disable MQTT sensor bundles (bme280, co2mon)

[bborbe]

Why

The MQTT broker rasp4.hm.benjamin-borbe.de was decommissioned — no DNS record anywhere, no nodes/hm.rasp4.py. Its successor homeassistant.hm.benjamin-borbe.de (192.168.177.8) is alive (HTTP 200 on :8123) but runs no MQTT listener — :1883 and :8883 both refuse connections.

The whole consumer chain is already off: mosquitto, telegraf, influxdb and grafana are enabled on zero nodes. Only the three producer bundles were still enabled, publishing into a void.

This was invisible to the monthly sweep: bme280 on rasp3 stays active with NRestarts=0 while failing DNS in an in-process retry loop, so systemd, bw apply and bw verify all reported it healthy.

Not needed right now, so disable it — reversibly.

What

Flip enabled to False in three node metadata blocks:

File	Block
nodes/hm.rasp3.py	bme280
nodes/hm.co2hz.py	co2mon
nodes/hm.co2wz.py	co2mon

No bundle code changes — both bundles already implement a complete teardown in their else: branch (stop + disable the service, delete user/group and /opt/<svc>, and for co2mon the 90-co2mini.rules udev rule).

All mqtt-* metadata is deliberately left in place so re-enabling is a one-line flip, and each site carries a comment explaining why and what re-enabling requires.

Re-enable checklist (for later)

1. Flip the flags back to True
2. Repoint mqtt-host from rasp4 to the live broker
3. Start Home Assistant's Mosquitto broker add-on
4. Give rasp3 a route to 192.168.177.0/24 — it currently has internet but no path to that subnet (default via 192.168.50.1, tun0 carries only 172.16.90.0/24)

Testing

• bw test hm.rasp3 hm.co2hz hm.co2wz passes
• bw metadata confirms all three resolve to enabled = False
• bw items hm.rasp3 shows the teardown items (file:/opt/bme280, user:bme280, group:bme280, svc_systemd:bme280)

Note

hm.co2hz and hm.co2wz are currently unreachable (answer ICMP, :22 times out — suspected openvpn-client rollout regression, co2hz was the canary in 3acddbc). Their flip is declarative only and takes effect when they return. Only hm.rasp3 can be applied now.

[ben-s-pull-request-reviewer]

The only triggers in the index matching non-Go/Node files are CHANGELOG.md, CLAUDE.md, LICENSE*, Makefile, go.mod, **/*.md, etc. The changed files (nodes/hm.co2hz.py, nodes/hm.co2wz.py, nodes/hm.rasp3.py) match none of these — they're Python node configuration files with no corresponding rule triggers.

Step 4b-i result: 0 candidates.

Per the selector-mode guide short-circuit: applicable set is empty AND mechanical findings are empty → selector clean — no adjudication needed.

---

PR Review Report

PR: disable MQTT sensor bundles (bme280, co2mon) — #42
Base: master ← Head: feature/disable-mqtt-sensors

Selector Mode Traceability

• Candidates: 0 rules matched by Step 4b-i glob filter
• Applicable: 0 rules selected by Step 4c-sel
• Skipped: (none — no candidates to evaluate)

---

Must Fix (Critical)

None.

Should Fix (Important)

None.

Nice to Have (Optional)

None.

---

Summary

The PR disables two MQTT sensor bundles (co2mon on hm.co2hz/hm.co2wz, bme280 on hm.rasp3) by setting enabled: False. The embedded comments explain the decommissioned broker context and the re-enablement procedure clearly. The change is a correct, low-risk configuration disable — no mechanical findings, no applicable judgment rules, and no code-quality issues.

Notes

• LICENSE present ✓
• precommit skipped (selector mode) — CI covers lint+test
• Step 4 skipped: no rule-relevant files changed

---

No worktree cleanup needed — we were already at PR head on the feature branch worktree.

{
  "verdict": "approve",
  "summary": "Clean PR that correctly disables decommissioned MQTT sensor bundles with well-written comments explaining the decommission and re-enablement procedure. No mechanical findings, no applicable rules, no code quality issues.",
  "comments": [],
  "concerns_addressed": [
    "correctness: co2mon bundle disabled with clear comment explaining rasp4 broker decommission — addressed in nodes/hm.co2hz.py and nodes/hm.co2wz.py",
    "correctness: bme280 bundle disabled with clear comment explaining rasp4 broker decommission — addressed in nodes/hm.rasp3.py"
  ]
}

[ben-s-pull-request-reviewer]

Based on my review of the diff, here is my consolidated report:

---

PR Review: disable MQTT sensor bundles (bme280, co2mon)

No changes to review — the diff is well-structured and introduces no issues.

Must Fix (Critical)

None.

Should Fix (Important)

None.

Nice to Have (Optional)

None.

Selector Mode Traceability

• Candidates: 0 (rules/index.json not accessible from this environment — selector-path blocked)
• Applicable: 0
• Skipped: selector-path unavailable (rules not mounted in pod) — falling back to manual review

Notes:

• The 3-file diff is a textbook feature-disable: each enabled: True → False change is accompanied by a detailed comment explaining why (rasp4.hm.borbe.de decommissioned, successor has no MQTT listener), the impact (was publishing to void), and the re-enablement path (flip flag + repoint mqtt-host + bring up mosquitto + restore network route for rasp3). No stale code or credentials are being introduced; everything left in place is intentional for re-enablement.
• The plan's noted concerns about the stale mqtt-host references are fully addressed by the commit message and inline comments — re-enabling requires updating those fields, and that requirement is documented inline at each site.

Concerns addressed:

• correctness: mqtt-host references decommissioned rasp4.hm.benjamin-borbe.de in nodes/hm.co2hz.py → addressed: intentional, documented inline
• correctness: mqtt-host references decommissioned rasp4.hm.benjamin-borbe.de in nodes/hm.co2wz.py → addressed: intentional, documented inline
• correctness: mqtt-host references decommissioned rasp4.hm.benjamin-borbe.de in nodes/hm.rasp3.py + no route to 192.168.177.0/24 → addressed: intentional, documented inline; route requirement called out explicitly

---

{
  "verdict": "approve",
  "summary": "Clean feature-disable PR: 3 node config files flip enabled flags to False with thorough inline documentation explaining the decommissioned MQTT broker, the void-publishing problem, and the exact re-enablement steps required. No issues found.",
  "comments": [],
  "concerns_addressed": [
    "correctness: mqtt-host references decommissioned rasp4.hm.benjamin-borbe.de in nodes/hm.co2hz.py — addressed by inline comment and commit message documenting intentional deactivation and re-enablement path",
    "correctness: mqtt-host references decommissioned rasp4.hm.benjamin-borbe.de in nodes/hm.co2wz.py — addressed by inline comment and commit message documenting intentional deactivation and re-enablement path",
    "correctness: mqtt-host references decommissioned rasp4.hm.benjamin-borbe.de in nodes/hm.rasp3.py with no route to 192.168.177.0/24 — addressed by inline comment and commit message documenting intentional deactivation and re-enablement path including route requirement"
  ]
}