Abstract
By default, a Linux bridge br0 is created on the administration interface in the hypervisor. This is useful to access all the VMs and the hypervisor with the same physical interface.
This bridge is also created inside the VM, but has no use. It should be removed.
Technical implementation
This bridge is applied in the cluster_setup_network playbook by the systemd_networkd role and the vars/network_vars.yml file. One simple solution could be to not apply this role inside the virtual machines, but in that case, the IP address of the VM will never be set.
In my opinion, we should provide a default network for the VM (just applying the admin IP address) and allow the user to modify it at will. This could be done, for example by creating two variables seapath_network and seapath_netdev that can be fully customized.
This is already the behavior if you use netplan. But the combined use of both netplan and systemd-netword is not clear now.
Abstract By default, a Linux bridge br0 is created on the administration interface in the hypervisor. This is useful to access all the VMs and the hypervisor with the same physical interface. This bridge is also created inside the VM, but has no use. It should be removed.
Technical implementation This bridge is applied in the cluster_setup_network playbook by the
systemd_networkdrole and thevars/network_vars.ymlfile. One simple solution could be to not apply this role inside the virtual machines, but in that case, the IP address of the VM will never be set.In my opinion, we should provide a default network for the VM (just applying the admin IP address) and allow the user to modify it at will. This could be done, for example by creating two variables
seapath_networkandseapath_netdevthat can be fully customized. This is already the behavior if you use netplan. But the combined use of both netplan and systemd-netword is not clear now.NB: This subject was already discussed in https://github.com/seapath/ansible/pull/565#discussion_r1680546842