I am trying to replicate the trick that Kinsta uses on GCloud to run PHP on Ubuntu VMs to allocate the RAM and handle caching for all the deployed sites instead of running multiple sets of cache threads and pools, and then deploy multiple independent Wordpress sites as LXC containers under this shared resource pool to fully bypass the limitations of Wordpress Multi-Site without going broke spawning a bunch of additional VMs.
The problem I am having is I cannot seem to get the LXDBR0 bridge interface to come up since Netplan came into being without losing connectivity to my VM, because when I apply the new plan it takes away the base NIC for the OS and puts it under the bridge's control. After that I am unable to reach it ever again unless I reset (see reboot) the instance, and for whatever reason a reboot wipes my new config out and resets everything back to the base config it was provisioned with initially.
network: version: 2 renderer: networkd ethernets: enp1s0: dhcp4: no dhcp6: no bridges: lxdbr0: dhcp4: no interfaces: - enp1s0 addresses: [192.168.0.10/24] gateway4: 192.168.0.1 nameservers: addresses: [192.168.0.1,8.8.8.8,8.8.4.4] parameters: stp: false forward-delay: 0
I expected the root vNIC to drop offline while it releases its address and then the bridge makes a new DHCP call to the IaaS, but then I could reach my VM again in a little while. What happens instead is the IaaS loses all connectivity to the VM that is IP based.