Managed network for lxd cluster

I have three hosts (kvm VMs), each with a single public IP and a second vNIC that I can freely configure as interconnect between the three.
Now I want to setup a LXD cluster (to run juju/k8s on top) on those three nodes and I would like to use a “managed” network to make use of auto DHCP/DNS.

I was thinking about using lxdbr0 with an “bridge.external_interfaces” link to a vlan on the interconnect.
However, only the bridge.external_interface is a per-target config item, the IP address is bridge-wide and as result binding on all three nodes at the same time. Obviously this is creating some weird issues. Is this intentional or am I missing something? What would be the recommended approach to get a “managed” network on a fresh LXD cluster?