The migration flow looks roughly like this.
1. cluster level - Create target pod
2. target node - prepare target pod (setting up network and creating ephemeral disks for things like cloud-init)
3. source node - perform migration
4. source node - hand off VMI object to target node
5. cluster level - source pod tears down
It's possible to add lifecycle hooks for items 2, 3 and 4. Which hooks would you need to satisfy your use case?