# Cluster Administration

Running workloads on a cluster someone else built is one skill. Being the person who built, upgrades, and keeps that cluster alive is another. This section covers the control-plane and node-lifecycle work: standing up a cluster with `kubeadm`, understanding what etcd actually does for you, safely draining and rejoining nodes, upgrading a live cluster without an outage, backing up and restoring cluster state, and isolating tenants inside a shared cluster.

If you're deploying applications *onto* an already-running cluster, [Workloads and Scheduling](<https://computecentral.in/kubernetes/workloads-and-scheduling/index.md>) and [Configuration and Packaging](<https://computecentral.in/kubernetes/configuration-and-packaging/index.md>) are more directly useful. This section is for the people responsible for the cluster itself.

## Read in this order

1. [kubeadm Cluster Setup](<https://computecentral.in/kubernetes/cluster-administration/01-kubeadm-cluster-setup/index.md>) — `kubeadm init`/`join`, control-plane prerequisites, and why nodes stay `NotReady` until you install a CNI
2. [etcd and Control Plane Internals](<https://computecentral.in/kubernetes/cluster-administration/02-etcd-and-control-plane-internals/index.md>) — why etcd is the cluster's single source of truth, and why only the API server ever talks to it
3. [Node Management](<https://computecentral.in/kubernetes/cluster-administration/03-node-management/index.md>) — `cordon`/`drain`/`uncordon`, adding and removing nodes, and the node conditions that drive eviction
4. [Cluster Upgrades](<https://computecentral.in/kubernetes/cluster-administration/04-cluster-upgrades/index.md>) — the version skew policy, the kubeadm upgrade order, and how managed platforms change the picture
5. [Backup and Restore](<https://computecentral.in/kubernetes/cluster-administration/05-backup-and-restore/index.md>) — etcd snapshots and Velero, for both disaster recovery and cluster migration
6. [Multi-Tenancy](<https://computecentral.in/kubernetes/cluster-administration/06-multi-tenancy/index.md>) — namespace isolation with quotas, `NetworkPolicy`, and RBAC, versus separate clusters or virtual clusters

> [!TIP]
>
> **Managed clusters still need this**
>
> Even on EKS, GKE, or AKS, where the control plane is someone else's problem, the node-management, upgrade-coordination, backup, and multi-tenancy chapters here still apply directly to your worker nodes and workloads.

## Next

Once the cluster itself is stable, continue to [CI/CD and GitOps](<https://computecentral.in/kubernetes/cicd-and-gitops/index.md>) to automate what gets deployed onto it.
