#HA url not reachable after update

1 messages · Page 1 of 1 (latest)

spark breach
#

Hello!
Trying to find a solution. I am on holiday and after updating a HA component and restart the interface is not reachable.
I can connect to my priavte network via VPN but I am not able to access the HA in any way.
I installed it via Raspberry Pi Imager some months ago. There are several plugins installed also EVCC to charge my EVs.

Is there any chance to get a connection to HA or reboot it to safe mode from long distance?

Home Assistant has an IP adress given in my UniFi Network. It appears as connected but without significant activity last 24 hours.

I can ping the server ip via mac terminal.

A SSH connection is refused.

Thanks in advance!!! ❤️

untold torrent
#

You need a CLI to troubleshoot. If SSH doesn't work you need a keyboard and monitor.

spark breach
#

Thx. Then I will do it after my holiday

light mulch
#

Had similar issue on two systems at different locations yesterday when upgrading the previous day.

Found that the Home Assistant URL for local network was on automatic and assigning an address that wasn't in my network. This also caused Home Assistant Cloud not to work.

I ended up changing it to the correct local network address via the CLI command prompt using sed but I was unsure how to turn auto off.

I needed to be very quick when the system came online after restarting, I used the GUI and turned off auto.
Funny thing is if you turn off auto then it turns it on again. You need to quickly turn it off twice and then quickly press save.

After all of this one system was then fine but the other needed the router to be rebooted also.

you can use this to check what it is currently set to via the root command prompt when you get back.

root

cat /mnt/data/supervisor/homeassistant/.storage/core.config

untold torrent
#

ha network info is the proper way.

light mulch
#

I don't think that network info shows the external_url and the internal_url.

untold torrent
#

Indeed. There seems to be no CLI subcommand to get/set it either :<

lone prawn
lone prawn
# untold torrent Indeed. There seems to be no CLI subcommand to get/set it either :<

The internal_url & external_url settings are used for mapping the name from IP address.

HA should still work via http://ip.adress:8123

You could also try http://homeassistant.local:8123 when connected to your VPN - although this will depend on whether you are on the same subnet (if on a different subnet you would need an mDNS reflector)

The external_url may not work as that will probably be secured via SSL which will require connections using the name otherwise the certificate won't match. You could get a browser warning which you can continue past, but HA may write out the links with name

You can use nmcli to configure Network Settings on HA OS, but you will need to drop into the host shell (type login enter from the ha> cli prompt)

light mulch
#

HA didn't work on two systems using http://homeassistant.local:8123. On boot you could see HA starting up with this address but it wouldn't take log for it to display connection lost. Reconnecting. Then it wouldn't reconnect. 100% a bug, hopefully someone looks at it before to many users get frustrated.

#

And the way I fixed both was turning off auto for internal_url and setting it to my static network address.

#

Also both systems only have one DHCP server.

#

Both systems have been working for years until upgrade was done to OS 16.1 or 2025.8.2. They were done almost at the same time.

#

Ok on nmcli, I might try that next time.

lone prawn
#

1). Providing a single place to see the IP addresses in use on your network
2). reducing the possibility of you manually configuring 2 devices with the same IP.

light mulch
#

I do have a reserved IP address with the Mac address on my DHCP router and always have had this.

#

Also the IPv4 address has always been static. What I am saying is the issue is with the Home Assistant URL local network that is for text-to-speech or other hosted media. I think it shouldn't cause an issue but it did when the local network address was set to auto. Also someone else seems to have had a similar issue core #150782 and maybe issue OS #4162

lone prawn
#

I think you need to look at your network. You have said you found the HA URL for local network was on automatic but getting assigned an address that wasn't in your network. As I said, this indicates an issue with DHCP.

You are also saying http://homeassistant.local:8123 didn't work and if this is on a flat network (ie. same subnet as client PC and home assistant server) you have something blocking mDNS.

light mulch
#

I also mentioned it happened on a friends system also. It hasn't failed on both systems since I have changed the Home Assistant URL local network.

#

http://homeassistant.local:8123 did work for a short while but then failed while loading integrations and wouldnt come back online. I was able to ping the HA sever with the static IP when this happened.

lone prawn
#

@Markus by any chance are you running teh DuckDNS addon and Nginx SSL proxy add-on?

light mulch
#

No.

dark sandal
#

I having same issues as well

#

not able to access using ip as well as homeassistant.local. It works for a day or two and then I have to reboot the whole system

lone prawn
dark sandal
#

I am running it on Dell R720X Proxmox - all other VM's are working fine

#

IP address is reserved

lone prawn
dark sandal
#

yes it does

lone prawn
# dark sandal yes it does

anything in the home assistant logs that could indicate an error? In the HA UI go to System -> Settings -> Logs use the selector at the top right to choose a log to look at. In particular check Core, Supervisor and Host

dark sandal
#

so I think it is fixed and working much better , had to disable/delete few plugins - proximity , tp-link , roborock , imap gmail, google calendar

lone prawn
#

for eveyone still having this issue, try the following:

Clear browser cookies & cache → removes outdated auth tokens
Delete all old refresh tokens under Profile → Security
Reinstall the Home Assistant app on mobile → forces new login with fresh tokens