Please fill in the following information:

Issue description

The lab requires that the VM are deployed on worker nodes to attach the network attachment definitions into the correct NodeNetworkConfigurationPolicy.

However, the way the VM is define does not enforce the VM to be deployed on worker nodes.

If the scheduler decides to deploy it on the master nodes, the VM will be pending  because the networkattachmentdefinition cannot find the network device. Worse, it will not be automatically destroyed as it does not have a probe.

The only way to solve the problem is by live migrate the VM 

Steps to reproduce:

It depends on the scheduler. It is completely reliant on the conditions of the cluster

Workaround:

Click the VM and ask it to live migrate to another node that has these devides.

Expected result:

Everytime the VM muste be deployed on worker nodes