_MainOpensTheFamilyStore,
red-proved.

Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
2026-10-17 09:56:12 +02:00
admin f75b3aa778 v0.324.0: the database password comes back with the data (R-1106, R-127b, R-926); restic 0.19.1 (R-1107); R-1108, R-1065, R-1017; dashboard hardening (R-983, R-976)
gates / gates (push) Successful in 44s
2026-10-16 22:46:00 +02:00
admin fbcc3cf042 Debian 13 runtime with restic 0.18.1 pinned outside the OS; restic assumptions re-proven; felhom-samba 1.2.0 (R-1105, R-972)
gates / gates (push) Successful in 49s
2026-10-16 20:33:57 +02:00