I have a USB Sonoff Dongle-P that is connected to one of my 2 Proxmox nodes, working with ZHA with no issues.
I want to make Home Assistant be able to fail over to either node.
When I go to migrate radio on HA it does its backup, resets the radio, then I specify that I want to restore an automatic backup. I indicate the socket://<ip>:8888 correctly but when I click sumbit it aborts quite a few seconds later with “Unknown Error”.
I do see that the connection has reached the zbbridge.
In the log I see no errors.
I did flash the correct zigbee coordinator firmware.
Any step I may have missed?
Can I activate/extract extra logging that may help?
This is very strange… I insisted a lot (since 10 hours ago trying everything).
Suddenly, I launched Wireshark and noticed that I could hold a tcp connection to port 8888 of the IP and that data was coming out…
I recognized the pattern FExxxxxxxxx FExxxxxxxxxxx of ZNP and my hopes went up.
So… I tried once again… it failed with “Unknown error” and went to the menu where it asks if I want to restore a backup (not the automatic), create a new network etc (even if I had come from a migration where I asked to move to another device). I had the backup and, like more thatn 100 times before, I uploaded and pressed submit and it worked.
Sooo… I power-cycled the znbpro and it recovered after a 20-30 seconds with no ZHA/HA intervention.
Conclusion. Flash it… leave it for almost a day… use it…
PS: (I have 200+ zigbee devices, but I’m pretty sure that does not influence the first radio migration where only the IEEE, NWK and others are set)
Note that even though it technically works, Zigbee Coordinator adapter over WiFi-based Serial-to-IP bridges are not recommended because using WiFi. Instead a Ethernet (wired) connection is recommended if want a network connected Zigbee Coordinator. So suggest check out ex. ZigStar and Tube’s Zigbee Gateway products. See warnings in ZHA integration and Zigbee2MQTT documentations:
Thanks!!! I do know about serial-over-wifi issues and understant that it may be a problem.
I do have a stable network and the AP (Cisco) is 30cm away from the gateway with <1ms latency. Have a dedicated SSID and VLAN for this bridge too on switching that is GBit.
The 200+ devices are working fine. Once in a time (once per week) one may misbehave, but returns by itself, with ZHA restart or even device powercycle.
But this happened even with the dongle connected to HA anyway (and much less times now!!).
WiFi radios are high-energy and will drown our low-energy radios like Zigbee. So always need to place Zigbee devices a little away, regardless of which channels are used, just due to general EMF/EMI/RMI.