Currently using a Google Nest Hub Gen 2 and... it sucks. When the nest hub reboots all the matter/thread devices shared to HA become unavailable until re-added.
#Recommend me a good Thread border router
quiet otter
dense marten
Currently using a Google Nest Hub Gen 2 and... it sucks. When the nest hub reboo...
Hi Brandy, this is definitely not normal, this should be a well tested scenario and these hubs get rebooted in multiadmin environment all the time
Can you elaborate on what you see a bit more? When you reboot the hub, do these devices appear online at Google and offline in HA? Offline everywhere? What are the other Thread devices you have in use?
If you swipe up on the screen and select ⚙️ -> Device Inoformation -> Technical Information? and note Fuchsia version?
quiet otter
They all come back Google side (after a short period, I believe this is due to the sleep characteristics of thread devices). But they never come back HA side, remaining unavailable. The thread devices I have so far are an Aqara P2 door sensor and 4x Zemismart thread blind motors
dense marten
thank you, so that rules out Thread network being down
Were you able to see the fuchia version perchance?
quiet otter
Will have to grab that in a few hours I think, doesn't appear to be exposed in the Google home app. Thanks for helping!
dense marten
np, that's okay, unless it's somehow very old, it's not super critical
so the other suspicion is SRP server issues, but it would be odd if the google records are readily available, but HA ones are not
to do some sanity testing, if you download the Service Browser app (I know there is a version on Android, probably also on ios)
With that app, there is going to be a record for _matter._tcp, if you click on that, you should see all devices records in all ecosystems
format is fabricid-nodeid as an alphanumeric sequence. All google ones will have the same fabric id, all HA ones will have another one.
So you should see ~12 records 1 for HA, one for nest hub itself, and 5x each for your devices * 2 ecosystems
if you only see around half, it would suggest that somehow SRP records got lost
With that info in hand, we can see what to look for next
quiet otter
Will do this soon as I'm home, that should be a great start. I'm a huge believer in matter/thread and hopefully all these little things can be worked through to make it accessible for everyone
dense marten
💯
I'm about to go offline for the night in my neck of the woods, but I'll check in tomorrow and we'll go from there, I'm confident we'll sort it out
quiet otter
if you only see around half, it would suggest that somehow SRP records got lost
Well, it looks very interesting... and this is a nice app for troubleshooting!
The current status of the thread devices:
1x HA server, 1x Nest hub
Added to Google Home app - 1x Door Sensor, 2x Blind Motors (assuming these fabric IDs end in CC as the hub is B7.local) 2 Blinds seem to have dropped off Google Home, greyed out now.
Allocating new controller with CaIndex: 1, FabricId: 0x0000000000000002, NodeId: 0x000000000001B669, CatTags: [] - this is in the matter server log. So the top two records are the HA matter server (F0000.local), and the blind I shared back to HA from Google Home to test.
Not sure what the middle fabricid is.
quiet otter
If you swipe up on the screen and select ⚙️ -> Device Inoformation -> Technical ...
Seems current according to: https://support.google.com/googlenest/answer/7365257?hl=en-AU
Preferred thread network in HA is set to the Google nest hub one
dense marten
So at least initially, it doesn't look like there are any obvious Border Router issues. Given the screenshot above it seems all the devices advertisements are coming through
the other fabric might be a local one from the phone, so it's fine
So steady state now, what do you have in both ecosystems? All of the devices again?
Don't remove them and recommission them just yet, I want to see what happens when we reboot certain things
quiet otter
Two of the blinds are still unavailable Google home side, but there's two available still plus the Aqara door sensor
one blind recommissioned via share into HA is still there as my original test
current state
Top two are HA fabric, so the server and the test blind motor reshare
Bottom 5 are the Google Nest fabric, all blinds, hub, door sensor
dense marten
And everything was powered on for a while now?
Next step, if you click into those, they'll show you IPv6 addresses for those devices, can you ping that ipv6 address from the console?
Also, what router / switch combo do you have. Sometimes issues manifest at that level
What you are describing seems like for whatever reason, hubs are not able to sucessfully establish sessions to the end devices reliably, so we should understand why
quiet otter
Also, what router / switch combo do you have. Sometimes issues manifest at that ...
It's a rack mount ubiquiti dream machine pro se
quiet otter
Next step, if you click into those, they'll show you IPv6 addresses for those de...
From HA core console?
dense marten
From HA core console?
Sorry, not the HA console, I meant the terminal in the OS, in the service browser app, if you click on any one of the device records, it'll show you the published IP address from that device, then you can use the ping6 command with that address
dense marten
It's a rack mount ubiquiti dream machine pro se
And as for ubiquity, I was afraid you'd say this. 🙂 I use Ubiquity WiFi AP's and older gigabit switches with an ISP router, which work well, but their routers can sometimes cause trouble depending on what settings are enabled.
I don't know what they have now in terms of settings, but often their optimizations do more harm than good, so that's something to consider looking into, but let's not go there yet.
Generally, as cliche as it sounds, if would be interesting to understand what happens when you reboot various devices.
There is reconnection logic that looks for devices if they haven't checked in for a while, generally it's around 30s for most ecosystems, so if a device that was online before drops off, in about 30 seconds, hubs will go looking for it again on the network.