admin/felhom-controller
1
0 Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

REPORT + CONTEXT: v0.348.0 the global floor; clear_hold proven on the release build
gates / gates (push) Successful in 31s
Details

Browse Source
This commit is contained in:
admin committed 2026-10-29 15:11:17 +01:00
1 parent 2a0a17a760
commit a28db2d17f
2 files changed
+37 -39
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

CONTEXT.md
+8
View File
@@ -1,5 +1,13 @@
# CONTEXT.md — Project Memory
> **2026-10-29 (afternoon, attended) — v0.348.0 RELEASED, the global floor since 14:48 CET (MinAgent 0.158.1).** Image
> `sha256:672fc1c1…` from `2a0a17a` (CI 2515). D64 = B (`09` §3 decision 283): `clear_hold` is the fifth verb of the closed
> operator-action list; any hold reason; the running controller clears memory and `settings.json` together through
> `<data>/ctl/hold.sock` (0600, peer uid checked), and `--clear-restore-hold` goes through it when a controller runs
> (R-1343). R-1340 (both producers name the real cause), R-1341 (OOM line once per kill), R-1342 (no delete during the
> stable-start watch). Proven live on Tester 1 on the release build: a crash-loop hold cleared from the hub stayed
> cleared after the next settings save. Open: R-1350 (an `app_start_failed` 30 s after a clear), R-1347, R-1348.
> **2026-10-29 (attended release run) — v0.347.0 RELEASED, the global floor (MinAgent 0.158.1).** Image `sha256:c79b06e6…`
> from `fd75c7c` (CI 2491); rc.1 = `e3fce09` on scratch 9202. Carries the night's R-1256 (all four edges, H4 + H5's
> failed stop proven live on Tester 1 with a `-tags faultinject` test build — D63), R-1254 controller half, R-528, the
REPORT.md
+29 -39
View File
@@ -1,50 +1,40 @@
# REPORT — v0.347.0 released as the global floor; D62 built and proven; R-1256's last two branches proven live (2026-10-29, attended)
# REPORT — v0.348.0 released as the global floor: „Clear hold" from the hub (2026-10-29, attended)
**Plain words first.** The controller 0.347.0 is the floor for every box since 10:31 CET. It carries the night's restore
edges, the escrow wait, the host's OOM count and the 404 fix, plus D62: a restore copy nobody needs stays 30 days, and the
household sees it in the kept-copies list and can delete it with „Törlés most". D63 held: the two restore branches that
had only tests were proven live on Tester 1, so nothing was reverted.
**Plain words first.** The controller 0.348.0 is the floor for every box since 14:48 CET. The operator can now clear any
held app from the hub (D64 = B); the running controller clears it and saves at once, so a later save cannot bring it
back (R-1343). The box command (`--clear-restore-hold`) goes through the running controller too. The held-app mail names
the real cause (R-1340), the OOM line is logged once per kill (R-1341), and a restore copy cannot be deleted during the
30-minute stable-start watch (R-1342).
## Commits
| Commit | CI (run id / #) | What |
| Commit | CI (job id) | What |
|---|---|---|
| `e3fce09` | 2486 / #537 green | **D62** — `restoreCopyUnusedAge` 30 days; `backup.RestoreCopies` / `DeleteRestoreCopy` (check + removal under `m.mu`, path guard); the list rows and the typed-name form (`POST /backup/restore-copy/delete`, JSON 200/400/404/409/500); `restore_copy_deleted` event. **D63** — `faultPoint` (test builds only, `-tags faultinject`). Tests `TestD62_*`, `TestD63_*`, each seen failing by a mutation. |
| `fd75c7c` | 2491 / #538 green | the v0.347.0 CHANGELOG heading — the release build's commit. |
| `652a690` | 2500 green | R-1342 (from the morning, first released now). |
| `5b3b541` | 2508 green | `clear_hold` in the closed operator-action list; the socket `<data>/ctl/hold.sock` (0600 in a 0700 folder, peer uid checked); `ClearHoldByOperator` also drops a waiting put-back from the app-stop marker; R-1340 hold mail from the hold record; R-1341 OOM line once per kill. Tests red first. |
| `6077ac0` | 2509 green | R-1343 fallback: with no controller running, the box command's own edit goes through the same clear door. Test red first. |
| `99fac06` | 2514 green | R-1340 second producer: `RestoreHold.Why()` builds the restore_mixed text from the record (found live on rc.1). Test red first. |
| `2a0a17a` | 2515 green | v0.348.0 CHANGELOG heading — the release build's commit. MinAgent 0.158.1. |
## Release
- rc.1 (`e3fce09`, digest `b6640c0a…`) on scratch 9202: walk green.
- 0.347.0 (`sha256:c79b06e6…`, from `fd75c7c`): per-box floors for Tester 1, demo-hp, demo-felhom at 09:57 CET — all three
moved within a minute; walk green on all three (controller healthy, every container up, 0 refusals, 0 errors, login and
file manager 200, https-only OK); one local restore on Tester 1 (2/2 volumes, the app ran, no hold); the OOM line (one
`app_oom`, one mail); global floor 10:31 CET, MinAgent 0.158.1, read back from the hub page and its log.
- The release image carries no fault switch (`grep -c "FAULT INJECTED"` = 0; 2 on the test image).
- rc.1 (`5b3b541`+) and rc.2 (`2a0a17a`, `sha256:1d96f60c…`) on 9202; the walk green on rc.2 (6 containers, 0 errors,
login and file manager 200; the hold socket present).
- 0.348.0 from a clean clone of `2a0a17a`: `sha256:672fc1c1deaaf1cbeecfc2c5b04da731b076a9194f4ce7ed9678339fcd4548bf`.
- Per-box floors 14:41 CET (Tester 1, demo-hp, demo-felhom; MinAgent 0.158.1): all three on 0.348.0 within 25 s; the
walk green on all three (no unhealthy container, 0 refusals, 0 ERROR lines, login and file manager 200, the socket
0600) and `https_only_check.py` green.
- Global floor 14:48 CET on the operator's yes (303 `floor_set`; the hub page and the hub log read 0.348.0 / 0.158.1).
## Live proofs (Tester 1; felhom.eu `documentation/audits/d63-2026-10-29/`)
## Live proofs (Tester 1)
- **H4** (test build): the snapshot held 3.8.1, the app ran the true 3.8.0; the fault failed the definition write after
the compose was written; „the live definition is written back before the start", then the start; the app ran 3.8.0, no
hold, both seeded events read back.
- **H5's failed stop** (test build): a busy volume was not written into; the put-back's stop failed; nothing was put back,
the copy stayed, the app was held with the D20 sentence; the held copy's „Törlés most" was off with its reason.
- **D62** (test build, then the release build with hub 0.175.0): the copy listed with its date and „automatikusan
törlődik"; a wrong name 400; the right name 200; gone; the hub stored `restore_copy_deleted`.
- **Clear from the hub, crash-loop hold (rc.1):** pressed 13:44 CET; gone on the box and on the hub; still gone after the
next settings save (14:00:45). No false `app_start_failed` after the clear in 18 minutes (F2: NOT-A-FINDING).
- **Clear from the hub, restore_mixed hold (fault test build):** the page showed the data warning; cleared; the copy kept.
- **R-1340:** the mixed-hold mail names a local restore, the failed put-back stop and the kept copy; no „database".
- **R-1341:** one WARN line over 4 scans; one `app_oom` event.
- **R-1342 (rc.1):** the list says „checking", the button is off, a delete gets 409; removing the app ended the watch
with no alarm.
- **On the release builds:** see the felhom.eu report `REPORT-release-2026-10-29b.md` (one restart, one clear).
## Teardown
Tester 1: throwaway Radicale and Gokapi removed through the product; the 21 empty anonymous volumes this session's backup
presses left (R-1267's PrivateBin class) removed by name — 268 volumes, as before; controller 0.347.0 (the release).
**Radicale's restore hold record stays — clearing it is the operator's.** Scratch 9202: controller 0.347.0 (from rc.1),
359 volumes as before. Hub: nothing provisioned.
## Rows
Filed: felhom.eu R-1340 (the mixed-hold operator mail tells the option-C story for any cause), R-1341 (the OOM WARN every
30 s). Closed by this release: R-1256, R-528 (with the agent).
## After the release — R-1342 (fixed on main `652a690`, NOT released)
At 10:39 CET Tester 1 mailed the operator a false „Gokapi did not start … the copy is kept": the D62 live test had deleted
the copy during the 30-minute stable-start watch and then removed the app. Now the watched copy reads „checking" and is not
deletable, and a removal ends the watch quietly. Tests red first; rides the next controller release.
Evidence: felhom.eu `documentation/audits/release-2026-10-29b/` (`p3-clear-hold/`, `release-controller-0.348.0/`, `final/`).