Tenet I
The Control Plane
Governance, truth, and desired state. In Kubblai’s language: the chamber where intent becomes law.
Ceremonial Statement
Fictional scripture; real mechanics beneath it.
“Declare the shape. Seal the intent. Let governance be quiet and absolute—so the cluster may converge without confusion.”
Plain-language Explanation
What this means for operators.
The control plane is Kubernetes’ authority: the API, persistence, scheduling decisions, and controller loops that observe and act. Tenet I is a reminder to treat the API as the source of truth and to encode intent declaratively, not as ad-hoc imperative steps.
If you can’t describe your platform in objects and constraints, you don’t control it—you negotiate with it.
Kubernetes Mapping
Concrete components and why they matter.
- kube-apiserver: the front door and contract; everything is mediated through it.
- etcd: persistence of desired state; your archive of truth.
- kube-scheduler: placement decisions; the rite of stewardship (see Tenet III).
- kube-controller-manager: reconciliation loops; convergence toward intent (see Tenet II).
- Admission control: policy gates that protect the archive from unsafe objects.
Operator Mindset
How to think under pressure.
Protect the API contract
Treat the API server and etcd as sacred infrastructure: secure them, observe them, back them up, and understand failure modes. If the control plane is unstable, everything else is theatre.
Prefer declarative change
Favor GitOps-like workflows, manifests, and repeatable pipelines. Imperative commands are fine for diagnosis, but lasting change should be recorded as intent.
Governance is kindness
Policies (RBAC, Pod Security Admission, resource quotas, admission controllers) aren’t bureaucracy—they prevent accidental harm and preserve cluster stability.
Practical Exercises
Small rites that build real skill.
- Inspect the control plane components in your cluster (managed or self-hosted). Identify where API server metrics and logs live.
- Write a minimal RBAC Role + RoleBinding that grants read-only access to Pods in a namespace. Verify it with
kubectl auth can-i. - Add resource requests/limits to a Deployment and confirm the scheduler’s decisions align with your intent.