Loading video player...
Most Kubernetes engineers use kubectl apply every single day — but very few actually understand what happens internally. In this video, we go deep inside the Kubernetes control plane and trace the complete journey of a single kubectl apply command — from your terminal all the way to a running container. This is not a surface-level tutorial. We are talking about real internals — API Server request flow, ETCD writes, Controller Manager reconciliation loops, Scheduler node selection algorithms, Kubelet container lifecycle, and CRI runtime execution. Everything demonstrated on a real production-grade kubeadm cluster with 1 master and 3 worker nodes. ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ WHAT YOU WILL LEARN IN THIS VIDEO ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ - What kubectl actually does before sending the request - How the API Server handles authentication, RBAC, and admission controllers - Why ETCD is the single source of truth and what gets written there - How the Controller Manager detects desired state changes - How the Scheduler scores nodes and picks the best one - What the Kubelet does on the worker node when a pod is assigned - How the CRI runtime (containerd) pulls the image and starts the container - What Kubernetes self-healing looks like internally in real time - Why image pull time drops from 17 seconds to 1.5 seconds on second pull - What happens when you delete a pod vs delete a deployment ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ COMPONENTS COVERED ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1️⃣ kubectl — converts YAML to JSON, sends HTTP POST to API Server 2️⃣ API Server — authenticates, checks RBAC, runs admission controllers 3️⃣ ETCD — stores desired state as source of truth for entire cluster 4️⃣ Controller Manager — reconciliation loop, ReplicaSet creates pod objects 5️⃣ Scheduler — scores nodes by CPU and memory, assigns pod to best node 6️⃣ Kubelet — watches for pod assignments, calls container runtime 7️⃣ CRI Runtime — pulls image, creates and starts the container ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ CLUSTER USED IN THIS VIDEO ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ This demo is performed on a real kubeadm cluster built from scratch on Rocky Linux — not minikube, not a managed cloud cluster. 1 master node and 3 worker nodes running in production for 225 days. Every event you see is real output from kubectl get events --watch running live on the cluster. ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ WHO IS THIS VIDEO FOR ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ → DevOps Engineers preparing for CKA (Certified Kubernetes Administrator) → SRE and Platform Engineers who want to go beyond surface-level Kubernetes → Software Engineers moving into cloud-native and infrastructure roles → Anyone who has used kubectl but never understood what happens internally → Engineers preparing for Kubernetes interviews at product companies ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ ⏱️ TIMESTAMPS ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 00:00 — Intro 00:06 — kubectl command — what it does internally 00:12 — kubectl apply internal flow diagram 00:20 — Cluster overview — real kubeadm 4-node cluster 00:41 — Live demo setup — events watch + commands side by side 00:47 — kubectl apply -f deployment.yaml executed live 1:00 — Deep dive — journey through each component 1:03 — Controller Manager & ReplicaSet — reconciliation loop 1:06 — SuccessfulCreate event — pod object written to ETCD 1:26 — Scheduler — pod assigned to worker node 1:40 — Kubelet + CRI runtime — image pull begins 1:50 — Why first image pull takes 17 seconds explained 1:57 — Container started — pod is running ✓ 2:08 — Self-Healing demo — Kubernetes desired state model 2:15 — kubectl delete pod — forceful pod deletion 2:17 — Killing event fires on workernode-1 2:21 — ReplicaSet instantly creates new pod automatically 2:33 — Second image pull — 1.5s vs 17s — cache explained 2:44 — Why Kubernetes always recovers — self healing internals 3:03 — Outro — next video on ETCD internals ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 🔔 SUBSCRIBE TO CLOUDSUTRA ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ CloudSutra publishes deep technical videos on Kubernetes internals, DevOps engineering, GitOps, observability, security, and real production infrastructure. Not beginner tutorials — real depth from real production systems. Next video — How ETCD stores Kubernetes data internally and what happens when ETCD goes down. Subscribe so you don't miss it. ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ USEFUL LINKS ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ Kubernetes Official Docs — https://kubernetes.io/docs kubeadm setup guide — https://kubernetes.io/docs/setup/production-environment/tools/kubeadm/ ETCD documentation — https://etcd.io/docs CKA Certification — https://training.linuxfoundation.org/certification/certified-kubernetes-administrator-cka