EKS Experiments #2: Handling Good & Bad Deployments with GitOps

Описание: 🧪 EKS Experiments #2: How Kubernetes and GitOps Handles Good and Bad Deployments

GitOps isn’t just a buzzword; it's a safer, repeatable way to ship to Kubernetes. In this video I walk through how GitOps works in practice using Argo CD + GitHub Actions on an AWS EKS cluster. Then I stress the system with k6 and run three live deployment experiments to see how Kubernetes behaves under load:
✅ Good deployment → rolls out cleanly while k6 is hammering the API.
✅ Bad deployment (1 replica) → the site goes down.
✅ Bad deployment (2 replicas) → the site stays up but drops load while the bad pod rolls in.

You’ll see how desired state in Git drives cluster changes, what Argo CD reconciles, and how replica count directly affects resilience during failures.

💡 You’ll learn:
GitOps fundamentals (PR-driven deploys, desired vs. current state, reconciliation loops)
How Argo CD applies and continuously enforces what’s in Git
GitHub Actions pipeline from commit → build → deploy
Why replicas (and readiness/liveness probes) matter for availability under failure

📂 GitHub Repos:

Infrastructure as Code (Terraform): https://github.com/robertdippolito/ek...

API (Go CRUD): https://github.com/robertdippolito/go...

🌐 More from me:
Blog: https://robertdippolito.me

Discord:   / discord  

💡 If this helped you, please like the video, subscribe, and drop a comment with questions or suggestions for future experiments.

▶️ Watch the next video in the EKS Experiments series here:    • EKS Experiments #1: How Kubernetes Handles...  

Video Breakdown

0:45 Architecture
1:30 What is GitOps
3:02 How it works
3:42 Good Deployment Experiment
12:05 Bad Deployment - 1 Replica Experiment
20:37 Bad Deployment - 2 Replicas Experiment