Z-Wave JS UI Upgrade from 9.16.2 to 9.17.0 resulted in lost of half of the zwave network

Hello,

A few days ago I updated from 9.16.2 to 9.17.0. When the update was complete only 10 of my 20 devices are showing on the network.

I restored the 9.16.2 backup HA made, but that did not fix the problem.

I am using HA version 2024.7.4 with zwavejs-iu as an addons. I am using a Zooz 800 Series GPIO Module ZAC93 (FW v1.20 SDK v7.19.3) over Ethernet in a Olimex ESP32-PoE with ESPHome firmware. https://tubeszb.com/product/z-wave-poe-kit/

Attached are the logs I have and node config. I really do not know where to even start troubleshooting. I have a Proxmox backup from 6 days ago, but if the issues is with the controller it would be a waste. The last controller backup I have is from March when I switched from a Zwave.me controller to the Zooz controller.

Here is a log from the 5th with a controller restart showing all the correct nodes.

Here is a log from the 8th. I am not sure if this is pre or post update.


My 9.17.0 upgrade went without issues.

Every other update has been no problem. It is either a coincidence or or borked the network.

That implies this was coincidental with something else. If it were the Z-Wave update and you rolled it back with the same results then it’s not a software bug most likely.

Have you tried healing your network or trying to re-interview edge devices?

I agree, maybe the reboot after the update caused it. Again I have no clue. I have not performed a heal or anything because I did not want to potential mess something up.

I pulled the node list and the ones with the pound just disappeared from JS UI and by extension HA.

{"1":{},
# "2":{"name":"Living Room Lights"},
"6":{"loc":"Laundry Room","name":"Washing Machine"},
"7":{"loc":"Laundry Room","name":"Dryer"},
"8":{"name":"One Plug Outlet"},
# "14":{"loc":"Party House","name":"Party House Inside"},
# "15":{"name":"Living Room Lamp"},
"17":{"loc":"Party House","name":"Pary House Door"},
"19":{"loc":"Party House","name":"Party House Motion"},
"28":{"name":"Master Lights","loc":"Master Bedroom"},
"29":{"name":"Porch Lights","loc":"Back Porch"},
# "30":{"name":"Master Scene"},
"33":{"name":"Office Switch","loc":"Office"},
# "34":{"name":"Front lock","loc":""},
# "35":{"name":"Themostat T6","loc":"Hallway"},
# "36":{"name":"3D printer","loc":""},
# "37":{"name":"Party House Outside","loc":"Party House"},
# "39":{"name":"Front yard plug","loc":"Front yard"},
# "40":{"name":"Big gate","loc":"Side Drive"},
"41":{},
 # "42":{"name":"Workshop Light","loc":"Workshop"}}

Should I use one of the rebuild routes options or Re-Interview nodes?

I don’t see anything wrong with that.

Whenever my ZWave mesh network stops working (maybe once every 4 months) I power cycle my Yellow Box and Zwave comes back.

I have tried restarting the zwave controller, Zwave JS UI and the entire HA vm nothing has fixed it. The problem is the controller does not recognize the nodes at all. It’s as if they never existed.

I tried a rebuild routes and that only rebuild for the current nodes. It did nothing for the nodes that disappeared.

Log snippet from 08.05:

2024-08-05T22:55:46.997Z CNTRLR   received additional controller information:
                                    Z-Wave API version:         1 (official)
                                    Z-Wave chip type:           EFR32ZG23 / ZGM230S
                                    node type                   Controller
                                    controller role:            primary
                                    controller is the SIS:      true
                                    controller supports timers: false
                                    Z-Wave Classic nodes:       1, 2, 6, 7, 8, 14, 15, 17, 19, 28, 29, 30, 33, 34, 35, 36, 37, 39, 40, 41, 42

Log snippet from 08.08 after the update:

2024-08-09T02:48:11.644Z CNTRLR   received additional controller information:
                                    Z-Wave API version:         1 (official)
                                    Z-Wave chip type:           EFR32ZG23 / ZGM230S
                                    node type                   Controller
                                    controller role:            primary
                                    controller is the SIS:      true
                                    controller supports timers: false
                                    Z-Wave Classic nodes:       1, 6, 7, 8, 17, 19, 28, 29, 33, 41

I am failing to understand how the controller lost nodes between restarts.

Have you tried unplugging the dongle and plugging it back in - or even moving it to a different USB port?

1 Like

I have rebooted the controller a few times. It is a Zooz ZAC93 plugged into a POE ESP32. No USB ports to switch around.

The node list is stored in memory on the ZAC93 board, so it’s pretty unusual for it to just forget 10 nodes. Have you ever saved a NVM backup by chance? Is there anything the missing nodes have in common (all S2, or 800-series, or battery-powered, etc)? What firmware version is on the ZAC93?

There is a decent chance something was wrong prior to the upgrade/reboot, which just revealed the issue. Hopefully your board isn’t dying, but if you don’t have a backup to restore, consider trying to exclude and re-include a few missing nodes and see if they survive a reboot (of the esp32 and zwavejs).

There is no link I can find between all the devices missing. The logs before the update and restart all seem fine. I guess since I did the restore I lost the logs from directly after the update.

It is possible the controller is failing. I will try a NVM backup that is a 6 months old at this point. Hopefully it will work. Controller is on FW v1.20 SDK v7.19.3, which is the latest zooz has.

Shutdown and Remove power from the device with the zwave module. Give it a few minutes and then lug it back in.

If that doesn’t fix it take a new backup before restoring.

Already did that. Interestingly the backup after the failure is significantly smaller than the backup with all the nodes still in tack.

image

The restore worked and seems to be functioning as expected. It has survived multiple reboots so far.

That’s bizarre the backups afaik are always 256kb. So looks like an NVM corruption. Glad you had a backup.

Well that’s bizarre. I restored from a backup, then did another backup and its 40kb again.

I have added more devices since the restore and I just did another backup this morning and its 40kb too. The network is functioning as expected right now.

I looked at the zwavejs change logs and see nothing regarding NVM backup size changes. May be worth while asking a question there to see if thats ok or if your controller is on it’s last legs.

Yeah I went there first, but the issues has been open for a while. Bigger fish to fry and all. I guess I am reaching out to Zooz since this things is only a few months old.

Was your 256KB backup from a 500-series?

500-series NVM backup is 256KB, 700-series is 48KB, and 800-series is 40KB

1 Like

Yes!! The old backup was from a Zwave.me controller which is 500 series. The Zooz Zac93 is 800 series.

Thank you for that tidbit.

1 Like