I am the proud owner of the aforementioned NAS system. Everything’s working, except homeassistant. As I have quite some HA addons in my current (proxmox-based) installation, I have no option to use Docker. Therefore I chose the built-in virtual machine manager.
Imported it in the virtual machine manager, which is obviously KVM-based and after restoring a back-up from my proxmox environment nearly everything was smooth sailing. The one thing that isn’t working is the zwave usb dongle (aeotec gen 5). It, for some reason, cannot seem to pass through the USB device. I am seeing the device in ugreen host through SSH, but I cannot get it to pass through properly to the HA VM.
After thorough troubleshooting, I sent a support request to ugreen and their virtualization experts are working on it.
Are there people running into similarities, or were people able to get this to work successfully? Is it a dongle-specific issue, or am I doing something wrong here?
Okay so I’ve managed to get homeassistant in a docker to work, with zwavejsui in a separate docker with usb support that actually worked. However I noticed there are long interview times of the devices and the names etc. are gone. Reusing the usb stick on my proxmox environment shows everything’s in order again. So what portions of this data is on the stick, versus in homeassistant itself? How can I move the data over from proxmox instance to the docker instance?
Quick update:
the formal response of Ugreen is that they observed that homeassistant does recognise the usb stick, but immediately closes the connection, therefore, not being able to use the usb stick. They say it’s an issue with the HA image. Has anyone observed this behaviour, or can anyone help me figure this out?
I have found the problem in the meantime, now to find a solution:
The problem is that if I restore the full system back-up (which includes zwavejsui) it will error out on the driver portion, saying it cannot open the USB drive. This makes sense as the backup is from another instance (the one on proxmox) so I change /dev/serial/by-id/… to dev/ttyAMC0. This obviously is configurable in zwavejsui.
So I made the change and it still errors out, restarting zwavejsui or HA entirely makes no difference (it just changes the usb device reference in the error).
Now for the weird part:
If I install HAOS and not do the system restore, HA finds out of the box devices on your network, including the zwave controller (which is listed as AMC0). Going forward with the configuration, leads me to a functional zwave network. The downside obviously is that everything is named ‘node ’, effectively riddling me with random names and breaking all automations.
So I’m led to think that the reconfiguration of the zwavejsui (with the address my ootb zwave autodiscovery thingy) doesn’t work. I really have no idea how to go about this. I’ve exported a list of all zwave devices with entity names and node ID’s, maybe that’s way forward?
Obviously I could also manually install zwavejsui in the new box and do a NVM backup of the usb drive. But how does that differ from the system restore?
Regarding the proxmox move, nothing too special: The proxmox host is really old nuc that has few resources. The nas has a better CPU, way more ram and a 2tb lightning fast ssd, therefore faster in every way.
New test results:
I tried restoring HA fully, except for zwaveJSUI. The result was as expected, HA couldn’t initialize zwave. Which makes a lot of sense as zwaveJSUI hosts a zwave server. All zwave devices were there as ‘node X’, X being nodeID. In zwave I did a reconfigure and opted for the builtin zwave service. After that, reinterviewing of all devices occured and slowly but steadily devices reported in and names resolved back to their original states. I checked some entities and they had their proper names back as well. I’m gonna let this sit for a while and see if all devices come back properly.
The whole point of /dev/serial/by-id/ is that the name doesn’t change. It’s the way to go and should “just work”.
Can you share the output of the commands above as well as the error log of zwave and ls -l /dev/serial/by-id/?
I don’t use zwave so I can’t tell you specifics about it.
Did you reboot the VM after restoring a backup? Addons can get confused without that.
You could simply install Proxmox VE on the NAS.
Immediately after I had finished going through the support case with ugreen, the guys at HA released a new image, namely “haos_ova-17.0.qcow2.xz”, I believe I was on 16.3 before. In this 17.0 release, everything ‘just worked’. I then tried to see if my zwavejsui docker instance worked as well, and this was definately the case, but ‘just work’ now. I am using the by-id string, no issues since the new image. I can’t see as to why this change (https://github.com/home-assistant/operating-system/releases) affects me in a positive way, possibly due to the kernel update for this release?
I asked for confirmation at ugreen, they changed nothing in my setup, so it must be the image. Why the zwavejsui docker is suddenly working now (which was giving the exact same error), remains a big mystery to me though.