Attempting to migrate from the official Z-Wave JS to UI version using instructions here and here. Installation proceeds normally and devices are discovered. There are no errors in the logs. However, the devices show as “unavailable” in my lovelace dashboard.
QUESTION:
Do the device entity names change in the UI version, or should that information be preserved? More to the point, do I need to go through all my automations and lovelace cards and update them, or should all that just work after the migration? It’s the difference between an onerous day+ long project and a quick 10 minute setup.
Yes. Zwave JS UI is running (green light). I watched the debug log as the devices were discovered. It took about 10-20 minutes and then stopped after devices were discovered. I also switched back and forth to the device list as it updated, and could see the devices updating. So, I believe it’s working. I did disable the official Z-Wave JS, according to the instructions on the HA setup page. But, on the JS UI Github setup page, they didn’t mention doing that. So, I’m not exactly clear whether I’m supposed to leave the official JS running.
Yes. That’s exactly the line I saw in those instructions. Okay… I’ll go ahead and uninstall it. I’m pretty sure I didn’t get step 4 right either. I don’t have the option to add the UI again in my add-ons. I can stop it and start it.
Step 4 is re-configure the integration (by adding it again), not the add-on. Nothing will work until you reconfigure the integration to connect to ZUI.
Damn it. I clicked “delete” under the Z-Wave JS integration instead of “uninstall” in the add-on. Now all those devices are completely gone, not just unavailable. Did I just delete them from my stick? I did backup the stick yesterday. And, have a full HA system backup from this morning.
No. You removed them from HA. Unless you have a backup, just install the integration as described in the add-on docs. You’ll have to rename everything as it was before.
After restoring HA from a backup and re-installing the UI add-on, all is well. Turns out… not explicitly botching every step along the way seems to work. Thanks guys!