Matter Devices Unavailable After VM Reboot

[inland lagoon]

My matter devices were online and operational for over a week, however after I rebooted my NAS, and consequently the home assistant VM, all of my matter devices are unavailable. I am able to ping some of them, however I get a red !
Is this a known issue? Has it happened to anyone before?

[noble sierra]

Are these devices wifi or thread?

There might be more information available in the logs of the Matter Add-On.

The usual reason for Matter devices being unavailable is that HA cannot look up the devices via mDNS, but there's also other things that can go wrong, like IPv6 routing issues.

[inland lagoon]

They're thread

After the matter and haos updates through the devices started working again

[outer dagger]

If this happens again, I've found that this happens sometimes to my Thread devices. The best way I've found to get them back is to restart the Thread add-on. I used to need to do this every time HA restarted, but recently it's only sometimes and only a few devices when it does.

[inland lagoon]

it happened again, but this time without a reboot or power failure or anything
I restarted the matter server addon however the devices didn't reconnect
when I try to ping the device none of them have IP addresses

Maybe this is a networking issue @noble sierra can assist with? my network stack is all ubiquiti hardware

[noble sierra]

Ubiquity hardware has often been problematic especially if vlans are involved. Their mDNS relay doesn't seem to work well

[outer dagger]

Thread is a different Add-On than Matter, restarting Matter Server makes it rediscover everything again, whereas restarting Thread just makes Matter think they aren't available for a touch. If it's happening during active use though, you have a more concerning issue than mine.

[inland lagoon]

everything is on the main vlan

[inland lagoon]

the thread addon isn't configured, should I reload it anyway?

I'm not sure how to configure it, this is the configuration screen but there aren't any options

either way reloading it doesn't bring the matter devices online

[reef isle]

You do not have a preferred (thread) network.

This is how looks here:

Do you see any button to make the NESTPAN thread network your preferred network? Did you have any other Thread Border Router in the past that you removed lately?

I am unsure about your situation. I configured my preferred Thread network in the end of last year, if I remember right. Too long ago… 😅

[inland lagoon]

no that nest hub is my only tbr, and I've never removed it

I also don't see any where to set it as preferred

[whole osprey]

If I recall correctly, you can't set a Thread network to preferred without having the Thread dataset/credentials in HA

[inland lagoon]

I can do this but it doesn't look like that's correct

[inland lagoon]

so should I be adding a dataset from TLV? my only other option is an openthread border router

[whole osprey]

I guess to back up a second, if the Nest Pan hub is your only TBR and it has been working for you, then you don't really need to deal with HA's Thread.

HA's Matter Server doesn't need HA's Thread/OTBR as it is using the Nest Pan hub

[inland lagoon]

apparently you have to do it in the android app

[whole osprey]

yes that the way to do it

[inland lagoon]

but from what you're saying that's not why my devices disconnected

[whole osprey]

Since you had this all running before, then it is more likely a Nest Pan Hub issue, but really hard to say. Is it all devices at one time or one or two devices being disconnected?

[inland lagoon]

it's all devices

I tried deleting & readding and them and the provisioning fails during "Connecting Device to Home Assistant"

[reef isle]

Did you have any changes in your network? Which ubiquiti equipment, which firmware versions, which settings?

[inland lagoon]

could have been a unifi auto update I suppose, but I didn't change anything in my network settings to begin with

my devices & versions are

• UDM SE: v4.0.21
• USW Pro 48 PoE: 7.1.26
• U6 LR: v6.6.78

@reef isle

the issue in the provisioning process at least is the matter device connecting to home assistant

home assistant itself is running in a vm with a passed through USB nic plugged into the switch

reinstalling the matter addon did allow me to successfully add a device though

[reef isle]

Not directly comparable to my setup:

udm-se: 4.0.21

• network application: 8.6.9
  u-lte-backup-pro: 7.0.70
  switches:
• 1x usw-pro-aggregation: 7.1.26
• 1x usw-pro-max-48-poe: 7.1.26
• 1x usw-pro-max-16-poe: 7.1.26
  access points:
• 2x u7-pro-max: 7.0.84
• 4x u7-pro: 7.0.90

There were a lot of changes in the network application, switch and ap firmware regarding mDNS. But I can’t say, if they also solved the issues with ap firmware 6.x.

[inland lagoon]

the issue may have been related to an invalid recorder definition that was an artifact from an old import
I've fixed that issue and I'll see if the switches remain stable

[inland lagoon]

all the devices have become unavailable again

for some reason the boarder router isn't found

[reef isle]

How good is the WiFi connection of your Google Nest Hub?

[sacred wasp]

I have "No border routers found" too, although thread devices work. It used to show 3 routers - Nest Hub Max, Nest wifi router and point.

[noble sierra]

seeing "no border routers found" there probably means that home assistant isn't able to resolve the mdns services from the thread border routers (they should be publishing a service named _meshcop._udp)

[inland lagoon]

It should be quite good, it's got line of sight to the wap

[inland lagoon]

Are there troubleshooting steps to confirm this?

[noble sierra]

I'm not sure how to do that from HAOS, unfortunately. If you have another linux machine on the same network, you can try running avahi-browse -r -t _meshcop._udp

[reef isle]

Do you have any ping timeouts between your machine and your Google Nest Hub? Does the Google Nest Hub work without any other problems? Line of sight doesn’t mean that it’s properly connected and also doesn’t say anything about wireless interferences. 😉

Your HA machine and your TBR are in the same VLAN?

[inland lagoon]

yea HA & the nest are on the same network & vlan

I can try pinging

[inland lagoon]

connection between them looks fine

[reef isle]

Ok, can you do 1000 or more pings without timeout? Just to be sure… However… Looks like there must be another issue.

Did you try the avahi-browse command suggested by kepstin?

You have to find out, if your mDNS works as expected.

[inland lagoon]

I tried running this, didn't return anything

I do have mDNS enabled on that network, but for some reason my HAOS can't see my TBR

[clear pebble]

Disable that mDNS checkbox on Unifi as that enables a mdns forwarder which is bad.

[inland lagoon]

can you elaborate?

[clear pebble]

mdns forwarders in general are bad but the unifi one is just complete garbage. We've seen wireshark dumps where packets were malformed and/or duplicated like crazy. Let alone the fact that IPv6 mdns forwarding does not exist. Just keep IoT devices in one single (IoT) vLAN and do not attempt to route it.

[inland lagoon]

I tried to put them in an IOT vlan but I couldn't get the matter devices working unless everything was on the main vlan

[clear pebble]

Exactly - either put EVERYTHING in the IoT lan (so devices and BR's and HA) or just use main vlan for BR/matter + HA

[inland lagoon]

I put everything in the iot vlan, it didn't work

[clear pebble]

If you insist on having multiple vlans, then moving HA into the IoT vlan is the best option

[inland lagoon]

I had HA in the iot vlan

[clear pebble]

But in general just a flat network save syou from a lot of headache

[clear pebble]

And what about any border routers ?

[inland lagoon]

had it in the iot vlan as well

[clear pebble]

In any case, do not enable that mdns forwarder in Unifi

[clear pebble]

If both all the border routers and HA is in the IoT vlan then it should work. Be careful to also put your phone on that wifi ssid if you are commissioning any devices

But like I said before, better just flatten the network

[inland lagoon]

I had my phone on that vlan anyway, I don't like having flat networks but it wouldn't function any other way

[clear pebble]

If you dont like the flat network, Matter is not for you

[inland lagoon]

you said it should work if they're all on the same vlan, that seems contrary to "if you don't like flat it's not to you"

[clear pebble]

flat network = all in the same (v)LAN - it will not work if you have multiple vlans it will even stop working if you have everything on the same vlan and enable a mdns resonder (because that corrupts the multicast packets for matter)

Also, if you have (semi)professional network gear like Unifi or Omada, ensure to disable any "multicast enhancements" options - that setting only makes sense in a high density enterprise environment but it cripples domestic protocols that rely on multicast

[noble sierra]

note that if you're using android, you need to have your phone on wifi on the same vlan as the TBR, and your phone needs to have internet access for google play services to work.

lots of things that can go wrong with complex network setups :(

[inland lagoon]

I was on android, the wifi the phone was on was the same vlan as HAOS and the tbr

also the matter devices are now offline again, though this time I can still see the TBR in the thread integration

[inland lagoon]

I'm going to change the vm from nic passthrough to usb passthrough and see if that makes a difference