☸️ Kubernetes

Core → Advanced

• Architecture: API server, etcd, scheduler, controller manager, kubelet, kube-proxy; the reconciliation loop ⭐
• Workloads: Pod, ReplicaSet, Deployment, StatefulSet, DaemonSet, Job/CronJob
• Networking: Service (ClusterIP/NodePort/LoadBalancer), DNS, Ingress vs Gateway API ⭐, NetworkPolicy, CNI (Cilium awareness)
• Config: ConfigMaps, Secrets (+ External Secrets Operator / Sealed Secrets)
• Health: liveness vs readiness vs startup probes (and how wrong probes cause outages)
• Resources: requests vs limits, QoS classes, OOMKilled, CPU throttling
• Scaling: HPA (CPU/custom metrics), KEDA (scale on Kafka lag ⭐), VPA, Cluster Autoscaler / Karpenter
• Storage: PV, PVC, StorageClass
• Rollouts: rolling update strategy, PodDisruptionBudgets, Argo Rollouts (canary)
• RBAC, ServiceAccounts, Pod Security Standards
• Packaging: Helm (and Kustomize)
• Operators & CRDs (build one with kubebuilder: advanced)
• Debugging: kubectl describe/logs/events/exec/debug, k9s
• Local clusters: kind / k3d / minikube

🧪 Labs (🟢 warm-up → 🟡 core → 🔴 hard → ⚫ boss)

• 🟢 kind cluster; deploy with Helm; k9s
• 🟡 Probes + preStop + graceful shutdown: 0 errors during a rollout under k6
• 🔴 KEDA autoscaling of workers on queue depth / Kafka lag
• 🔴 A kubebuilder operator: OrbitWorkerPool CRD
• ⚫ Kubernetes the Hard Way

🧠 Cognitive tasks

• Trace kubectl apply through every component (write it, then verify it)
• Symptom → hypotheses: “1% 502s for 20 s after every deploy”

🛰️ Orbit integration

• Orbit v3 - Cloud Native runs entirely on K8s

Go deeper

⚙️ Kubernetes Internals · Container Internals · 🧩 Distributed & Cloud Patterns (operator, sidecar)

Resources

• kubernetes.io docs ⭐ · Kubernetes Up & Running 3e · Kubernetes in Action (Lukša)
• KodeKloud CKAD course · Kelsey Hightower’s Kubernetes the Hard Way · learnk8s.io articles ⭐