Skip to content

AWS-DEPLOYMENT.md

Latest commit

 

History

History
226 lines (169 loc) · 4.92 KB

AWS-DEPLOYMENT.md

File metadata and controls

226 lines (169 loc) · 4.92 KB

AWS Lambda Deployment Guide for RAG Backend

Prerequisites

  1. Docker Desktop installed and running

  2. AWS CLI installed ✅ (Already installed)

    aws --version

  3. AWS Account with credentials configured

    aws configure

    You'll need:

    • AWS Access Key ID
    • AWS Secret Access Key
    • Default region (e.g., us-east-1)
    • Output format: json

  4. Environment Variables (from .env file)

    • GROQ_API_KEY
    • MONGODB_URI

Deployment Steps

Option 1: PowerShell (Windows) - Recommended

cd backend
./deploy-lambda.ps1

Option 2: Bash (Git Bash/WSL)

cd backend
chmod +x deploy-lambda.sh
./deploy-lambda.sh

What the Script Does

  1. ✅ Creates ECR (Elastic Container Registry) repository
  2. ✅ Authenticates Docker with AWS
  3. ✅ Builds Docker image with optimizations
  4. ✅ Pushes image to ECR
  5. ✅ Creates IAM role for Lambda
  6. ✅ Creates/Updates Lambda function
  7. ✅ Sets up Function URL (no API Gateway needed!)
  8. ✅ Configures CORS for frontend

After Deployment

You'll get a Function URL like:

https://xxxxxxxxxxxxxx.lambda-url.us-east-1.on.aws/

Test Your API

# Health check
curl https://your-function-url.lambda-url.us-east-1.on.aws/api/health

# Root endpoint
curl https://your-function-url.lambda-url.us-east-1.on.aws/

Update Frontend

  1. Open frontend/.env or create it:

    VITE_API_URL=https://your-function-url.lambda-url.us-east-1.on.aws

  2. Rebuild frontend:

    cd frontend
npm run build

Cost Estimate (Free Tier)

AWS Lambda Free Tier includes:

  • 1M requests per month - FREE
  • 400,000 GB-seconds of compute time - FREE

Your configuration:

  • Memory: 2048 MB (2 GB)
  • Timeout: 900 seconds (15 min)
  • Free tier gives: 400,000 / 2 = 200,000 seconds = ~55 hours of execution time

Estimated free usage:

  • ~2,200 requests/month at 900s each = FREE
  • Or ~220,000 requests at 90s each = FREE

After free tier: ~$0.0000166667 per GB-second

Important Lambda Limits

⚠️ Size Limits:

  • Docker image: 10 GB (should be fine)
  • Unzipped: 250 MB (might be tight with ML models)
  • Ephemeral storage: 512 MB - 10 GB

⚠️ Cold Start:

  • First request after idle: 5-10 seconds
  • Container image cold start: slower than ZIP

⚠️ Memory:

  • Set to 2048 MB (2 GB) for ML models
  • Increase if OOM errors occur

Troubleshooting

1. Docker not found

Error: docker: command not found Fix: Install Docker Desktop and ensure it's running

2. AWS credentials not configured

Error: Unable to locate credentials Fix:

aws configure
# Enter your AWS Access Key ID and Secret

3. Image too large

Error: RequestEntityTooLargeException Fix: The sentence-transformers model might be too heavy. Consider:

  • Using Groq embeddings API instead
  • Switching to a smaller model
  • Using ECS/Fargate instead of Lambda

4. Function timeout

Error: Task timed out after 900.00 seconds Fix: Increase timeout (max 15 min) or optimize code

5. Out of memory

Error: Runtime exited with error: signal: killed Fix: Increase memory in script (currently 2048 MB, max 10240 MB)

Update Deployment

To update after code changes:

cd backend
./deploy-lambda.ps1  # or ./deploy-lambda.sh

The script will:

  • Rebuild Docker image
  • Push new version to ECR
  • Update Lambda function automatically

Delete Resources

To remove all AWS resources:

# Delete Lambda function
aws lambda delete-function --function-name rag-backend-api --region us-east-1

# Delete ECR repository
aws ecr delete-repository --repository-name rag-backend --force --region us-east-1

# Delete IAM role (detach policies first)
aws iam detach-role-policy --role-name rag-lambda-execution-role --policy-arn arn:aws:iam::aws:policy/service-role/AWSLambdaBasicExecutionRole
aws iam delete-role --role-name rag-lambda-execution-role

Alternative: Deploy to ECS (If Lambda fails)

If the Docker image is too large or cold starts are problematic, use ECS:

# Will create separate ECS deployment script if needed

Frontend Deployment (Free Options)

Vercel (Recommended)

cd frontend
npm install -g vercel
vercel

Netlify

cd frontend
npm run build
npm install -g netlify-cli
netlify deploy

Monitoring

View logs in AWS CloudWatch:

aws logs tail /aws/lambda/rag-backend-api --follow --region us-east-1

Or in AWS Console: CloudWatch → Log groups → /aws/lambda/rag-backend-api

Support

If deployment fails, check:

  1. Docker is running
  2. AWS credentials are valid: aws sts get-caller-identity
  3. Region is correct (default: us-east-1)
  4. No firewall blocking Docker/AWS

Next Steps:

  1. Run deployment script
  2. Test the Function URL
  3. Update frontend with API URL
  4. Deploy frontend to Vercel
  5. Share your app! 🚀