Skip to content

Describe the hardware tests generically #12

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
MarkParker5 merged 1 commit into from
Oct 14, 2026
Merged

Describe the hardware tests generically #12

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
17 changes: 5 additions & 12 deletions majordom_matter/readme.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,11 +38,9 @@ Two test tiers:
- **Virtual** — `tests/test_controllers/test_matter/test_matter_controller.py`, driven by
`docker-compose.matter-tests.yml`. Prebuilt **x86-64** Matter Virtual Device (MVD) binaries under
`bins/` act as fake devices, commissioned over the network (no BLE). Runs in CI on x86-64 runners.
- **Hardware** — `tests/test_controllers/test_matter/test_matter_controller_hardware.py`, on the
self-hosted `lab-pi5` runner: a real Thread bulb commissioned over **BLE → Thread** through a real
OpenThread Border Router (OTBR + SkyConnect RCP), verified physically via the IoT-cage photoresistor.
Manual only (`workflow_dispatch`). The target device is selectable in one line — `_TARGET` in the
test, or the `MATTER_HW_TARGET=ikea|nanoleaf` env var (default `ikea`); both bulbs pass.
- **Hardware** — kept in a private repository and run manually: a real Thread bulb commissioned over
**BLE → Thread** through a real OpenThread Border Router (OTBR + SkyConnect RCP), with its light
verified physically. Both an IKEA and a Nanoleaf bulb pass.

---

Expand All @@ -68,14 +66,12 @@ Two test tiers:
re-fires it → "no commissionable device discovered." Nanoleaf uses a resolvable address, so it's fine.
- **The hub does not push the Thread dataset.** `MatterController.pair_device` just calls
`commission_with_code`; the server must *already* hold the operational dataset.
- **The Arduino cage resets all relays OFF on every serial open** (DTR pulse) — bulbs lose power
unless one fd holds the port open for the whole run.
- **The "Nanoleaf RX-wedge" was a broken-OTBR artifact, not firmware.** We'd blamed Nanoleaf fw 4.1.3
for wedging its Thread RX after one `Invoke`. Once the border router was healthy (fresh image, above),
it never reproduced: 25 back-to-back Toggle+read+ping cycles with an active subscription, plus
Level/Color commands and a 90 s idle soak, all clean — `ot-ctl ping` never dropped a packet. The old
symptom was a degraded Thread mesh from the spinel-failing RCP. **Both the IKEA and the Nanoleaf are
CI-usable.** (The Nanoleaf's shared A0-photoresistor light and its slot-3 power are wired in the test.)
CI-usable.**

## What it takes to make it work

Expand All @@ -91,9 +87,6 @@ Two test tiers:
4. **Flush the BlueZ cache immediately before each commission** and do not scan afterward:
`for m in $(bluetoothctl devices | awk '{print $2}'); do bluetoothctl remove "$m"; done; systemctl restart bluetooth`.
5. **Power the device on** (IKEA opens a ~5-min pairing window on any power-on; a decommissioned
Nanoleaf re-advertises on power-on), driving the cage from a single held serial fd.
6. **Pick the target**: IKEA (`2455-383-5850`, cage slot 2) is the default CI device; switch to the
Nanoleaf (`1321-631-8363`, cage slot 3) with `MATTER_HW_TARGET=nanoleaf`. Both pass the full
discover → pair → OnOff-with-photoresistor → unpair flow.
Nanoleaf re-advertises on power-on). Both pass the full discover → pair → OnOff → unpair flow.

See `otbr-skyconnect-fix` in the assistant memory for the exact commands and recovery steps.
Loading