From Running Containers to Maintaining a Service
Connect Docker experience to Kubernetes responsibilities and distinguish desired state from observed state.

Hands-On Kubernetes with kind — From Core Concepts to Deployment and Operations — 2026 English Edition
Overview
Move from a first Kubernetes deployment to a system you can operate and explain. Twenty chapters follow Mini CloudShop through Pods, Deployments, Services, Traefik Ingress, configuration, persistent storage, probes, HPA, Jobs, StatefulSets, RBAC, NetworkPolicy, and Gateway API. Recorded lab output, 44 diagrams, a 125-term glossary, four worked cases, and forty additional questions with reasoning connect concepts to observed behavior. Finish by rebuilding the application in a fresh three-node cluster and running recovery drills.
Docker 29.2 · kind 0.33 · Kubernetes v1.37 · Traefik 41.6 · Gateway API 1.5.1
25,431 words (20 chapters + 4 appendices)
Local edition complete — preparing KDP publication
An Nginx Web front end calls an Express API backed by Redis. The same small application makes request paths, workload recovery, storage lifetime, and policy boundaries visible.
Enterprise cloud architect and founder of AIDevOps. His teaching connects application architecture, platform operations, and evidence-based troubleshooting.
Lab Environment
Every manifest in this book can be applied to a three-node cluster created with kind (Kubernetes in Docker) on one computer with Docker installed. The cluster has one control plane and two workers, so you can watch Pods being placed on and replaced across several nodes locally, with no cloud account or separate server. Apply the manifests from the free companion files in chapter order.
Create a kind cluster with one control plane and two workers (named cloudshop-en) and load the API and Web images you build into its nodes.
Pods, Deployments, Services, Ingress, ConfigMaps and Secrets, PVCs, probes and resource settings, and HPA are applied in turn on the same cluster. HPA is checked after installing metrics-server.
Jobs and CronJobs, StatefulSets, quota, LimitRange and RBAC, NetworkPolicy, and the Gateway API run on the kind cluster so you see what is allowed, what is blocked, and how state changes.
Diagnose representative failures from evidence, then build a fresh cluster, redeploy the whole of Mini CloudShop from scratch, and practice recovery.
The companion files include kind.yaml, which defines one control plane and two workers. The cluster name is cloudshop-en throughout, with context kind-cloudshop-en.
cat kind.yaml
# kind: Cluster
# apiVersion: kind.x-k8s.io/v1alpha4
# nodes:
# - role: control-plane
# - role: worker
# - role: worker
kind create cluster --name cloudshop-en --config kind.yaml --wait 5mBuild the images from the api and web folders in the companion files and hand them to the cluster with kind load, because a kind node cannot see the images on the host.
docker build -t intro-api:1.0 ./api
docker build -t intro-web:1.0 ./web
kind load docker-image intro-api:1.0 intro-web:1.0 --name cloudshop-enApply the files in the order of the chapter number at the start of each file name. Do not apply the whole folder at once; pick the files for each chapter.
kubectl apply -f k8s/07-deployment.yaml
kubectl apply -f k8s/09-service.yaml
kubectl apply -f k8s/11-config.yaml
kubectl apply -f k8s/12-redis-pvc.yaml
kubectl apply -f k8s/13-api-operational.yamlInstall Traefik with Helm. A local kind cluster has no external load balancer, so the Service stays ClusterIP and you connect with port-forward. Chapter 18 installs the Gateway API CRDs and turns on the Gateway provider of the same Traefik.
helm repo add traefik https://traefik.github.io/charts
helm upgrade --install traefik traefik/traefik --version 41.6.0 \
--namespace traefik --create-namespace \
--set providers.kubernetesIngress.enabled=true \
--set providers.kubernetesGateway.enabled=false \
--set service.spec.type=ClusterIP
kubectl port-forward -n traefik service/traefik 8080:80
# Chapter 18: Gateway API CRDs
kubectl apply --server-side -f https://github.com/kubernetes-sigs/gateway-api/releases/download/v1.5.1/standard-install.yamlOn a local cluster such as kind, metric collection can fail on kubelet certificate verification, so the --kubelet-insecure-tls option is added for local labs only.
kubectl apply -f https://github.com/kubernetes-sigs/metrics-server/releases/download/v0.8.0/components.yaml
kubectl patch deployment metrics-server -n kube-system --type='json' \
-p='[{"op":"add","path":"/spec/template/spec/containers/0/args/-","value":"--kubelet-insecure-tls"}]'
kubectl apply -f k8s/14-hpa.yamlThe final lab deploys the whole application from scratch on a separate, fresh cluster named cloudshop-final-en. Use the delete commands to clean up when you finish.
kind create cluster --name cloudshop-final-en --config kind.yaml --wait 5m
# Clean up
kind delete cluster --name cloudshop-final-en
kind delete cluster --name cloudshop-enThe lab environment recorded for this edition is Docker 29.2, kind 0.33, Kubernetes v1.37, Traefik Helm chart 41.6.0, and Gateway API v1.5.1, and all 20 live validation checks passed on it. Pod names, timestamps, and counter values differ per environment. kind is a local learning cluster, so production conditions such as an external load balancer or cloud storage are not covered.
Table of Contents
Chapters 1–6 establish the control model and local environment. Chapters 7–14 connect workloads, networking, configuration, storage, health, and scaling. Chapters 15–20 add finite tasks, StatefulSets, policy, Gateway API, diagnosis, and a fresh-cluster recovery lab. Kubernetes Learning Series, Volume 2.
Connect Docker experience to Kubernetes responsibilities and distinguish desired state from observed state.
Trace declarations through the control plane and follow user requests through the application.
Inspect contexts, namespaces, resources, logs, and Events before changing a cluster.
Create a three-node kind cluster and load locally built application images.
Observe Pod lifetime, container sharing, and what happens when an unmanaged Pod disappears.
Read API version, kind, metadata, and spec as a resource contract.
Follow Deployment and ReplicaSet reconciliation, scaling, rolling updates, and rollback.
Use namespaces, labels, and selectors deliberately and diagnose mismatched targets.
Connect stable Service names to changing Pods and inspect EndpointSlices.
Install Traefik, route Web and API requests, and distinguish rules from the controller.
Deliver ConfigMap and Secret values and verify when changed configuration reaches a process.
Compare temporary storage with PVC-backed Redis data and test persistence after replacement.
Set resource requests and limits; distinguish startup, readiness, and liveness probes.
Use measured CPU for HPA decisions and recover a failed application release.
Run bounded Jobs and CronJobs; inspect completion, retries, and failure limits.
Preserve StatefulSet identity and storage while understanding independent Redis members.
Verify quotas, default resources, RBAC permissions, and denied versus allowed network clients.
Replace the Ingress entry point with Gateway API and inspect route acceptance.
Diagnose scheduling, image, memory, and process failures with status, logs, and Events.
Rebuild the full application in a fresh cluster and practice data, release, and policy recovery.
A question-first command reference.
Follow evidence through a justified diagnosis and recovery.
Authoritative references for the lab tools and APIs.
Two additional questions for every chapter.
Companion Materials
Every companion/... path referenced throughout the book maps to the same path inside the zip downloadable from this page. The free ZIP contains 34 files: English API/Web sources, locked npm dependencies, a three-node kind configuration, staged manifests, diagnostic examples, and a learning record. Start with README.md and follow the chapter sequence.
The same files are also on GitHub, where the README covers the folder layout and a chapter-by-chapter lab guide: github.com/aidevops-books/kubernetes-understanding-practice-en
README.md
Archive layout and lab sequence.
api/, web/
API and Web source used to build the lab images.
k8s/
Progressive workload, Service, storage, policy, and Gateway manifests.
Publication Status
The English EPUB, 196-page Reading PDF, and 200-page print interior passed local publication checks. EPUBCheck reported zero errors or warnings, and all 20 live checks passed. Kindle/KDP account-side preview and publication remain pending. The free companion is available here.
AIDevOps Cloud Native Series
Learning Volume 1 introduces Docker and a first Kubernetes application; this is Volume 2. Continue to Kubernetes Production Engineering for production architecture and operations. Repository series numbers differ from learning-volume numbers.
AIDevOps Books · Learn. Build. Operate. Evolve.
← Back to all books