Closed
Description
As a cluster operator of NGF
I want the control plane of NGF to fail-over to another pod when the control plane pod goes down
So that I can ensure I am providing a highly available control plane to the product teams using my cluster.
Acceptance
- NGF needs to elect a lead control plane pod.
- The lead control plane pod is marked as ready and all other pods as unready.
- All Agent connections from the data plane should be to the control plane leader.
- If the leader is not marked as ready, the leader should switch to another available control plane pod.
- Non-leader pods should not send any config.
Dev Notes
- Only the leader pod should be marked as ready by Kubernetes, all other pods are marked as unready.
- Any pod that tries to connect to NGF will only connect to the ready pod.
Dependencies
Metadata
Metadata
Assignees
Labels
Type
Projects
Status
✅ Done