TL;DR: Is there any good way to bring my networking configuration/operation up to the present, without borking my docker and lxc bridges, and keeping my custom lan (static)/wan(dhcp) stuff?
Details:
I run a bunch of web-servers in lxc containers for of some of my hobbies (that actually are used a fair bit) on an ubuntu server on my home connection, as well as local NAS, some docker containers (homeassistant!) and so-on. There is so so much legacy stuff on it. Networking is one of these, naturally.
The host server was originally Ubuntu 12.04 when that was brand new; I upgraded probably around 2017 to 16.04; and probably in 2021 to 20.04; and more recently to 22.04. As you are undoubtedly aware, networking underwent many changes during this time period...
I recently decided to install pi-Hole, in a docker container, for my home network, which is behind an appliance router (EdgeRouterX). It works great for that. But I had to disable dnsmasq - because installation instructions - and that broke the host DNS. I fixed it by manually editing resolv.conf
The server has a connection to the lan, as well as directly to the wan, so that it is kind-of outside the home network, and has a different external IP, but can still talk to the devices on the lan.
My problem is basically thus: Whenever I go to do anything new, all the instructions are assuming either NetworkManager.service, or the older service network-manager, but I'm still back in service networking days, and editing my /etc/network/interfaces, despite all the upgrades, and it becomes quite difficult to manage when all the things I have to do are different than all the newer guides. I've poked at this long enough that I can generally get it working, but there has to be a better way!?
Really, the /etc/network/interfaces is quite simple (though I can't remember how I aliased the ethernet ports):
auto lo
iface lo inet loopback
auto wan
iface wan inet dhcp
auto lan
iface lan inet static
address 192.168.2.88
netmask 255.255.255.0
network 192.168.2.0
Docker appears to use the NetworkManager service, since systemctl is-active NetworkManager.service returns active and nmcli lists all the docker containers and bridge; but everything else is "unmanaged", and LXC seems to handle itself some other way, and its ip's are all configured within the individual container configs. I'm using shorewall as a firewall, which seems to handle some amount of the magic.
And at this point I don't know how my dns resolution will happen if I ever restart, or if I'll have to manually edit resolv.conf every time.
I also generally just remote into the server, though I can plug monitors into it, and use gnome-fallback.
So the question is thus:
Is there any good way to bring my networking configuration/operation up to the present, without borking my docker and lxc bridges, and keeping my custom lan (static)/wan(dhcp) stuff, and fix however my dns resolution occurs? Or, is it better to just keep legacy-ing on?
