Skip to contentkubesim
A Kubernetes cluster in your browser

Learn Kubernetes by breaking it.

Type real kubectl and watch Pods get scheduled, crash, heal and roll out, with every step explained as it happens. Free, open source, nothing to install.

Start lesson 1: your first Pod →Open the Sandbox

Tap the logo for a node failure. They always come back.

lessons
19lessons
challenges
30challenges
live incidents
16live incidents
explainers
11explainers
practice exams
2practice exams

This cluster is live

A shop with real traffic. Kill its database and watch the errors travel up, then heal.

Full-size Sandbox →
Starting a cluster…

Daily Incident

You're on call. Production is down.

A shop is serving real traffic when something breaks. Find the cause with kubectl and get it serving again before the clock runs up. One new outage a day, the same for everyone: share your time, keep your streak.

Take today's incident →All 16 incidents

Three ways to learn

LessonsOne step at a time, with the cluster in the spotlight. Each step checks itself as you type.19 lessons in 7 tracksChallenges and examsA broken cluster and a goal, no instructions. Or a timed exam in the style of the CKAD and CKA.30 challenges · 2 practice examsSandboxFree play. An empty cluster that's all yours: any kubectl, any YAML, switch Nodes off. Nothing to pass.No rules · share a copy with a link

The path

Seven tracks, from your first Pod to RBAC. Every lesson gives you a fresh cluster, a terminal and a YAML editor, and most checks look at the cluster, not at what you typed.

All lessons →11 narrated explainers →
  1. 01Foundations5 lessons
  2. 02Workloads4 lessons
  3. 03Configuration2 lessons
  4. 04Networking2 lessons
  5. 05Storage2 lessons
  6. 06Operations3 lessons
  7. 07Security1 lesson

What's simulated

Workloads
PodDeploymentReplicaSetStatefulSetDaemonSetJobCronJob
Config
ConfigMapSecretServiceAccountrequests & limitsprobes
Networking
ServiceEndpointSliceEndpointsIngressNetworkPolicyDNS
Storage
PersistentVolumePersistentVolumeClaimStorageClass
Operations
NodeNamespaceHorizontalPodAutoscalerEventmetrics-serverkubectl top
Security
RoleClusterRoleRoleBindingClusterRoleBindingkubectl auth can-i

Back-off timings, events, scheduler reasons and kubectl output follow real Kubernetes defaults, and each behaviour is written up from the Kubernetes source and pinned by tests. It's still a simulation, so some corners are simpler than the real thing.