🐳 Docker Playground

Hands-on Docker examples for real-world patterns — networking, proxying, resilience, security, observability, and more. Each project is self-contained and runnable with a single command.

📦 Projects

🌐 Web & Proxy

Project	What it demonstrates
🔀 Traefik Reverse Proxy	Path-based routing (/service-a, /service-b) with Traefik using a static YAML file provider. No Docker socket required.
⚡ Traefik Circuit Breaker	Circuit breaker middleware that automatically stops forwarding requests to a failing backend and recovers after a timeout.
⚖️ HAProxy Load Balancer	Round-robin load balancing across multiple HTTP backends with HAProxy.
📄 Nginx Webserver	Static file server with gzip compression, security headers, cache control, custom error pages, and structured access logs.
🧩 Varnish ESI	Edge Side Includes — assembling HTML pages from cached components using Varnish.
🚦 Queue-Based Site Access	Virtual waiting room in front of a protected page — Redis-backed FIFO queue, live position polling, ETA, atomic slot claiming, and a bypass-proof architecture where the target has no published port.

🗄️ Data

Project	What it demonstrates
🍃 MongoDB Prefilled	MongoDB container that seeds a collection from a JSON file on startup.
🐘 PostgreSQL Read Replicas	Primary + two read replicas with streaming replication — ready to test read scaling.

🔒 Security

Project	What it demonstrates
🛡️ Secure Docker Container	Multi-stage build, non-root user, read-only filesystem, dropped capabilities — security hardening checklist in one Dockerfile.
🔑 Build-Time Secret Handover	RUN --mount=type=secret vs --build-arg — why secrets passed as build args end up in the image history and how to avoid it.
🏦 API Keycloak Security	Go banking REST API secured with Keycloak (OIDC login, roles) and JWE token encryption — access tokens are RSA-encrypted so only the API can read the claims.

📡 Messaging & Observability

Project	What it demonstrates
🐇 RabbitMQ Producer / Consumer	Work queue pattern with two Python services — one publishing messages, one consuming them.
📊 Spring Boot + OpenTelemetry + Elastic APM	Distributed tracing from a Spring Boot app through OpenTelemetry to Elastic APM and Kibana.

☸️ Kubernetes

Project	What it demonstrates
⎈ Local Kubernetes + ArgoCD	Fully local GitOps environment with kind, ArgoCD, and a local Git server — no cloud account needed.

---

🚀 Quick start

Every project has a Makefile. The common entry point is always:

make        # show available targets
make demo   # or: make up

---

🔍 Checkout a single project

You don't need to clone the entire repository. Use Git sparse checkout to get only the project you want:

git clone --no-checkout --depth 1 git@github.com:messeb/docker-playground.git
cd docker-playground
git sparse-checkout init --cone
git sparse-checkout set traefik-circuit-breaker
git checkout main

Replace traefik-circuit-breaker with any project folder name. To add more projects later:

git sparse-checkout add nginx-webserver

---

💡 Tips & Tricks

Port conflicts — each project uses its own ports. Check the compose.yml before running two projects at the same time. Common ports used: 80, 8080, 5672, 15672, 5601.

Named networks — projects that use Traefik define a named Docker network (proxy, cb-net, …). Run docker network ls if you see unexpected network warnings after make down.

Rebuilding images — if you change application code, pass --build to rebuild:

docker compose up --build

Or use make clean (where available) to remove cached images entirely.

Following logs — attach to a specific service without restarting it:

docker compose logs -f <service-name>

Inspecting a container — open a shell inside a running container:

docker compose exec <service-name> sh

Stopping cleanly — make down (or docker compose down) stops and removes containers but keeps images. Add --rmi local to also remove locally built images, or --volumes to wipe named volumes (e.g. database data).