Smart Home, Matter, Thread, Zigbee, HomeKit, Local-First Automation, Border Routers — Daily Update 2026-10-25
By Nora Thorne | myhomeio.com
October 25, 2026
Yesterday I closed with a tracking list. Today’s update is that list, answered with what I can verify in my own rack, plus the one drill every local-first stack should run once a month: the border router failover test. Short version — Thread 1.4 credential sharing is working in my house, Matter 1.7 drafts are shaping up around energy and water, and your second border router is only as good as the last time you actually let it take over.
Status Board (tracked since the 2026-10-24 update)
| Item | Status as of Oct 25 |
|---|---|
| Matter 1.7 scope | Drafting at the CSA. Energy management and water valve clusters are the two additions drawing the most vendor comment volume. Nothing to act on yet — but the water valve cluster is the first standardization of irrigation in Matter, and local-first homes are the only place it will actually be automatable without a cloud round-trip. |
| Thread 1.4 on Google/Nest | Still in Google’s internal testing window on Nest displays; no public firmware string change since last week. My Apple + HA mesh merged cleanly into one PAN; a Nest Hub in the same room remains a second, isolated mesh until that ships. |
| UniFi OS 6.0 gateway | No public gateway firmware meeting the 6.0.10 requirement as of yesterday’s check — the Network 11 Matter page remains gated. Cloud Keys and NVRs are ahead; gateways are not. |
| IKEA Liljebergge TRV | European retail January 2027, unchanged. If you want a Matter-over-Thread radiator thermostat today, the practical routes are a Shelly Gen 3 path or an existing Zigbee TRV behind a bridge. |
| First Aliro-enabled devices | Certification path open since 2 Sep; no consumer product announcements yet. Aliro is the pairing/credential layer, not the device layer — expect it to be invisible when it ships. That’s the point. |
Two of those rows are “wait for the vendor.” The one row that is in your control today — failover — gets the rest of this update.
Deep Dive: The Failover Drill
Thread border router failover is the weakest link in an otherwise solid local stack. When your primary border router reboots — power blip, firmware update, or a HomePod that decided to restart at 3am — Thread devices can sit unreachable for 30 to 90 seconds before a second router takes the mesh over. In that window, any automation reading state is reading a snapshot, not a measurement.
The fix is not firmware. It is a monthly drill that proves two things: (1) the second border router actually takes over, and (2) your automations don’t act on stale state during the handoff.
The 20-Minute Drill
- Pick a window where no automation should be firing — weekday morning, before any sunrise triggers.
- Identify your primary and secondary border routers. On HA: Add-ons → OpenThread Border Router → Log shows roles and the joined-router list. On Apple: Home app → Home icon → Thread network details. Write down which is which before you unplug anything.
- Unplug the primary. Not “restart” — unplug. A cold start is the worst case and the one that matters. Start a stopwatch.
- Watch, don’t touch. In the HA log, note the timestamp of the first device
unavailableevent and the timestamp of the first device reappearing. In my 14-device test fabric the median handoff is 42 seconds; the worst case I have measured is 81 seconds — a Thread-only sensor in the farthest room, which is exactly where you don’t want the blind spot. - Re-plug the primary. The harder question: does it take back leadership cleanly, or do you end up with two routers claiming leadership and devices ping-ponging? Thread 1.4 credential sharing means both routers should see the same fabric — if you see repeated rejoin noise, your firmware versions are mismatched. Update both, not one.
- Record it. One line in a log: date, median handoff, worst case, firmware pair. A year of these lines is your proof that the stack is resilient instead of an assumption.
The Automation That Keeps the Handoff Honest
Pair the drill with a state-reconciliation automation so the handoff window never poisons an automation. This is the pattern I run in production — entity IDs are mine; swap in yours:
alias: "Matter/Thread failover state reconciliation"
description: >-
After a border router or Matter Server restart, wait for the stack
to settle, then force a state refresh on safety-critical entities
so no automation ever acts on stale state during the handoff.
mode: single
max_exceeded: silent
triggers:
- trigger: event
event_type:
- matter_server.connected
- openthread_border_router.started
actions:
- delay:
hours: 0
minutes: 1
seconds: 30
- service: homeassistant.update_entity
data:
entity_id:
- lock.front_door
- lock.garage
- binary_sensor.garage_door
- climate.living_room
- switch.irrigation_main_valve
- service: log_entry
data:
message: "Failover reconciliation complete: critical Matter/Thread entities refreshed."
- service: notify.home_assistant
data:
title: "Mesh failover"
message: "Border router restarted; 5 critical entities re-polled and state is fresh."The delay is deliberate: 90 seconds covers the worst handoff I have measured. homeassistant.update_entity forces the integration to re-read device state rather than trusting the last cached value. And notice what the list contains — locks, garage, climate, irrigation valve — i.e., the devices where a stale state means a wrong physical action. A light being briefly stale costs you nothing; a garage door reading stale at 2am costs you the evening.
For safety-adjacent devices, go one step further and prefer command-over-state: don’t ask “is the garage door closed?” — send the close command and verify the open/close sensor confirms it within a timeout. That pattern is immune to the handoff window entirely, because it never trusts a cached answer.
Matter 1.7: Energy and Water, the Two Missing Clusters
The 1.7 drafts circulating at the Connectivity Standards Alliance add what local-first homes have been papering over with custom integrations: first-class energy management and water valve clusters. Two notes from reading the drafts:
- Energy management lands as device-side consumption reporting and tariff-aware scheduling hooks. The part that matters for a local stack: the cluster is designed to be readable without cloud mediation, so a local energy dashboard — or HA’s own energy dashboard — can drive scheduling off cluster data alone.
- Water valves are the first standardization of irrigation in Matter. If you have been running a cheap Wi-Fi valve behind a custom bridge, this is the cluster that finally makes a local-first irrigation schedule a first-class citizen instead of a hack. No shipping firmware yet — but the spec is the leading indicator, and local-first homes are where this will actually get used, because a cloud round-trip on a valve open/close is exactly the failure mode a water system can’t afford.
Nothing to act on this month. The right move is to start naming your water and energy devices in HA with stable, descriptive entity IDs now, so when cluster support lands you are swapping a binding, not rewriting a house.
HomeKit Offline: What Actually Survives a Severed WAN Link
I ran the physical test again last night — router up, WAN link down — because the list changes with every iOS and firmware update. What survived, all of it local:
- Locks: front door and garage, both Thread. Lock, unlock, state reads, and automation triggers all worked with zero internet.
- Scene execution from a physical button — the button press never left the house.
- Lighting: all Zigbee and Thread lighting — dimming, color, schedules.
- Sensors: motion, contact, temperature, occupancy — all reporting to HA over the local mesh.
- Frigate on the local NVR: camera detection and recording continued fully offline, the Coral accelerator handling the detection load.
What did not survive, and never will without a cloud: third-party app features that require a vendor backend, firmware update delivery, and Matter commissioning of a brand-new device from the phone app (commissioning over a local Thread network with a border router present can complete locally, but the app onboarding flow still wants to call home).
That list is the honest boundary. Local-first doesn’t mean “no cloud ever” — it means “your house keeps working when the cloud is gone,” and the failure modes are in onboarding and updates, not in the daily control loop.
Weekend Action Items
- Run the failover drill. Twenty minutes, one unplug, one log line. The monthly habit that turns “I think my mesh is resilient” into “I measured 42 seconds median, 81 seconds worst case.”
- Add the state-reconciliation automation from above, with your safety-critical entity list. Ten minutes of setup, one fewer class of 3am bugs.
- Verify firmware parity on both border routers. Mismatched Thread firmware is the most common cause of ping-pong rejoins, and it is the one thing a firmware update actually fixes.
- Name your water and energy devices with stable IDs while you remember to. Future Matter 1.7 support will thank you.
Next Update
Tracking: UniFi OS 6.0 gateway release (the last gate on the Network 11 Matter page), first Aliro-enabled device announcements, the Matter 1.7 draft comment period, and in-field reports of Thread 1.4 credential sharing on Google/Nest once the firmware ships. The local-first stack is past the “will it work” phase and into the “how fast does it fail over” phase — and that is a question you can answer with a stopwatch and an unplugged router.
— Nora Thorne, IoT Architect & Smart Home Systems Specialist, myhomeio.com