4.19 documentation for use of primary UDNs in VM specification wrongly shows the use of the UDN name instead of using 'default' as the network name.
The example contains the use of the custom UDN name:
apiVersion: kubevirt.io/v1
kind: VirtualMachine
metadata:
  name: example-vm
  namespace: my-namespace 
spec:
  template:
    spec:
      domain:
        devices:
          interfaces:
            - name: udn-l2-net   <==== THIS
              binding:
                name: l2bridge 
# ...
      networks:
      - name: udn-l2-net <===== THIS
        pod: {}
# ... 

However, the use of a custom name attaches IP addresses to the VM in question but the network is not accessible from the VM. When creating two VMs using the custom name of the UDN created inside the VM spec, no communication is possible between the VMs.
When changing the spec to use 'default' as the name of the network, the VMs can communicate.

4.19

always in this doc release

1. Create a UDN of name "myudn"
2. create VMs using the name of the UDN "myudn" as the network like:
spec:
  template:
    spec:
      networks:
      - name: myudn
3. ping one VM from the other

Following the documentation and using a custom UDN name instead of using 'default': The VMs have no access to the primary network and cannot ping each other.

The VMs have access to the network and can ping each other.

link to actual doc (4.19): https://docs.redhat.com/es/documentation/openshift_container_platform/4.19/html-single/virtualization/index#virt-attaching-vm-to-primary-udn_virt-connecting-vm-to-primary-udn