Kubernetes · Orchestration under pressure
What runs in the control plane and why does it matter?
Short answer
The control plane runs the API server, etcd, scheduler, controller-manager and often a cloud-controller-manager. It decides what should happen, while nodes do the work.
The API server is the only component that reads and writes etcd. It handles authentication, authorization, admission and validation, then persists objects. The scheduler watches for pods with no node and scores feasible nodes. Controller managers watch objects and reconcile them, for example ensuring a Deployment has the right ReplicaSet.
- Losing the scheduler stops new placements but does not stop running pods
- Losing controller-manager stops reconciliation of built-in resources
- Losing the API server blocks all control operations, though existing pods keep serving traffic
- Losing etcd quorum makes the API server unable to commit changes
- Cloud-controller-manager integrates load balancers, routes and node lifecycle with the provider
kubectl get pods -n kube-system
kubectl get leases -n kube-system
kubectl get endpointslices -n defaultFor production, run an odd number of control-plane nodes, spread them across zones and back up etcd. Managed clusters move this burden to the provider, but you still own admission policies, API availability expectations and upgrade timing.
Follow-ups you should expect
Interviewers use these to check whether the first answer was understanding or memorisation.
- 1What still works when the control plane is unavailable?
- 2Why must etcd have an odd number of members?
- 3What does the scheduler consider when placing a pod?
- 4How do you protect the control plane during upgrades?
Say it in the interview
The control plane is the decision layer: API server, etcd, scheduler and controllers, plus a cloud controller in managed environments. The API server is the only writer to etcd, and the scheduler and controllers react to objects through it. If it goes down, running pods usually continue, but nothing new gets scheduled or reconciled. That is why I treat etcd backups and control-plane availability as production-critical.
Use this as a model, not a script. Saying it in your own words is what makes it survive follow-up pressure.
Related questions
Reading is step one
Now say it out loud.
Understanding an answer and delivering it under pressure are different skills. Rehearse this topic in a live Kubernetes mock interview and get scored feedback on depth, clarity and confidence.