We want to check if the downtime of the live-migration is less if we mimic the bridge CNI solution of creating the pod interfaces with IFSTATE DOWN, and let KubeVirt set them up.

This could be done in two ways:

• create the pod interface with IFSTATE DOWN (as done for bridge), and rely on KubeVirt to set the interface UP
• create the LSP with IFSTATE DOWN, and once the VM is now available in the target pod set the LSP state UP

Both should be possible, understand which is preferable.

We'd still prefer to rely on the multiple requested-chassis OVN feature, but we're waiting for confirmation from the core OVN team that this is doable (requested feasibility study in https://issues.redhat.com/browse/FDP-770)