Connecting a second OTBR

Configuration Matter/Thread
1

Hi!

I have a HA running as a VM inside a Truenas server. It has a usb ZBT-2 thread radio, but I have a large house and ~70 thread devices, and some of these would currently need over 5 hops to reach the ZBT-2 border router. I am also seeing a lot of timeout errors when attempting OTA updates, and even just devices dropping out from home assistant.

I am therefore thinking about adding another (or maybe several) border routers, and I bought a couple of SLZB-06MG26 for this purpose. I am now trying to get HA to work with both the local ZBT-2 and the “remote” (ethernet) SLZB06. Note that it’s currently running firmware 3.2.4 (the latest “stable” version so far).

The ideal solution would be to have another OTBR add-on to handle the thread-over-ethernet SLZB, but that’s apparently not an option (yet?) so I instead ran a docker container (“custom app” in Truenas terminology) based on GitHub - ownbee/hass-otbr-docker: Stand-alone Home Assistant OpenThread Border Router docker container. · GitHub. This seems to be working, and I was able to add the OTBR integration from HA to connect to this.

At this point I have 2 OTBR integration services: one relying on the local OTBR add-on (which talks to the ZBT-2), and one that relies on the docker container that talks through ethernet to the SLZB06.

It sounds like it should be all I need, but if I go in the “Thread” page, I now have only one thread border router listed.

Here are the screenshots of the OTBR integration (with the 2 entries) and the Thread page (with only one TBR):

screenshots
screenshots658Ă—628 42.1 KB

(Screenshots combined because I can only include a single image…)

I restarted home assistant, with no change.

I can access the docker container’s web UI, where I see valid-looking entries in the “Status” page (in particular, I see the same IPV6 prefixes as the ones from the OTBR add-on). Though the “topology” page never seems to load in the docker container - it loads fine for the OTBR add-on. :person_shrugging:

Maybe I missed something obvious? Is anyone able to use multiple border routers reliably?

(I’ll try a GL-inet S200 next, I think it runs its own OTBR so it might be slightly easier to setup.)

EDIT: I have tried another device, a SLZB-06Mg26U. It’s very similar, but on the latest firmware, it has the ability to run OTBR on-device, instead of using serial-over-ethernet. Indeed, I just had to add a third OTBR integration entry, and that one is now showing in my Thread integration page. Not sure why the one that relied on a docker-contained OTBR server did not appear.

So things look better with this on-device OTBR so far. Things are still quite broken (trying to adopt a wired thread device sitting right next to my ZBT-2 runs into “Something went wrong” half the time), but

2

Interesting… this is the first I have heard of running OTBR on the SLZB itself. Do you have a pointer to this (I can’t seem to find it on the smlight website)? Is this essentially a SMHub Nano?

Is the SLZB based OTBR also “preferred”?

3

You might not need a second TBR. Have you tried the beta matter.js server? I have 102 matter over Thread devices and with the beta server things run smoothly, even with only one router.

4

I saw the ability mentioned on Thread setup (network ... | SMLIGHT Manuals, and indeed on my SLZB-06Mg26U, the latest 3.2.6.dev0 beta adds this option. Not that it’s working great so far… Before adding it, I’d say my system was working ~70% - everything was sometimes connecting, but sometimes disconnecting (and mostly timing out during OTA). With the extra OTBR, things are basically not working at all. It’s very possible the OTBR support in this SMLight device is not complete. In particular the “ipv6” option appears turned on, but it doesn’t show any ipv6 configured, and my unifi router doesn’t see any ipv6 associated (from ipv6 announcements).

At first the SLZB TBR was in its own thread network, but I moved it to “preferred” and they’re now together in the preferred network.

5

I’m sure there are cases where a single TBR is sufficient, but maybe we can agree that there are cases where that’s not enough?

I am currently running the beta matter.js server. But it’s ~40 meters between my HA computer and the other corner of the house, and I’d rather keep the routes to the nearest TBR shorter. I already noticed a much higher success rate of commissioning near the zbt-2 vs a few rooms away.

6

My point wasn’t to suggest you should have one TBR (I have six) but that you might have better luck with the beta matter server if you do.

7

I agree, things were indeed much worse before switching to the beta/matter.js server.

1 Like