Kubernetes Administrator CKA Fast Track

CKA (Certified Kubernetes Administrator) der Cloud Native Computing Foundation (CNCF) ist die weltweit anerkannteste Kubernetes-Zertifizierung — ein Hands-on-Examen in echten Clustern (kein Multiple Choice). Diese Fast-Track-Schulung führt in 2 Wochen zur Prüfungsreife. Themen entsprechend CKA-Curriculum (gewichtet wie Examen): 1. Cluster Architecture, Installation & Configuration (25%) — Kubernetes-Komponenten (Control Plane: kube-apiserver, etcd, kube-scheduler, kube-controller-manager, cloud-controller-manager; Worker Nodes: kubelet, kube-proxy, Container Runtime CRI wie containerd/CRI-O). Cluster-Installation mit kubeadm (Bootstrap, Init Control Plane, Join Worker, HA-Setup mit external etcd), upgrade von Clustern (kubectl drain, kubeadm upgrade plan, OS-Updates). RBAC (Role, ClusterRole, RoleBinding, ClusterRoleBinding, ServiceAccounts, ImpersonateUser). kubectl-Konfiguration mit kubeconfig (Contexts, Users, Clusters). 2. Workloads & Scheduling (15%) — Pod-Lifecycle (Pending, Running, Succeeded, Failed, Unknown; Init Containers; Lifecycle Hooks postStart/preStop). Workload-Resources: Deployment (Rolling Update mit MaxSurge/MaxUnavailable, Rollback, Pause/Resume), StatefulSet (Stable Network Identity, Ordered Deployment, Persistent Storage), DaemonSet (one Pod per Node), Job (run-to-completion), CronJob (scheduled). ConfigMaps und Secrets (Environment Variables vs. Volume Mounts, Best Practices). Scheduling — nodeSelector, Node Affinity/Anti-Affinity (Required/Preferred), Pod Affinity/Anti-Affinity, Taints und Tolerations, Resource Requests/Limits (CPU, Memory), Quality of Service (Guaranteed, Burstable, BestEffort), Priority Classes und Preemption. Horizontal Pod Autoscaler (HPA) mit CPU/Memory/Custom Metrics. 3. Services & Networking (20%) — Services (ClusterIP, NodePort, LoadBalancer, ExternalName), Service Discovery via DNS (CoreDNS), Endpoints und EndpointSlices. Ingress Controllers (NGINX, Traefik, HAProxy), Ingress-Ressourcen, IngressClass. Network Policies (Pod-level Firewall, Default Deny, Allow Specific). CNI Plugins (Calico, Cilium, Flannel, Weave). 4. Storage (10%) — Volumes (emptyDir, hostPath, configMap, secret, persistentVolumeClaim), PersistentVolume (Static Provisioning), PersistentVolumeClaim, StorageClass (Dynamic Provisioning), Access Modes (RWO ReadWriteOnce, ROX ReadOnlyMany, RWX ReadWriteMany), Reclaim Policies (Retain, Delete, Recycle). CSI Drivers für Cloud-Storage (AWS EBS, Azure Disks, GCP Persistent Disks). 5. Troubleshooting (30%) — Cluster-Komponenten debuggen (kubectl logs, journalctl, systemctl status kubelet), Pod-Issues (kubectl describe, kubectl logs, exec für Live-Debug), Service/Networking-Issues (kubectl get endpoints, DNS-Lookup im Pod, Network Policy Conflicts), Resource-Monitoring (kubectl top, kube-state-metrics, Prometheus). Cluster-Recovery (etcd Backup mit etcdctl snapshot save, Restore). 30%! des Examens ist Troubleshooting — Praxis-Drill. Examen-Details: Online-Proctored, 2h, 6-10 Hands-on-Tasks in Live-Clustern (5 Cluster-Versionen verfügbar), Pass 66%. Kann zweimal innerhalb 12 Monaten versucht werden. Empfohlene Vorbereitung: 200+ Lab-Übungen auf Killercoda, KodeKloud oder Killer.sh.