Hi all,
I have a strange problem that I can’t get my head around, hope one of you can help me with this.
I’m running my HA on a Debian 12 system (x86) in supervised mode and recently I changed my full network setup (all routers/switches/access-points etc.), with that change I also changed my lan ip address from 192.168.1.0/24 to 10.0.0.0/24 and made several other ranges for e.g. guest and iot networks. Another thing I changed is to disable all ipv6 in my network, I currently don’t see the benefit from that and only managing a ipv4 network seems easier.
Now the problem, in my HA the different addons (each running in their own docker container) cannot access the network correctly, few examples:
Advanced SSH & web terminal addon: I can use dig to get a valid ip-address from my dns server, I can to a ‘ping -4 device.lan’ succesful, but when I do ‘ping device.lan’ it returns a ‘bad address’ error.
Home assistant core: I use the opentherm-gateway integration to connect to my central-heating system. When I use a name in the url link ‘socket://device.lan:1234’ it fails to connect with a ‘name does not resolve’ error, but when I just use the ip address in url it does connect.
InfluxDB addon: fails to make the connection with an error ‘Failed to establish a new connection: [Errno -5] Name has no usable address’
When I check ourside of the supervised HA environment (just login via SSH directly on the Debian server) I can connect correctly and all seems fine (dig and ping work correctly, the network connection reported via NetworkManager look ok).
I think it somehow has to do with the ipv6 configuration, that the different docker containers are still using this protocol to try to make the connection while I explicitly disabled this on all interfaces in /etc/NetworkManager/system-connection.
Any clues about what can be happening here, or what I can do to further debug/fix this ?
Anyone a clue about what might be happening on my system ?
In the meantime I still haven’t fixed the issue
What I noticed is that when I disable all ipv6 on my host via sysctl I see the ipv6 addresses dissapear from all network interfaces. But when I then restart the docker service the ipv6 addresses re-appear on the interfaces. All my docker networks however are not cofigured with ipv6 and also I added a startup option for the docker-deamon not to use ipv6…
Issue is still the same, if I try to connect to some other devices from my HA by name this seems to use ipv6 and then it fails. To prevent this I’ve now changed all my connections to only use an ip address instead of a name, this works but is not very handy to maintain.
So any help in gettng the dns and/or ipv4/6 working as it should be (all using just ipv4) would be greatly appreciated.
Well, still talking to myself it seems issue seems to not be related to ipv6 config.
I’ve now re-enabled ipv6 in my whole network (router, switches etc.), including also the debian host, docker and home-assistant configuration. I do have a proper global ipv6 and ipv6 dns is also working alongside the full ipv4 network.
When I do a ‘dig device.lan’ I get a valid ipv4 address as answer (the device doesn’t support ipv6), I can do a ‘ping -4 device.lan’ which is succesful, but a ‘ping device.lan’ and a ‘ping -6 device.lan’ both fail with a ‘bad address’ error message. When I do a ‘ping -6 google.com’ this is succesful, so the ipv6 dns lookup and network are working ok.
Also I still cannot acces any of my devices based on name (e.g. I use motioneye add-on for some network camera’s and otgw integration for central-heating), but accesing these devices with just the ip-number does work.
I’m starting to run out of ideas on what to check, all things I know of I did check and all seem to be ok, but HA isn’t working as it should
If anyone has any suggestions about what I can try to get this fixed, help is greatly appreciated.