1. Proposed title of this feature request

OCP4: Ovnkube DB rebuild might could/should happen automatically during outage scenario

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

Considerations for methods or specific triggers that could lead to the ovnkube database to be proactive about when the db needs to be rebuilt. We should be able to trigger this process from the network operator directly, rather than waiting for support to refresh/clean up nat flows, or otherwise re-sync the database with peers.

• Alternatively - can we consider not persisting DB through reboots?  See this KCS - workaround 2:
  • https://access.redhat.com/solutions/7088619 If node reboot just populated the DB fresh every time, this could also alleviate a number of reconciliation tasks.

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

During a protracted outage scenario where the ovnkube db is leading to continued crashloops or is otherwise out of sync with peers, a db rebuild is often suggested as a method to forcibly trigger a sync and redeploy the ovnkube north and south databases. Customer has asked that we request about a method to automatically detect such a condition and trigger the rollout directly (after some timeout period or internal safe window to avoid interrupting normal recovery procedures) - to reduce downtime. 

4. List any affected packages or components.

ovnkube-node

network operator