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

docs: 11 §5.7 System page, §5.8 Docker slow lane BUILT, §5.9 crash restart; 00/03/07/08; decisions 90-94 (CC unattended); runbooks docker-undo + crash-guard; register R-852 R-835 R-848 R-849 R-851 R-854 closed, R-853 R-855 R-856 opened, R-812 R-840 narrowed (334 -> 332); live evidence
gates / gates (push) Successful in 33s
Details

Browse Source 
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS
This commit is contained in:
authentik Default Admin
2026-10-04 17:36:10 +02:00
parent 245886ee92
commit 0c55336fba
38 changed files with 1594 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files
documentation/backlog/CLOSED-ITEMS.md
+13
View File
@@ -26,6 +26,19 @@
---
## 2026-10-04 (evening) — System page, Docker slow lane, crash guard (agent v0.142.0, hub v0.132.0, installer 1.30.0)
> Evidence: `audits/os-docker-crash-2026-10-04/`.
| Row | What | Closed | Evidence |
|---|---|---|---|
| **R-852** | **The operator could see no box's Debian, Proxmox, kernel or Docker version anywhere.** The box reports them (`system` stanza: Proxmox + kernel from the API, the wrapper's read-only `facts`); hub v0.132.0 shows them on the new **System** page (with the ring / switch / approve buttons) and a Proxmox / kernel column on Hosts. **Reasoning kept: a value the box could not read says `unknown`, amber, with the reason — never empty, never guessed.** | CLOSED 2026-10-04 — FIXED (`09` decision 89) | `partA/` |
| **R-835** | **Turning Docker's live-restore OFF by a restart stops every container and starts none.** Resolved by design (decision 87): live-restore is ON everywhere — the golden bakes it, installed boxes get it once by a RELOAD (measured: same ids on 9202 6/6, demo-hp 24/24, demo-felhom 5/5), and the wrapper refuses a Docker step while it is off (R15). **Rule kept: never turned off by a plain restart.** | CLOSED 2026-10-04 — RESOLVED BY DESIGN | `partB/b1..b4` |
| **R-848** | **A held host package was invisible to the hub.** The facts report `held`; the System page shows it amber. | CLOSED 2026-10-04 — FIXED | `partA/` |
| **R-849** | **The guest's "restart needed since" never cleared.** The wrapper scans the guest on every pass too (agent v0.142.0). | CLOSED 2026-10-04 — FIXED | `partB/agent-redproofs.txt` |
| **R-851** | **A host that panicked stayed stopped (`kernel.panic = 0`).** The crash guard (decision 88): `kernel.panic = 10`; the 3rd unclean stop within 60 minutes leaves the box off; re-arms after 24 h or by `felhom-crash-guard rearm`. MEASURED on demo-hp: crash 1 and 2 restarted by themselves (54 s, 53 s), the guard tripped, crash 3 stayed off until the operator switched it on; the hub mailed the trip. **Reasoning kept: a crash, a power cut and a hard reset cannot be told apart on these boxes (pstore saved nothing for a real panic) — the guard counts every unclean stop.** | CLOSED 2026-10-04 — FIXED | `partC/` |
| **R-854** | **The first Docker steps were stored as GUEST reports** — they ran (agent rc) before hub v0.132.0 was live, and the older hub maps an unknown layer to `guest`; the guest candidate briefly read 0 packages. One-time; a pass with the released agent re-reported both layers. Nothing to fix. | CLOSED 2026-10-04 — ONE-TIME, NO ACTION | `partB/b5-*` |
## 2026-10-04 (afternoon) — OS updates, host fast lane + fleet view + alarms (agent v0.141.0/v0.141.1, hub v0.131.0/v0.131.1, controller v0.292.0)
> Evidence: `audits/os-host-lane-2026-10-04/`.