Matter devices are unavailable

[Rammstein]

Update - all Devices are Off during the night after 4 days without any problems, the complete Thread network was down (16 Devices) just the only one matter device was running!

Well after restart from homey all devices are immediately back, it takes not so long as when homey lost one ore two devices and must create the thread network new…seems for me this fault was other then they before? An answer can only give us Athom but the do not answer since over one week, i send them all informations and logs, screenshots and descriptions of the problems but not response till now! Right now i send them again all informations…we will see!

[deejayreissue]

Change your WiFi channel to just 1, not 1-6-11

11 will conflict with Zigbee

Also, keep an eye on your matter dashboard in Homey Developer Tools

If the IPv6 addresses keep vanishing then reappearing from the far right column, then its an IPv6 issue with Unifi

It could also be your ISP router thats the issue, it might not be passing IPv6 to your Unifi network. Do you have a separate ISP router above the Unifi network stack?

My Thread issues were all down to having an EE Router above everything else.

Once I removed the EE router entirely from my network and entered the PPPoE details for EE into my Flint2 router, every Thread issue vanished.

I have 50 Thread devices all working perfectly. Not a single outage when I upgraded to Homey Pro v13.5

[Rammstein]

Hey, thx for your suggestions! Here are my answers:

1. I have now changed the channel from the 2,4Ghz band only 1 and will monitor it

2. I will not 100% exclude that it is not a problem with my unifi network but i dont think so because there are so many facts which shows in direction of homey, for example, i run these all devices with apple Home or Home Assistant and they works over 2 years without problems, of course i have some restarts but you can count his on one hand…further i cant explain why it work somtimes for weeks without problems and other only a few days, or why you can solve this kind of problem with simple restart of homey, why homey loose at the last breakdown ALL Devices at once…

3. Yes i have an ISP Router but he is in “Bridge Mode”, only internet signal passthrough to the unifi Master, here short the network hardware, ISP Router in Bridge Mode ->UDR is Master->Unifi 16 POE Switch->4 x Unifi UAP-AC-Pro

4. I dont know if you have seen it but my config for Unify IPV6 is approx. this settings, sry its a german description : UniFi Netzwerk richtig für Matter Konfigurieren - Tipps - Allgemein - simon42 Community

5. Here a Picture from the Network Devices, i have done a lot of changes on my network but at all it always repeats the problem on Homey side, not on Apple side or Eve side…at the past i run the Apple thread network parallel to homey and the devices are working there…currently i have only one thread network running, i have delete all on Apple-Eve side to exclude all influences…

   

[Rammstein]

Thx for any further input in this topic!!!

[Mark1541]

What if you have multiple AP

Like ubiquity or omada they have different bands channels

[Rammstein]

All Unifi Devices are in the same band and channel…i only use right now 2,4Ghz, the 5Ghz is dissabled at the moment …

[Thyg0d]

Do you have access to Claude?
I do and had it access and assess both Homey and Ubiquiti network.

Ended up buying an ethernet adapter for the homey and it solved the issue, from the wireshark traffic it seemed to be an issue with homey not waiting long enough for the matter devices to communicate och just said “never mind, you’re dead, and you’re dead.. everyone’s dead”
Oprah style

[Rammstein]

What do you mean with Claude? sry i do not understand what you mean?

I have since the beginning installed the ethernet adapter, i also change the original power supply & cable to PD_Thunderbolt devices…

[Rammstein]

ah you mean the KI App “Claude” for Homey…no i never heard from it, i will test it today - thx for your suggestion!

[Thyg0d]

You don’t need the Claude app in Homey, it can use SSH for example.

But Claude is from Anthropic, like Chat-GPT, Gemini or Co-pilot.
An AI, but Claude is a different beast completely.

[ulrikb]

Hi Pascal,

You’re right on the terminology — Matter is the application standard, Thread/Wi-Fi is the transport. Worth keeping clean.

Where I’ve landed: I haven’t used my HP23 in over two months. I moved my Matter setup to Home Assistant on the MR3 mentioned earlier, with Matter and the OTBR in their own container next to HA (same VLAN). Very stable now — at times more stable than my Zigbee. Bonus: firmware updates on IKEA devices without needing the IKEA hub.

To be clear, I’m not suggesting moving to Home Assistant as a fix for the Homey issue — it isn’t one. But after being in dialog with Athom and getting the sense they didn’t see it as something they could or should act on, I had to try a different path to see how it could actually work for me. The idea of an all-in-one device is genuinely great, and I’d love to see Homey deliver on it.

So the standard itself was never the issue. What’s left is the question of whether the instability on Homey Pro comes from the shared 2-in-1 radio (Zigbee + Thread on the same chip), the product implementation, or both. My experience points mostly at the product side.

A proper test would be if Athom let the community try a self-hosted Homey with a dedicated Thread radio (TCP) natively for Matter over Thread — that would actually answer the question. I floated it here but no response so far.

I hope that I some time could come back for another try.

[Pascal_Nohl]

@ulrikb

I’m not suggesting moving to Home Assistant as a fix for the Homey issue

But in some cases, it could actually become a valid option.

There are moments when I get really frustrated with Athom. I hope the latest update fixed the Matter issues, as stated, but I still feel that Thread remains very unstable. And I suspect this may be related to a situation many of us experience without necessarily realizing it.

I have migrated almost all of my Matter devices to Homey’s native Thread network, except for eight roller shutters. Those are still in Apple Home and shared with Homey via Multi-Admin.

A few days ago, around 30% of my devices suddenly became unreachable. Several Homey reboots did not help, including some with a powerless phase of about 20 minutes. Then I had an idea: I unplugged all my Apple Thread border routers — Apple TV, HomePod and HomePod mini. After that, my devices came back.

After about 30 minutes, I started plugging the Apple border routers back in one by one, with intervals of around 20 minutes. At the moment, the system seems to be running again. What I find strange is that all my roller shutters continued to work even without a single Apple border router being online.

To me, it looks as if the Homey and Apple Thread networks are merged into one large Thread network. That is, of course, how Thread is designed to work. But in practice, it seems that, for some unknown reason, two border routers may sometimes behave as if they are both in control and somehow disturb the network. Once one of them is unplugged, the network seems to heal itself.

What frustrates me most, however, is Athom’s communication. My support ticket has been open for several weeks without any real follow-up. My latest app is waiting for certification, and my messages remain unanswered. If this is considered customer service, then it becomes very difficult to keep confidence in the company.

[smarthomesven]

@Pascal_Nohl

My latest app is waiting for certification, and my messages remain unanswered

For how long was it in certification? With 24 apps and counting, I’ve never seen any app take longer than 1-3 working days to get certified, usually published the next working day after submitting.

Does your app require the homey:manager:api permission? Athom states that those can take longer.

[Pascal_Nohl]

@smarthomesven

For how long was it in certification?

Certification is pending since 16 Mai.
A reviewer told me that it is to similar to another app and that I should contact the other developer but can also explain them why I think they should certificate my app instead. I replied and after a week I’ve send a second answer. But until now, radio silence. I’m completely ignored.

You know how much work it is to develop a good app that you have to bring to a publishing level before you get told that they do not value your work and do not want to let the user the choice of the app he wants to use. I can also invest my efforts in other platforms where developers are valued.

It does not use an API

[robertklep]

@Pascal_Nohl

You know how much work it is to develop a good app

And yet, AI-slop gets certified within proverbial minutes

[smarthomesven]

I think you should retract your submission and immediately submit again afterwards (just click “Retract submission” and then click the “Submit for certification” button), and then you can enter your comment again as well (it doesn’t have to be reviewed to be able to enter a comment, you can just type or paste it and click Send). Maybe something went wrong with the certification so you should try to submit again. It never takes 12 days for an app to get certified