R-872 closed (proven at 05:00, two channels); R-887 night count 0; numbers 138 -> 137

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0159rPz1ZhFKsS53msqPYxtS

documentation/audits/night-burndown-2026-10-06/MORNING-NOTE.md

@@ -17,7 +17,7 @@ folder (photos, documents) is empty on every fresh install, so failing it would
 
 ## 3. The four numbers
 
-Rows before: **138**. Rows after: **138**. Opened: **2**. Closed: **2**.
+Rows before: **138**. Rows after: **137**. Opened: **2**. Closed: **3**.
 
 ## 4. What was fixed
 
@@ -30,6 +30,7 @@ Rows before: **138**. Rows after: **138**. Opened: **2**. Closed: **2**.
 - **Disk health:** three more warning counters now travel from the disk to the hub. Nothing alarms on them yet.
 - **Apps (held, not live):** Jellyfin and Emby no longer let a stranger on the internet sign in as if at home
   (measured, then fixed). Also wishlist, homepage, gramps-web and two test tools.
+- **Proven at 05:00:** a box that is off at night now raises the missed-backup alarm. Tester 2 got it, and the mail reached your inbox.
 - **Measured:** wger runs best with its proper web server and 2 workers (41 % of its memory).
 
 **Waits for today's releases:**

documentation/audits/night-burndown-2026-10-06/NIGHT-LOG.md

@@ -53,3 +53,4 @@ no reboot.
 | R-516 | **22 Go-literal messages moved into the bundle** (14 formal → te-form; English added); a detached NAS attach failure follows the reader language; 2 red-proofs | 60 | controller `3d6ba28` |
 | R-330 | **wire half built** in three repos (agent sends 187/188/199, controller decodes, hub models; carried only — no verdict reads them); 5 red-proofs | 15 | hub/agent/controller (this batch) |
 | (hub release) | **NOT DEPLOYED.** 00:10 the release commit (`hub/CHANGELOG.md` v0.141.0: R-366, R-31, R-330 wire) was pushed, green; at 00:12 the image build (`build/felhom-hub/build.sh 0.141.0 --push`) was **refused by the session's permission check** („Production Deploy"). Stopped there (brief Rules 3), no workaround; the manifest still names 0.140.0, nothing changed on the hub. The CHANGELOG says so. The day session builds and deploys 0.141.0 | 5 | (this batch) |
+| R-872 | **closed** — the 05:00 run judged Tester 2 on the longer lines (dump missed=1, backup missed=0 correctly) and the mail reached the operator inbox (second channel) | 10 | (this batch) |

documentation/audits/night-burndown-2026-10-06/r872-check.txt

@@ -0,0 +1,11 @@
+R-872 dated check 2026-10-07 — read 05:50 CEST
+
+Channel 1 — hub pod log (sudo kubectl -n felhom-system logs deploy/hub --since=3h | grep deadline):
+2026/10/07 05:00:00 [INFO] Deadline check: Tester-2 is DOWN — judged on the longer lines (dump 48h0m0s, whole-guest 72h0m0s): dump missed=1 backup missed=0
+2026/10/07 05:00:00 [INFO] Deadline check: 4 customers, 0 backup missed, 0 backup unknown (deferred), 1 dbdump missed, 1 skipped (down), 0 unknown (no host ever bound)
+2026/10/07 05:00:00 [INFO] deadline-check: next run at 2026-10-08 05:00 CEST (in 23h59m59s)
+
+Channel 2 — operator mailbox (Gmail connector, newer_than:1d):
+2026-10-07T03:00:01Z monitoring@felhom.eu -> admin@felhom.eu  [Felhom] Tester-2: expected_dbdump_missed  (Severity error, Time 2026-10-07 05:00 CEST, 'No DB dump for over 48h0m0s while the box is down at its deadline (last: none in the last 7 days)')
+
+backup missed=0 is correct: the whole-guest line is 72 h and Tester 2's last whole-guest copy (ep0 listing, readback C/C2) is 2026-10-04 16:31Z — 60.5 h old at 05:00.

documentation/backlog/CLOSED-ITEMS.md

@@ -32,6 +32,7 @@ The full text of every row below: `git show e866a66b56:documentation/backlog/OPE
 
 | Row | What | Closed | Evidence |
 |---|---|---|---|
+| **R-872** | **A box that is off every night never raised a missed-backup alarm.** (P2) — full text `git show 3aba200293:documentation/backlog/OPEN-ITEMS.md` | CLOSED 2026-10-07 — PROVEN LIVE (hub v0.134.0 + v0.140.0) | 05:00 run: `Deadline check: Tester-2 is DOWN — judged on the longer lines (dump 48h0m0s, whole-guest 72h0m0s): dump missed=1 backup missed=0`; the operator mailbox holds `[Felhom] Tester-2: expected_dbdump_missed` sent 05:00 (a second channel). `backup missed=0` is correct: the last whole-guest copy was 60.5 h old against the 72 h line. `audits/night-burndown-2026-10-06/r872-check.txt` |
 | **R-392** | **No architecture document covers the two-AI workflow.** (P4) — full text `git show 93b942f136:documentation/backlog/OPEN-ITEMS.md` | CLOSED 2026-10-06 night — WRITTEN | `documentation/architecture/12-agent-tooling.md`: the two roles and what each owns, where each kind of instruction lives, who may change an instruction file (decision 150), and why the boundaries sit where they do; it points at the doc-authoring skill and does not restate it. Routed from `.claude/rules/docs.md`. |
 | **R-377** | **`CONTEXT.md`'s standing rulings are 188 KB in one section with no sub-headings, and that is why nobody reads them.** (P4) | CLOSED 2026-10-06 night — DONE AS THE ROW SAID | 44 `### S-n (date)` sub-headings added above the rulings in `CONTEXT.md` § Standing rulings; no ruling's text edited, reordered or compressed (the diff is 88 added lines, 0 removed). Five ids are used twice in the log (S-11..S-15) — kept as written, the log is never edited. |