I have a Pi 3 B+ using wired ethernet and it boots.
The HA image was installed using Raspberry Pi Imager.
After it boots I can access the web page but obviously the “onboarding” has failed.
It has no DNS name resolution.
"The network issue might be DNS related. The currently used DNS service is: .
To resolve this, you can try a different DNS server."
I tried clicking the buttons that are labeled “Use Cloudfare DNS” and “Use Google DNS” but they seem to have no effect.
I got into my router and tried specifying DNS servers
I connected a monitor and keyboard to see if I could figure out how to fix it in the CLI but I am confused. I tried the dns servers command, power cycled the device but it still says the DNS server is . when I reload the HA web page. How do I restart the onboarding process?
The IP address displayed in “supervisor info” command is not one that would work in this house for internet access. Is there some kind of VPN tunnel established? Is this VPN broken?
No other devices in the home have internet DNS problems.
The IP address you see is probably one for the internal Docker internet running in the Pi.
It will be routed to one on your home network, but it sounds like your network is not detected.
Have you tried a different cable and port?
Hi Wally,
Thanks for the suggestion.
I think that because my router has HOMEASSISTANT listed as 10.0.0.20 and I can connect to http://10.0.0.20:9123 and see the HA web page in a PC on my desk. HA did get an IP address. Is there an CLI command to view the IP configuration like ifconfig?
It is strange that when I ping 10.0.0.20 from my desktop, the reply seems to come from 10.0.0.3. That is the PC I am issuing the ping command from.
Is ssh not available? It would be nice to not have to switch my monitor and keyboard back and forth in order to interact with the CLI.
There is a SSH addon, but you need to have HA running then and you will have to be able to connect to the IP then. http://10.0.0.20 is probably the external IP address seen from the docker network, so that is the one you should use.
I took a second look and the ping command received no response. I guess ICMP may not be enabled so I won’t consider the lack of response conclusive.
Apache (I assume) is bound to 10.0.0.20 and can serve http but it seems another essential service is not bound to that address and is therefore unable to download anything.
I am trying to change or add a nameserver into the configuration using CLI. I can’t figure out the name of the interface. Network info made me think it is enu1u1u but the command fails for that as well as eth0.
Maybe someone can spot the problem looking at this photo.
The output shows your dns server is 10.0.0.1, that’s probably your router.
Some ISPs set their routers to funny upstream dns servers. Try setting the dns servers that the router uses to 8.8.8.8 or 9.9.9.9 (or both of it takes 2)
I have written a different image (Gladys) to the same SD card, run on the same device, same ethernet cable, same router, same router settings. It works as designed.
I have to conclude the problem is a bug in the current HAOS image for Raspberry Pi.
More likely it is your ISP.
ISPs do, from time to time, block GitHub and associated sites, because their malware scanner think they found something bad there.
Sometimes you can just use other DNS servers, but a few ISPs use deep packet inspection and then it is a lot harder to circumvent their blockade.
O2 have been a frequent ISP in this behavior.
If you have the opportunity to try this out, once back to the CLI (screenshot you showed earlier), try: ha dns options --servers dns://8.8.8.8 --servers dns://9.9.9.9
I don’t know how this would work for on-boarding failures, but restart the machine and see if the settings persist.
Today I setup a raspberry pi at the office and it completed the onboarding process.
I was able to move my Home out of Amsterdam and over the the US. I automatically detected a bunch of things in the network like the NAS.
I took that SD card back home to my own unit and started dinking around. I have a Z-Wave USB stick and I tried to add the Z-Wave JS integration. It seemed to know what model it was. But then when it got into the setup is got hung up and stopped responding.
Now it won’t open in the browser any longer. As I look at the activity light goes green and red as if it is trying to do something but I fear it is failing to connect to resources it needs.
I wonder if I can sneak past the ISP and defeat whatever it might be doing.
Good to hear, but the ISPs would those “malware” filters have them running with automatic updates, so it will often occur again and the GitHub site will be blocked again. Then you will have to wait for some one to complain and the filters to be manually opened again. It is a circle.
