Some of my Matter-over-Thread devices became unavailable after a while — always the same ones. A few of them are Thread routers, and I suspect that once one of those routers dropped, some nearby end devices followed.
After rebooting Homey everything usually came back, but once I had to restore Homey from a backup.
I started making a few changes to test their impact. Things look stable now, but it’s still too early to call it “fixed”. I also can’t say which change (or which combination) actually made the difference. For reference, here’s what I changed:
Added a handful of Thread router smart plugs so each battery device has at least two routers nearby, and placed one router plug close to Homey as well.
Reduced Wi-Fi transmit power on my access points by one step (from Powerful to Good).
Forced 2.4 GHz channel width to 20 MHz (instead of 40 MHz).
Disabled beamforming on 2.4 GHz.
Moved Homey farther away from my nearest access point (now ~1 m; previously ~50 cm).
Worth noting: my Thread network under Apple Home was rock solid without any of these changes. That makes me think Homey’s Thread radio might be more sensitive to interference (or simply weaker) in my environment.
Next step for me is to migrate to a more “pro” network setup (more control, proper segmentation/VLANs for IoT, cameras, etc.).
Just a small update after a longer period — I don’t have a lot of new findings to share, but happy to pass on some of the experiences I’ve had along the way.
Since I stopped auto-updating and started keeping a closer eye on what’s in each update before applying it, I feel more in control of my setup. I shared some thoughts on that in this comment in the podcast thread.
It’s a bit unfortunate to see these challenges showing up primarily on Homey — both Athom and IKEA deserve better, and I’m sure they’re working on it. Matter has so much potential, and getting it rock solid would be a win for everyone.
One thought I’ve been playing with: since Zigbee and Thread share the same radio on Homey Pro, I’m curious whether offloading Zigbee to a separate coordinator via Zigbee2MQTT would free up the radio enough to make Matter/Thread more stable. Just a theory at this point — but might be worth exploring.
This morning some of my devices were unavailable again. The strange thing ist that one of the devices is a Wifi device that isn’t Matter and has an app to work. So for this device, no Thread and no Matter problem can be the cuplrit.
And in my logs I’ve found a strange Message (I log errors and the error message the card throws)
Matter is not available ???. This is definitively a Homey Problem and has nothing to do with Wifi interferences.
When I restart my Homey, all devices come back within some minutes, and actually I have to do that every morning, as every morning some devices are missing and my heating did not start
Also, comparing the logs to backups … the first time a card throws an error ist always the first action after the backup. Even worth, I had to send my Homey back to Amazon as the USB connector had some problems. The devices that drop off now, were not restored on the new device by the backup and were completely missing. Coincidence ?
Thank you for the hint. I already activated the the Power User experiment and I also created a flow to restart my Homey by clicking virtual button. I did not automate it until now as I wanted to collect an error log.
Today I’ve created an official ticket for the support. My ticket is very precise and I attached the log I created. It is friendly and formal, but also makes very clear that this is not acceptable for users for a 400 bucks device that officially states to support Matter over both transport protocols. If Matter is not stable they should sell it as Matter compatible (beta) and I never had bought the device.
Today, after I have the logs and I created the ticket I created the following flow that I hope will work as expected. The plug is one of those that are always concerned.
