aws-infrastructure

name	aws-infrastructure
description	AWS infrastructure patterns for this monorepo. Infrastructure is managed in a separate turbo-infrastructure repo via Terraform modules. This app repo only contains deployment workflows (GitHub Actions) and a docker-compose for staging. Use when deploying to AWS, configuring CI/CD, or setting up GitHub environment variables.

AWS Infrastructure

Architecture

Infrastructure is split across two repositories:

Repository	Purpose
turbo-template (this repo)	App code + CI/CD workflows + docker-compose.staging.yml
turbo-infrastructure	Terraform modules (VPC, ECS, ALB, ECR, EC2, monitoring)

Deployment Strategies

Environment	Branch	Strategy	Infrastructure
Staging	staging	EC2 + Docker Compose + host Nginx	Single EC2 instance
Production	production	ECS Fargate + ALB	Auto-managed containers

Architecture Diagram

STAGING:
  Internet → Nginx (EC2 host) → Docker Compose
                                   ├── Web (port 3001)
                                   └── Backend (port 3000)

PRODUCTION:
  Internet → ALB (port 80/443)
               ├── / → Web Target Group (port 3001)
               └── /api/* → Backend Target Group (port 3000)
                      ↓
                ECS Fargate (private subnets)
                ├── Web Service (512 CPU, 1024 MB)
                └── Backend Service (512 CPU, 1024 MB)

App Repo Structure

.
├── docker-compose.staging.yml   # Docker Compose for staging EC2
├── docker-compose.yml           # Docker Compose for local dev
└── .github/workflows/
    ├── ci.yml                   # CI checks (lint, typecheck, format)
    ├── deploy-staging.yml       # Staging deploy (EC2)
    └── deploy-production.yml    # Production deploy (ECS Fargate)

All infrastructure templates (Terraform, nginx configs, task definitions) have been moved to the infra repo or inlined into CI/CD workflows. Developers only need to:

1. Set GitHub environment variables (see Deployment section in README.md)
2. Merge to staging or production branch

CI/CD Workflows

Staging (deploy-staging.yml)

1. CI passes (lint, typecheck, build)
2. Build Docker images → push to ECR
3. SSH into EC2 → copy docker-compose + generate nginx configs
4. docker compose up → configure nginx → health checks
5. Auto-rollback on health check failure

Nginx configs are generated inline in the workflow using heredocs. SSL detection is automatic — if Let's Encrypt certs exist, HTTPS configs are used.

Production (deploy-production.yml)

1. CI passes
2. Build Docker images → push to ECR
3. Register ECS task definitions (generated inline via heredocs)
4. Deploy to ECS services → wait for stabilization
5. Auto-rollback to previous task definition on failure

Task definitions use ECS_EXECUTION_ROLE_ARN and ECS_TASK_ROLE_ARN from GitHub variables (sourced from Terraform outputs).

ECS Task Definition Pattern

Task definitions are constructed inline in the production workflow:

{
	"family": "${PROJECT_NAME}-web-production",
	"cpu": "512",
	"memory": "1024",
	"networkMode": "awsvpc",
	"requiresCompatibilities": ["FARGATE"],
	"executionRoleArn": "${ECS_EXECUTION_ROLE_ARN}",
	"taskRoleArn": "${ECS_TASK_ROLE_ARN}",
	"containerDefinitions": [
		{
			"name": "web",
			"image": "${ECR_REGISTRY}/${ECR_REPOSITORY_WEB}:${IMAGE_TAG}",
			"portMappings": [{ "containerPort": 3001 }],
			"environment": [
				{ "name": "NODE_ENV", "value": "production" },
				{ "name": "NEXT_PUBLIC_APP_URL", "value": "..." },
				{ "name": "NEXT_PUBLIC_API_BASE_URL", "value": "..." }
			],
			"healthCheck": {
				"command": ["CMD-SHELL", "wget --spider http://localhost:3001/ || exit 1"]
			},
			"logConfiguration": {
				"logDriver": "awslogs",
				"options": {
					"awslogs-group": "/ecs/${PROJECT_NAME}-web-production",
					"awslogs-region": "${AWS_REGION}"
				}
			}
		}
	]
}

GitHub Environment Variables

See the Deployment section in README.md for the complete list. Key additions for production:

Variable	Source
ECS_EXECUTION_ROLE_ARN	terraform output ecs_execution_role_arn
ECS_TASK_ROLE_ARN	terraform output ecs_task_role_arn
ECS_CLUSTER	terraform output ecs_cluster_name
ECS_SERVICE_WEB	terraform output ecs_service_names
ECS_SERVICE_BACKEND	terraform output ecs_service_names

Troubleshooting

1. Service not starting → Check CloudWatch logs: /ecs/{PROJECT_NAME}-{service}-production
2. Health check failing → Verify endpoint responds: /api/v1/health (backend), / (web)
3. Image pull errors → Check ECR URI matches, verify execution role has ecr:GetDownloadUrlForLayer
4. Permission denied → Verify task execution role has ECR pull + CloudWatch policies
5. 503 errors → Check target group health, security group allows ALB → ECS
6. Env vars missing → Verify GitHub environment variables/secrets are set
7. Staging nginx errors → SSH to EC2, run sudo nginx -t, check /etc/nginx/conf.d/*.conf