Create EKS cluster with Auto Mode and two-tier application

[pulumi]

Summary

This PR creates a complete EKS infrastructure with Auto Mode and deploys a two-tier containerized application.

Infrastructure Changes

VPC & Networking

• Fresh VPC: 10.0.0.0/16 CIDR block
• Multi-AZ: 3 availability zones with public and private subnets
• NAT Gateway: Single NAT gateway for cost optimization

EKS Cluster

• EKS Auto Mode: Enabled for automatic node management
• Version: Kubernetes 1.31
• Authentication: API mode for Auto Mode compatibility
• OIDC Provider: Enabled for IAM roles for service accounts
• Logging: All control plane logs enabled

Kubernetes Gateway API

• Gateway API CRDs: v1.2.1 installed
• AWS Load Balancer Controller: v1.11.0 with proper IAM roles
• Gateway: HTTP gateway for application routing
• HTTPRoute: Routes /api to backend and / to frontend

Application Components

Backend (Express.js)

• Image: Node.js 20 Alpine
• Endpoint: /api/env - Returns environment details (hostname, platform, CPU, memory, etc.)
• Health Check: /health endpoint
• Resources: 100m CPU request, 500m CPU limit, 128Mi-512Mi memory
• Replicas: 2 initial replicas
• HPA: Scales 2-10 replicas based on 70% CPU utilization

Frontend (Tailwind CSS)

• Image: Nginx Alpine
• UI: Tailwind CSS dashboard displaying backend environment details
• Features: Auto-refresh every 10 seconds, responsive design
• Resources: 50m CPU request, 200m CPU limit, 64Mi-256Mi memory
• Replicas: 2 replicas

Deployment

The GitHub Actions workflow is configured to:

1. On PR: Run pulumi preview to show planned changes
2. On merge to main: Run pulumi up to deploy the infrastructure

Preview Results

✅ Preview succeeded - 61 resources to create:

• VPC with subnets, route tables, NAT gateway, internet gateway
• EKS cluster with Auto Mode, IAM roles, OIDC provider
• Kubernetes resources: namespace, deployments, services, HPA, Gateway, HTTPRoute
• AWS Load Balancer Controller with IAM policy

Testing

After deployment:

1. Get the Gateway address: kubectl get gateway -n two-tier-app
2. Access the frontend at the Gateway's load balancer URL
3. The dashboard will display real-time environment details from the backend
4. Test autoscaling by generating CPU load on the backend pods

Configuration

• ESC Environment: aws-login_v2/pulumi-dev-sandbox-env
• AWS Region: us-west-2
• Stack: dev