Comment 1 for bug 1956846

As a workaround I tested setting `proto 0` to all virtual_routes inside the keepalive.conf by adding `output += ' proto 0'` to the build_config function of KeepalivedVirtualRoute. (https://opendev.org/openstack/neutron/src/branch/master/neutron/agent/linux/keepalived.py#L140)

This works fine as a workaround and fixes the issue for me, but it does not feel like the right solution to do.