Can't connect ZWave-JS-UI (remote) on RPi to HA Yellow

I have a Nortek HUSBZB-1 stick on my HA Yellow and a couple of devices in the house. I updated to ZUI some days back and all is generally good.

I added a ZNET in a remote shed, connected on the SAME subnet as HA and then added an RPi4 on which I (finally!) managed to get ZUI installed and running. The ZUI instance on the RPi shows the ZNET controller, but not the Flex Sensor device attached.

The two do not appear to talk, so I assume something is wrong in where either ZUI or HA is looking/pointing.

HA base setup:

  • only the Nortek and 2 locally installed devices show up
  • ZUI shows the three devices above, plus a deleted one I can’t seem to get rid of, even though deleted in the integration
  • all keys are entered, and Soft Reset is off
  • Home Assistant section has WS Server on, no server host and Port 3000, with DNS discovery on.
  • The map shows the controller, two connected nodes and one greyed out (the deleted one)
  • ZWave integration shows ZUI under “Hubs”, with the Server URL as ws://a0d7b954-zwavejs2mqtt:3000
  • MQTT is NOT selected

RPi setup:

  • ZUI Serial port is 192.168.yyy.22:2001 (this is the confirmed address of the ZNET and the RPi can ping it successfully)
  • all keys are the same as in HA ZUI
  • WS Server Host is set to the address of HA: 192.168.yyy.210, port 3000 (is this even vaguely right?)

Clearly I’m doing something wrong. The device is clearly working in the shed as when it’s exposed to light it beeps.

HA talks to Z-Wave via the web socket server. So you need to add a new integration and point it to the Z-Wave JS UI that is connnected to the Z-NET.

No, this is completely wrong. You should leave it blank and connect HA to the server, which is the IP address of your RPI.

These keys should come from the Z-NET, if this was an existing network, if not new ones are OK. Ideally they should be unique per network for security purposes.

I don’t really understand what to do here; sorry…I’m totally lost with this I think.

So you need to add a new integration and point it to the Z-Wave JS UI that is connnected to the Z-NET.
When you say “add a new integration”…I already have ZUI on the HA…I can’t add a second one. Do I add the “old” ZWave JS?

No, this is completely wrong. You should leave it blank and connect HA to the server, which is the IP address of your RPI.
I’ve blanked this, but is the “connect HA to … your RPi” the above?

I’m not getting how these devices are supposed to connect at all.

ZUI is an add-on, not integration. https://www.home-assistant.io/docs/glossary/

Z-Wave is the integration.

Have you looked at the Z-Wave docs or any of the others?

At Z-Wave - Home Assistant there is a diagram:

The integration (which is Home Assistant) connects over the websocket (network connection) to the remove Z-Wave JS Server (ZUI), which is your RPI.

Or, you can see the HomeSeer docs: Z-NET Integration with Home Assistant. In step 9 you use your RPI’s hostname/IP address instead of the add-on.

Or here are the ZUI docs for setting up the integration with a similar diagram to illustrate the relationships: Z-Wave JS UI

This part isn’t clear to me. You have this?

HA Yellow (Z-wave Integration) RPI4 (Z-wave-JS-UI) Znet

You realize that the ZNET and RPI4 are both Raspberry Pis? You could run Z-wave-JS-UI directly on the ZNet, no need for a separate PI?

1 Like

Now we’re just circling back to your post over here. :laughing:

@pragdogforshins I would take another good look at this post above, it clearly shows the possible configurations. You are currently attempting the second option, with the Nortek being used in place of one of the Z-Nets (the one connected via HAOS).

Yes - this is what I have. I have no effective way of installing ZUI on the ZNET – I read the part about “how to” do this on another thread and I just don’t think my budding linux skills are up to that (although for sure that would be the best and easiest solution perhaps).

My issue is that I have two quite distinct locations where I need ZWave. One is in the house (supported by the Nortek hub). The other is several hundred feet away, well out of the range of the Nortek, in a pump shed. This installation needs only one sensor for a well pump. The ZNET and sensor are both there, and the ZNET is connected via ethernet.

The RPi is in my office (at the house) and was an attempt to bridge the gap. So far I’m struggling mightily to understand how the two sets of services interface. I’ve read stuff for hours; and while I know linux-and HA-capable people can manage a lot of what’s necessary, for me it’s a challenge. (I’m an accountant!)

When I look at the diagram above from @freshcoast (which I did see elsewhere) I see only one hub…not two, which is how I’m configured. I understand that in theory one of them should be just a relay…but how to get there eludes me right now it seems.

So I think I’ve tried this…and I keep getting a “failed to connect” message, regardless of address:
ws://192.168.yyy:217:3000 (or 8091) or using tcp://

Here’s a copy and paste of the image from the other thread.

Replace the top Z-Net with your Nortek. Your RPi is the “External Windows or Docker Container”. Z-Wave JS UI is the server. Each Z-Wave integration instance is a client, which connects to a server.

TCP is the serial port address for the Z-Net, it is not involved with the integration. The ws address is the correct one. 192.168.yyy.217 (hope you aren’t actually using 192.168.yyy:217 which has a colon in the IP) would be the IP address of the RPi. You’ll have to provide more details on how you’ve installed ZUI on the RPi. Either the WS server is not actually enabled (make sure you saved the setting), or you’re using something like Docker and not exposing the port, or there’s a firewall blocking it, etc.

1 Like

Sorry - colon was a typo!!

I’m restoring back to my original base system and will reinstall …I’m not sure how screwed up it might have gotten, so a fresh start seemed wise. I do SO MUCH appreciate the help, time and explanations!

Worked!

Mostly. The ZNET is now picked up in HA as is the Flex Sensor…but the sensor is not “identified” (this may be an inclusion error on my part, so will try to re-interview or re-included it).

Also may be related to the fact that the ZNET’s security settings didn’t seem to work so I had to use “no security”.

@freshcoastI have to extend my deep thanks for your assistance – the Flex Sensor which will monitor my “well is out of water” light works like a charm now! I have two instances of Zwave-JS-UI both working, one located on the HA Yellow and the second on an intermediate RPi (which I’d really prefer was on the ZNET…but that may be for a different day!).

I blanked the ZWave keys in the RPi ZUI and let it establish new ones (using the refresh buttons) and re-included the Flex Sensor, which is now properly identified and whose entities are pushing through to HA’s reporting. Woo hoo!!

I assume that if I turn off the intermediate RPi with ZUI installed, that I’ll no longer be able to “see” the Flex Sensor…but I’ll test and see if that’s true.

1 Like

If you turn off the intermediate RPi then 1) the Z-Wave network is turned off, and 2) the connection from the integration to the zwave-js-server will also be lost. So, correct, nothing will work.

Hah! Beat me to testing it…thanks!!