1. Proposed title of this feature request

Backup/restore and disaster recovery support in UDN

2. What is the nature and description of the request?

Ensure that UserDefinedNetworks play well with OADP's backup/restore and ACM's disaster recovery.

The flow would be roughly the following (although I'm no SME in this):

• Take a snapshot of namespace's application. e.g. store a Deployment, the ConfigMaps it references, ...
• During recovery, have admin create projects, have the tooling apply all the backed up resources in bulk in this namespace

Primary UDNs are not compatible with this flow, since they cannot be applied in bulk with the application itself. That is because primary UDNs cannot be created if a Pod already runs in the given namespace. i.e. if we apply a bulk of one Pod and UDN, if the Pod gets created first, UDN creation will be blocked.

Some ways of addressing this issue:

• Teach OADP and ACM about primary UDNs and how to apply them first
• Require partially manual restore, where UDNs will need to be created by a human before automation recovers the rest of the project
• Make primary UDNs eventually consistent, e.g. by reacting to their creation by restarting all the Pods in the given namespace to pick up the new network configuration

3. Why does the customer need this? (List the business requirements here)

Backup/restore and disaster recovery are part of our offering today. UDNs aim for seamless integration into the rest of the product.

4. List any affected packages or components.

• OVN Kubernetes
• OADP ?
• ACM ?