Why does my network stop working when adding a second IP to a second NIC in netplan and starting a docker container exposed on that nic's IP?

"Stop working" might be an overstatement, but parts of it do, and I don't know enough about what's happening to phrase it shorter for a title.

Basically, I have

• A machine with 2 nic srunning one VM and a bunch of Docker containers. NIC#1 has IP 10.0.30.21 and NIC#2 was recenlty set up and assigned 10.0.30.22.
  • One VM running Home Assistant OS (IP: 10.0.30.61:8123, and served by NGINX at hass.my_domain.tld)
  • A bunch of Docker containers, incl. of (potential) interest:
    • NGINX (linuxserver/swag) (using ports 80 and 443)
    • deCONZ (using ports 8080 and 8443)
    • Z-WaveJS2MQTT Server (using port 8091 and 3000)

All docker containers of interest expose ports on 10.0.30.21.

Until now, I have only used the one nic, but for unrelated reasons, I want to move my unifi controller to a docker container. Since it uses some ports already in use on the 10.0.30.21 IP (e.g. the ones used by deCONZ), I wanted to set up the second NIC and use a different IP.

So, I added the second interface to my /etc/netplan/00-installer-config.yaml:

network:
  ethernets:
    eno1:
      dhcp4: true
    eno2:            # <-- These two lines are what I added
      dhcp4: true    # <-- These two lines are what I added
  version: 2
  bridges:
    br0:
      dhcp4: yes
      interfaces:
             - eno1
      parameters:
        stp: true

It seems at first to work fine, the new mac shows up and I can assign IP and all that in the dhcp server.

For the unifi controller, I used the port bindings that should tie everything to the "new" 10.0.30.22 IP (from the unifi-controller section in my docker-compose):

    ports:
      - 10.0.30.22:8443:8443
      - 10.0.30.22:3478:3478/udp
      - 10.0.30.22:10001:10001/udp
      - 10.0.30.22:8080:8080
      - 10.0.30.22:1900:1900/udp #optional
      - 10.0.30.22:8843:8843 #optional
      - 10.0.30.22:8880:8880 #optional
      - 10.0.30.22:6789:6789 #optional
      - 10.0.30.22:5514:5514/udp #optional

But when I then start the unifi-controller docker container (linuxserver/unifi-controller), a few things happen:

1. The Home Assistant VM can no longer access deCONZ or Z-waveJS2MQTT services
   • The error logs just say that they are not available at their IP (10.0.30.21)
   • I can still access both of their web UIs at local IP:port (10.0.30.21:8080 and 10.0.30.21:8091), and I can control devices through this web UI (so it seems fully functional)
2. The Home Assistant VM web UI is no longer accessible at hass.my_domain.tld despite:
   • The web UI being reachable and functional at the local IP (10.0.30.61:8123)
   • No errors or warnings in NGINX logs
   • No errors or warnings in Home Assistant logs
   • Maybe errors in docker logs deconz, I'm not sure if related.
     • deCONS: Unbind response failed with status 0x88 for 0x84fd27fffe70d3ed ep: 0x01 cluster: 0x2000
3. I did not loose FQDN access to any other services served by NGINX, which include both services on that machine, and on other machines, e.g.:
   • Nextcloud (local IP 10.0.30.67)
   • DokuWiki (local IP 10.0.30.21)
   • TrueNAS (local IP 10.0.30.27)
   • PiHole (local IP 10.0.30.43)
   • etc.

When I then kill the unifi-controller, noting changes, there's still no access. When I remove the unifi-controller section from my docker-compose, and completely reboot the system, nothing change. Still no access.

The only way I could restore was the remove the eno2 lines from the netplan file, then reboot, and everything came back as before.

What on earth am I doing wrong here?

(Ubuntu Server LTS 22.04)