podinfo-js

Small Cloud-native nodejs project inspired by Stefan Prodan's podinfo template

About

This is a bit of a refresher/learning project that shows a tiny nodejs app integrated with Docker/Kubernetes. Intentionally similar to the excellent podinfo project/template, just written in NodeJS. In my mind the podinfo serves as the prototypical example of building a cloud-native application, so I figured it would act as a solid starting point. Any code reuse will be attributed appropriately.

Run (local)

Docker

1. From this directory, docker build . -t podinfo-js:latest
2. docker run --rm -p 3000:3000 -it podinfo-js:latest

Node

1. npm install
2. npm run dev

Testing (local)

Node

Run npm test (see scripts in package.json)

Github Actions

Use act: act -W ./github/workflows/ci.yml

Routes

Routes are documented below

Route Path	HTTP Method	Component / File	Purpose	Parameters / Notes
/	GET	Info (src/routes/info.ts)	Returns application runtime information such as hostname, version, uptime, node version, a custom message, and list of mounted secrets.	None
/healthz	GET	Health (src/routes/health.ts)	Cheap liveness probe to verify that the application process is running.	Returns 200 OK with { status: 'ok' }.
/readyz	GET	Health (src/routes/health.ts)	Readiness probe for load balancers. Fails gracefully during graceful shutdowns.	Returns 200 OK with { status: 'ready' } normally, or 503 Service Unavailable with { status: 'shutting down' } when shutting down.
/error	GET	Chaos (src/routes/chaos.ts)	Returns an intentional internal server error to test error handling, alerts, or retries.	Returns 500 Internal Server Error with { error: 'on purpose' }.
/delay	GET	Chaos (src/routes/chaos.ts)	Simulates network lag or slow responses for testing timeouts and drain behaviors.	Query parameter: ms (optional, duration in ms; defaults to 1000, capped at 30000).
/cpu	GET	Chaos (src/routes/chaos.ts)	Intentionally blocks the JavaScript event loop in a tight loop to simulate high CPU load. Used to test Horizontal Pod Autoscaling (HPA).	Query parameter: ms (optional, duration in ms; defaults to 1000, capped at 10000).
/panic	GET	Chaos (src/routes/chaos.ts)	Forces the application server process to crash. Useful for verifying automated container orchestration restarts.	Returns 500 with { status: 'panicking' } and triggers process.exit(1).
/ui/*	GET	UI (src/app.ts)	Serves the static frontend assets from the public/ directory (e.g., public/index.html).	Served via @fastify/static.

CI/CD

The project will be deployed and managed several different ways across multiple Kubernetes environments as a comparative learning exercise. The idea is that the same workload will be targeting a variety of different platforms.

Target Environments

I have access to the following environments

• AWS EKS
• Oracle OKE
• Homelab (mini-pc's running proxmox+k3s)

GitOps Platforms

The two dominant pull-based CD platforms in the CNCF landscape. Both watch Git and reconcile cluster state.

• ArgoCD — declarative GitOps with a UI, ApplicationSet templating, and broad ecosystem support.
• FluxCD v2 — CLI-first GitOps platform with strong multi-tenancy primitives and Helm-native flow.
  • Flux AIO — single-pod Flux distribution for resource-constrained environments (homelab fit).

Progressive Delivery

Sit on top of either GitOps platform. They manage Kubernetes CRDs that any reconciler can sync, but each has a cultural home in one ecosystem.

• Argo Rollouts — blue-green and canary deployments via the Rollout CRD. From the Argo ecosystem.
• Flagger — metric-driven progressive delivery and automated rollback. From the Flux ecosystem.

Promotion / Environment Management

• Kargo — orchestrates artifact promotion across environments (dev → staging → prod) on top of GitOps. Complements rather than replaces ArgoCD/Flux.

Traditional CI/CD (Baseline)

• GitHub Actions — vanilla push-based kubectl apply deployment, as a comparative baseline to see what we gain with dedicated tooling

Acknowledgments

This project is inspired by podinfo by Stefan Prodan, reimagined in Node.js/TypeScript as a learning exercise.