Learn light/switch identifies non 433 MHz devices.

[Menno]

I have some old doorbell transmitters that I would like to incorporate into Domoticz.
I click on "learn light/switch", press the button and Domoticz says this is already in use as an LED-strip.
Which is strange, because this LED-strip runs Tasmota and has no 433 MHz connectivity at all.
Then, as I was goofing around with it, I noticed that I don't even have to push the doorbell.
Just click on learn " light/switch", wait a few seconds and it gives the same message.
Usually it points to a LED-strip (Tasmota), occasionally to "new firmware watermeter available", another WiFi gizmo.
Now I can imagine if there are lots of 433 MHz transmitters around, my RFLink will pick it up, but why it gets mapped to a WiFi device is beyond me.
What gives?

[waltervl]

Do these devices communicate with Mqtt or other ways? It seems that the learn option is acting on a device updatetrigger from a gateway, RF link or another.

[Menno]

Yes, the Tasmota LED-strip works via mqtt:
https://blog.ledez.net/iot/anavi-light- ... in-a-boat/
And the watermeter is this one:
https://smartgateways.nl/product/slimme ... r-gateway/
Works via mqtt as well.

So how can this behaviour be avoided?

{EDIT} URL fixed

[Dave21w]

Are you also using the Auto Tasmota Plugin, this may be trying to add something you already have ?

[Menno]

No, I don't use any plugins.

[Menno]

If I stop Mosquitto, the behaviour persists.

[waltervl]

If I stop Mosquitto, the behaviour persists.

That is strange if it is then showing MQTT related devices as new switches/lights
Are there Lua/Blockly/dzVents scripts running and triggering these switches/lights?

[Menno]

Yes, sometimes one of the "switches" it finds relates to my watermeter, which counts pulses via a LUA script.
I put switches between quotation marks because the hardware is not an actual switch, but might for example be "Watermeter: new firmware available".
The mqtt appears to still need some work by the way:

Code: Select all

2021-08-31 17:13:54.803 Error: Mosquitto: MQTT: disconnected, restarting (rc=7)
2021-08-31 17:14:01.511 Error: Mosquitto: MQTT: Failed to start, return code: 14 (Check IP/Port)
2021-08-31 17:14:10.383 Error: Error opening url: http://192.168.178.112/control?cmd=GPIO,0,1
2021-08-31 17:14:31.547 Error: Mosquitto: MQTT: Failed to start, return code: 14 (Check IP/Port)
2021-08-31 17:15:01.581 Error: Mosquitto: MQTT: Failed to start, return code: 14 (Check IP/Port)
2021-08-31 17:19:55.147 Error: Mosquitto: MQTT: disconnected, restarting (rc=7)
2021-08-31 17:20:09.265 Error: Mosquitto: MQTT: Failed to start, return code: 14 (Check IP/Port)
2021-08-31 17:33:36.299 Error: Mosquitto: MQTT: disconnected, restarting (rc=7)

[waltervl]

It is better to disable (not delete!) the MQTT hardware in domoticz. It will stop updating the devices.

[waltervl]

I just tested the button myself (I have no RF) and I see that the button triggers on a device update. I instantly get a motion sensor popup (it is in the room i was sitting in). So you have to be fast in learning the device or stop all other device updates.

[Menno]

Yes, I was experimenting a bit too and found that if you click the button and immediately activate your device, it might pick it up.
This would appear to be a bug in Domoticz however.

[waltervl]

Why is a bug? Learning switches should not only be restricted to RTL 433 and 868 devices? The button does not say Learn RTL433 devices ...
You have 5 sec to activate the switch. If inbetween those 5 seconds another already existing RTL433 switch would be activated you would get the same issue.

[Menno]

I consider it a bug because it is called "learn new switch", yet picks up things that are clearly not switches.
It also means that if you have a Domoticz installation with lots of hardware and sensors, this function becomes very hard to use, or even impossible to use.

[waltervl]

For Domoticz everything that can be switched (lightbulb, dimmer, switch etc) is a switch. So from that point of view it is as designed.
I agree that in current time this learn switch function has limited functionality as other protocols take care of the switch creation on their own (Zwave, Python Plugins, MQTT etc). So perhaps we should consider limiting this Learn Switch function to 433 devices only. Perhaps other protocols need also a learn function. Or let the learn function neglect the existing switches.

But I am no 433 user so only venting out my personal opinion.