Six-week CKA study plan

About an hour a day for six weeks, if you already use Linux and containers. Each week ends with hands-on practice, because the exam has no multiple choice.

Last reviewed . Lab lists come from our current CKA catalogue. You can follow the plan with any course or book.

  1. Week 1: Workloads and scheduling

    Workloads & Scheduling 15%

    Start with the objects every other task uses. Speed here pays off in every later domain.

    Study

    • Pods, Deployments and rolling updates
    • ConfigMaps, Secrets and how Pods consume them
    • Probes, resource requests and limits
    • Taints, tolerations, node affinity and PriorityClasses
    • HorizontalPodAutoscaler

    Practise: 31 labs, about 190 minutes

    • W1Rolling update and rollback5 min
    • W9Tune a rolling update strategy6 min
    • W19Switch a Deployment to Recreate6 min
    • W20Finish a paused rollout6 min
    • W21Cut a Service over to the green Deployment8 min
    • W22Rolling update with percentage surge6 min
    • W2ConfigMap and Secret in a pod7 min
    • W10Pods stuck on missing configuration6 min
    • W18Roll out a ConfigMap change6 min
    • W26Immutable ConfigMap and Secret6 min
    • W27Rename a ConfigMap key and inject one Secret key6 min
    • W28Roll a Secret change into running pods6 min
    • W7HorizontalPodAutoscaler (v2)6 min
    • W12Autoscaling on two metrics with scaling policies7 min
    • W23Autoscale a StatefulSet6 min
    • W24Fix an HPA that watches the wrong workload6 min
    • W25Autoscale up, and do not scale down7 min
    • W3Native sidecar container7 min
    • W11PodDisruptionBudget for safe maintenance5 min
    • W13Batch Job that runs out of time7 min
    • W14Tame a CronJob7 min
    • W17Slow starter killed by its liveness probe7 min
    • W29Add readiness and liveness probes6 min
    • W30Drain connections before a pod stops5 min
    • W4Taints, tolerations, nodeSelector6 min
    • W5Required node affinity5 min
    • W6ResourceQuota and LimitRange7 min
    • W8PriorityClass4 min
    • W15DaemonSet on every node, control plane included7 min
    • W16Spread replicas evenly across nodes7 min
    • W31Prefer to keep replicas on different nodes6 min
  2. Week 2: Services, networking and storage

    Services & Networking 20%, Storage 10%

    How Pods find and talk to each other, and how data outlives them.

    Study

    • Service types, selectors and EndpointSlices
    • NetworkPolicy: default deny, then allow
    • Ingress and the Gateway API
    • CoreDNS and service discovery
    • PersistentVolumes, claims, StorageClasses and access modes

    Practise: 58 labs, about 400 minutes

    • N7Pod networking facts5 min
    • N21Host networking: hostNetwork pods and a hostPort7 min
    • N28Pods in different namespaces share the pod network6 min
    • N29ClusterIP and Pod IP are different addresses6 min
    • N30Pin a name to an IP with hostAliases5 min
    • N2NetworkPolicy: allow only from one app7 min
    • N3NetworkPolicy: default-deny egress, keep DNS8 min
    • N13NetworkPolicy: namespace AND pod selector8 min
    • N14NetworkPolicy egress allow-list with an ipBlock8 min
    • N15NetworkPolicy: default deny and named ports9 min
    • N36Allow ingress from one namespace8 min
    • N37Limit one pod's egress to DNS and one app8 min
    • N1NodePort Service5 min
    • N8Headless Service and stable pod DNS7 min
    • N16Multi-port Service with a broken target port6 min
    • N17Service without a selector and a manual EndpointSlice8 min
    • N18Keep an old Service name working with ExternalName6 min
    • N19LoadBalancer Service with externalTrafficPolicy Local7 min
    • N20Sticky sessions with Service sessionAffinity6 min
    • N5Gateway API: Gateway + HTTPRoute8 min
    • N10HTTPRoute with header matching and traffic weights8 min
    • N11Migrate an Ingress to the Gateway API9 min
    • N22Gateway HTTPS listener with a TLS certificate9 min
    • N23Cross-namespace HTTPRoute with a ReferenceGrant10 min
    • N24HTTPRoute filters: rewrite, headers and redirect9 min
    • N4Ingress with path routing5 min
    • N9Ingress with TLS and a default backend7 min
    • N25Default IngressClass and a multi-host Ingress9 min
    • N31Ingress backend by port name5 min
    • N32Point an Ingress at the Service port name5 min
    • N33Move an Ingress off the deprecated class annotation5 min
    • N6Service and pod DNS4 min
    • N12Custom DNS records in CoreDNS7 min
    • N26CoreDNS stub domain: forward one zone9 min
    • N27Pod DNS settings: dnsPolicy and dnsConfig7 min
    • N34CoreDNS Service and cluster domain5 min
    • N35Pod resolver attempts and timeout5 min
    • S2Default StorageClass5 min
    • S4Dynamic provisioning with a provisioner8 min
    • S7Local PersistentVolume with delayed binding8 min
    • S12StatefulSet volume claim templates and retention9 min
    • S14Rework a StorageClass that is in use7 min
    • S16Per-pod scratch space with generic ephemeral volumes7 min
    • S20Create a StorageClass with delayed binding5 min
    • S3Troubleshoot a Pending PVC5 min
    • S6Projected volume and in-memory scratch space7 min
    • S9Raw block and single-pod volume claims7 min
    • S10Retain the volumes of a running StatefulSet8 min
    • S13Share one volume with subPath and subPathExpr8 min
    • S18Mount a host directory into one pod6 min
    • S19Cap an emptyDir scratch volume5 min
    • S1PV + PVC + pod7 min
    • S5Recover data from a Released PersistentVolume8 min
    • S8Bind a claim to a PersistentVolume by label6 min
    • S11Claim and volume stuck in Terminating7 min
    • S15Move an application's data to a bigger claim10 min
    • S17Report which pods use which claims6 min
    • S21Keep a volume before anyone claims it7 min
  3. Week 3: Cluster architecture: access, kubeadm and lifecycle

    Cluster Architecture, Installation & Configuration 25%

    25% of the exam is about the cluster itself. Learn it on a cluster you're allowed to break.

    Study

    • Roles, ClusterRoles and bindings; kubectl auth can-i
    • Users through CertificateSigningRequests
    • Static Pods and the kubelet
    • Upgrading control plane and workers with kubeadm
    • etcd snapshot and restore, drain and cordon

    Practise: 21 labs, about 145 minutes

    • C5RBAC for a ServiceAccount5 min
    • C6New user via CertificateSigningRequest10 min
    • C17Reuse a ClusterRole in selected namespaces6 min
    • C18Aggregated ClusterRoles6 min
    • C19RBAC audit with kubectl auth can-i7 min
    • C20A kubeconfig for a new user7 min
    • C21A ServiceAccount for a CronJob7 min
    • C3Upgrade the control plane with kubeadm12 min
    • C4Upgrade the worker node8 min
    • C10Static pod on a worker4 min
    • C22Bootstrap tokens and the worker join command5 min
    • C23A kubeadm JoinConfiguration file9 min
    • C24Check and renew kubeadm certificates8 min
    • C47List the images kubeadm will use4 min
    • C1Back up etcd6 min
    • C2Restore etcd10 min
    • C11Drain and cordon a node4 min
    • C25etcd snapshot: save, check and report6 min
    • C26Node maintenance with a PodDisruptionBudget9 min
    • C48Restart kube-proxy safely6 min
    • C49Inventory node versions before an upgrade5 min
  4. Week 4: Cluster architecture: installing and extending

    Cluster Architecture, Installation & Configuration 25%

    The newer parts of the curriculum: package managers, extension points and operators.

    Study

    • Preparing a node: container runtime, kernel modules, sysctl, swap
    • Helm install, upgrade and rollback; Kustomize overlays
    • CNI, CSI and CRI: what each one does
    • CRDs, custom resources and operators
    • What a highly-available control plane needs

    Practise: 28 labs, about 205 minutes

    • C13Prepare a node for kubeadm7 min
    • C27containerd and the systemd cgroup driver8 min
    • C39Load the IPVS kernel modules6 min
    • C40Persist vm.swappiness for kubeadm5 min
    • C41Node preflight report6 min
    • C14Prepare a highly-available control plane join8 min
    • C36etcd membership and control-plane topology7 min
    • C37Add load-balancer SANs to the API server certificate12 min
    • C42Read the kubeadm ClusterConfiguration5 min
    • C43Static pods of a stacked control plane6 min
    • C7Helm: install, upgrade, roll back7 min
    • C8Kustomize overlay6 min
    • C15Helm: inspect values and install with a values file8 min
    • C16Kustomize: images, generators and patches8 min
    • C33Helm: install from a chart repository8 min
    • C34Find and fix a failed Helm release7 min
    • C35Kustomize: a reusable component in two overlays9 min
    • C38Migrate a Helm release to plain manifests8 min
    • C12Extension interfaces: CNI, CRI, CSI5 min
    • C28Inspect the container runtime with crictl6 min
    • C29RuntimeClass with scheduling and overhead6 min
    • C30Investigate the CNI setup6 min
    • C46Register a CSI driver6 min
    • C9CRDs and custom resources6 min
    • C31Harden a CRD: validation, printer columns, status10 min
    • C32Install an operator: cert-manager12 min
    • C44Define a namespaced custom resource8 min
    • C45Install an operator from a local manifest8 min
  5. Week 5: Troubleshooting

    Troubleshooting 30%

    The biggest domain at 30%. Everything from weeks 1 to 4 comes back here, broken.

    Study

    • Nodes NotReady: kubelet, runtime, certificates
    • Control-plane components that won't start
    • Pods that are Pending, crashing or OOMKilled
    • Services without endpoints and broken DNS
    • Logs, events and resource usage

    Practise: 39 labs, about 240 minutes

    • T1Worker NotReady: broken kubelet config7 min
    • T2Worker NotReady: kubelet not running5 min
    • T9Pods stuck Pending: scheduling constraints5 min
    • T19Worker NotReady: container runtime problem7 min
    • T20Static pod missing on node016 min
    • T21Pods can't get their image5 min
    • T31A CronJob never runs5 min
    • T34DaemonSet doesn't cover every node6 min
    • T37Report a node's pressure conditions5 min
    • T3API server down: broken etcd connection8 min
    • T4API server down: invalid flag or value6 min
    • T5Pods stuck Pending: scheduler broken6 min
    • T6Deployment creates no pods: controller-manager broken6 min
    • T32kubectl fails with a team's kubeconfig6 min
    • T33API server down because etcd is broken8 min
    • T39Which images are the control plane running5 min
    • T12Resource usage: metrics-server and top8 min
    • T15Container killed for running out of memory6 min
    • T16ResourceQuota blocks a Deployment's pods6 min
    • T28Find the heaviest workload with kubectl top6 min
    • T29Which pods go first under memory pressure? QoS classes5 min
    • T36Give a Deployment Guaranteed QoS6 min
    • T38Report ResourceQuota usage6 min
    • T8CrashLoopBackOff and a failing probe7 min
    • T11Logs, events, and top5 min
    • T22Pods stuck initializing7 min
    • T26Why did the container crash? Previous logs5 min
    • T27Investigate a namespace with events5 min
    • T30A Job never completes7 min
    • T35Read one container's stderr5 min
    • T7Service with no endpoints5 min
    • T10Cluster DNS broken6 min
    • T13Readiness probe keeps pods out of a Service6 min
    • T14NetworkPolicy selects the wrong client7 min
    • T17New Services unreachable: kube-proxy is down8 min
    • T18Pods can't start on node01: CNI broken8 min
    • T23Service port doesn't reach the pods6 min
    • T24Egress policy breaks name resolution8 min
    • T25Pod can't resolve Service names6 min
  6. Week 6: Mock exams and weak spots

    All domains

    Sit a full two-hour mock under exam conditions early in the week. Spend the next days on the domains where you scored lowest, then sit a second mock you haven't seen.

    Study

    • Re-read the exam environment tips
    • Redo every lab you needed the solution for, without it
    • Time yourself: aim for under seven minutes on routine tasks
    • Book the exam for the end of the week while it's fresh

    Practise: 15 mock exams of 120 minutes

    • MockCKA Domain Mock: Cluster Architecture, Installation & Configuration16 tasks
    • MockCKA Domain Mock: Services & Networking16 tasks
    • MockCKA Domain Mock: Storage16 tasks
    • MockCKA Domain Mock: Troubleshooting16 tasks
    • MockCKA Domain Mock: Workloads & Scheduling16 tasks
    • MockCKA Full Mock 0119 tasks
    • MockCKA Full Mock 0219 tasks
    • MockCKA Full Mock 0317 tasks
    • MockCKA Full Mock 0417 tasks
    • MockCKA Full Mock 0517 tasks
    • MockCKA Full Mock 0617 tasks
    • MockCKA Full Mock 0717 tasks
    • MockCKA Full Mock 0817 tasks
    • MockCKA Full Mock 0917 tasks
    • MockCKA Full Mock 1017 tasks

Follow the plan in the labs

Every lab listed above is built for a real kubeadm cluster and graded automatically.