deCONZ - Official thread

Configuration Zigbee
1192

I am facing a weird problem with conbee 1. I moved from RPi to Intel NUC and i have deconz in docker. Now the network is not stable and also i see in my deconz multiple entries for the same light but only one works.

I have 3 in wall (with neutral) switches (2 double and 1 single). When i;m trying for example to add the switch to the network i’m going at phoscon - add new lights and i’m pressing for few seconds and then it starts to pair and i see something like in the photo.

[Album] Imgur

The light “main” and the smart plug “kitchen” are the buttons from the double in wall switch.
Bathroom is the single switch which also created on/off light 5, 6 etc.

Smart plug “Door” and “smart plug 10” are the buttons from the second in wall switch. I can add “Door” to a group but not the “smart plug 10”.

In general i cannot find a way to fix it so when i add a light it will add only one light and not multiple and also to be able to assign them to a group. Any help? Update to latest firmware recently,deconz latest version and HA as well… Thanks in advance,

1193

OK so I updated to latest hassio, updated immediately to latest deconz, then immediately rebooted. Everything fine and ingress working fine too, but like others I can’t access phoscon via the IP address and port. I can access ok via ingress so it’s not urgent but will investigate further at a later date.

1194

Is it possible now make deconz API available outside docker? I need to add deconz to another instance.

1195

A list of supported devices can be found here:

Others may work to

1196

Silly question to all the above guys not getting the api.
Did you unlock it.
In the deconz app gateway config button authenticate
It then goes blue for 60 secs
Then you try and connect

1197

Yeah i can’t speak for everyone having api problems but i did the ”unlock” procedure. Has anybody gotten api access to work outside the hassio instance deconz is running on since 5.0?

Edit nevermind more people like me reporting the same problem when running two instances of Hassio. Hope this will be fixed and that the option will come back to manually input deconz ip and port in the integration flow

1198

"linux runtime spec devices: error gathering device information while adding custom device "/dev/bus/usb": no such file or directory"

20-01-10 15:55:39 ERROR (SyncWorker_9) [hassio.docker] 
Can't start addon_core_deconz: 500 Server Error: Internal Server Error 
("linux runtime spec devices: error gathering device 
information while adding custom device "/dev/bus/usb": no 
such file or directory")
  • RPi4 4GB / hass.io supervisor 193 / HassOS 3.7 / RaspBee HAT.
  • Everything worked perfectly before crash
  • Reflashed SD, reboot, restored last working snapshot.

Tried deleting, rebooting, redownloading, restarting, everything … everything else appears okay.
Error as above when trying to start add-on.

Addon config as before:

{
  "device": "/dev/ttyS0",
  "vnc_password": "...redacted..."
}

What am I missing? Low WAF at the moment. Can’t turn on or off lights! :]

1199

If bridge discovery fails integration flow will fall back to manual input

1200

What showing in the
hassio / system / hardware

1201

Rejoice guys! I found working solution how to connect several Hass.io instances to one deconz:

  1. Don’t use Hass.io add-on Deconz, since it is not visible outside the Hass.io. Instead, install https://github.com/marthoc/docker-deconz if you use docker or official deconz distributive otherwise.
  2. Use ‘host’ network mode for docker-deconz container. Set DECONZ_WEB_PORT = 40850 and better DECONZ_UPNP = 0 for environment variables.
  3. Remove existing deconz integration on your Hass.io instances.
  4. Manually add deconz integration for every Hass.io instance with usual procedure (with prior authentification in phoscon/gateway/advanced).
  5. Enjoy!
1 Like
1202

Does not seem to work have tried putting in

deconz:
  host: ip adress if my RPi
  port: 40850

Into configuration.yaml but no success :slightly_frowning_face:

1203

I am using the Phoscon App with a Conbee II through the Home Assistant/Hassio addon. The device is connected to /dev/ttyACM0 port and works fine.

I am running the 2.05.72 with the firmware 26420700.
In the log I can see:

11:25:30:068 GW firmware version: 0x26420700
11:25:30:068 GW firmware version shall be updated to: 0x26490700

In the UI I can run the update and it ends successfully.
However when I open again the UI the firmware is still 26420700 and it suggests again the update to 26490700 .

Someone suggested me that the Firmware updates from the web UI will fail silently and to update manually following the instructions reported here.
I am not sure that I can do it using the hassio-addon.

1204

Through the config flow

1205

Here is a guide for updating the firmware with the hass.io addon.

1 Like
1206

Hi, what do you mean with this?

I also split everything in 2 machines

  1. HASSIO on NUC with all the automations scpts and so on
  2. HASSIO on a pi3 with SSD, with very few things, but since deCONZ was not working (not beeing discovered by the other HASSIO-NUC installation) I put deCONZ under raspbian in the same pi3.

Now I wish to move deCONZ to HASSIO on the pi3, and of course I need this to be visible (via HASSIO integration), on my main HASSIO installation.

And it would be nice to be able to see the MESH.

Is it now possible, working?

P.S. the deCONZ integration is visible in the HASSIO-pi3, but is not visible in the HASSIO-NUC (maybe because already added as integration …?)

1207

I had the same issue and got it fixed. Short version:

Reset the network config in the add-on, put the port 40850 back again as you need (and others if needed) and restart the add-on.

Long version:
I found the following link:


where frenck mentioned the solution. In some update of the deConz Add-on the internal port got changed from port 80 to port 40850.

Before the network config reset to defaults:
041e4ff74c50 homeassistant/armhf-addon-deconz:5.1 "/run.sh" 17 minutes ago Up 10 minutes 0.0.0.0:40850->80/tcp, 0.0.0.0:5901->5900/tcp, 0.0.0.0:40860->8080/tcp addon_core_deconz
After resetting the network config to defaults and putting back port 40850:
eb9b309ea867 homeassistant/armhf-addon-deconz:5.1 "/run.sh" 12 seconds ago Up 8 seconds 0.0.0.0:40850->40850/tcp, 0.0.0.0:5901->5900/tcp addon_core_deconz
(I skipped the websocket port as I don’t think I’ll need it).
One more thing. I changed my Pi not to use IPV6 as netstat -a showed that port 40850 was using tpc6 but that didn’t fix the issue (the problem was in the docker port forwarding) and I’ll revert that.
HtH

1208

I think something is broken in that case, running on 103.6.

It seems like more people are having this issue after 103.3/103.4 (manual input option not showing in the config flow)

It just says ”Couldn’t get an API key” and then nothing more

1209

Hi again, I bought two single and double, unfortunately the single could not be added to Deconz. The switch reports as two and has entities on / off, is visible in HA as lights and also added to GH as two lights. They work in all possible combinations, we turn on the action-turn off the button and vice versa. The switch has a slight click and a slight key stroke.

1210

Well that is probably because the discovery works :slight_smile:

What you could do is to disable the add-on and then start the config flow and discovery should fail and prompt you with manual input, then start deconz again and unlock the api.

Or you can wait for 104 which has few improvements that might be what you’re missing.

1211

Thanks :blush::pray: Will try that!