In OVN-K we have a use case to attach an interface to a VRF.
The VRF is managed by OVN-K.
The interface is managed by the cluster admin.

The most straight forward way to do so is with an NNCP like

apiVersion: nmstate.io/v1
kind: NodeNetworkConfigurationPolicy
metadata:
  name: udn-test-vlan
spec:
  desiredState:
    interfaces:
    - name: enp3s0
      state: up 
      controller: udn-test

where "udn-test" is the OVNK managed VRF and "enp3s0" is the admin managed interface.

However, applying that NNCP when the VRF doesn't exist fails and it's not retried.

We cannot guarantee the order in which the NNCP is applied and the VRF creation happens in scenarios like day-0 configurations, node scale ups, reboots, backup restore, etc and thus we have conncerns that without NNCP retries eventual convergence in these scenarios won't occur.

More details in OCPBUGS-55325.

We would like to have some recommendations:

• Are NNCP retries a reasonable ask? If so, what would be the timeline of such a request being delivered?
• Is there any alternative NNCP configuration that would better address our concerns?
• Are we missing some detail that would make our concerns invalid?