Skip to content

DEPLOYMENT.md

Latest commit

 

History

History
268 lines (192 loc) · 6.19 KB

DEPLOYMENT.md

File metadata and controls

268 lines (192 loc) · 6.19 KB
Raw

DocuMind Deployment & Teardown Guide

This guide covers deploying DocuMind to AWS and tearing down all resources.

Table of Contents

Prerequisites

Before deploying, ensure you have:

  1. AWS CLI configured with credentials

    aws configure

  2. AWS CDK installed globally

    npm install -g aws-cdk

  3. Docker running (for building container images)

  4. Node.js 20+ and Java 21+ installed

  5. OpenAI API Key ready

Deploy to AWS

Step 1: Bootstrap CDK (First Time Only)

cd infra
npm install
npx cdk bootstrap

Step 2: Deploy Infrastructure

cd infra
npx cdk deploy

This creates:

  • VPC with public/private subnets
  • ECS Fargate cluster with backend service
  • RDS PostgreSQL database with pgvector
  • S3 buckets (frontend assets + documents)
  • CloudFront CDN
  • Secrets Manager secrets
  • Application Load Balancer

Deployment takes ~15-20 minutes.

Step 3: Note the Outputs

After deployment, CDK outputs important values:

Outputs:
DocuMindStack.CloudFrontUrl = https://xxxxxx.cloudfront.net
DocuMindStack.ApiUrl = http://DocuMi-Backe-xxxxx.us-east-1.elb.amazonaws.com
DocuMindStack.FrontendBucketName = documind-frontend-xxxxx
DocuMindStack.DocumentsBucketName = documind-documents-xxxxx
DocuMindStack.DatabaseEndpoint = documindstack-xxxxx.rds.amazonaws.com
DocuMindStack.OpenAiSecretArn = arn:aws:secretsmanager:...

Post-Deployment Steps

Step 1: Configure OpenAI API Key

aws secretsmanager put-secret-value \
  --secret-id documind/openai-api-key \
  --secret-string '{"apiKey":"sk-your-actual-openai-key"}'

Step 2: Restart ECS Service (to pick up the secret)

aws ecs update-service \
  --cluster documind-cluster \
  --service documind-api \
  --force-new-deployment

Step 3: Build and Deploy Frontend

cd frontend
npm install
npm run build

Upload to S3 (replace with your bucket name from outputs):

aws s3 sync dist/ s3://documind-frontend-YOUR_ACCOUNT_ID-YOUR_REGION/ --delete

Step 4: Invalidate CloudFront Cache

Get your CloudFront distribution ID:

aws cloudfront list-distributions --query "DistributionList.Items[*].{Id:Id,Domain:DomainName}" --output table

Invalidate cache:

aws cloudfront create-invalidation --distribution-id YOUR_DIST_ID --paths "/*"

Step 5: Access Your Application

Open the CloudFront URL from the deployment outputs.

Teardown (Delete All Resources)

⚠️ WARNING: This will permanently delete all data including uploaded documents and database content.

Step 1: Verify Resources to Delete

cd infra
npx cdk diff

Step 2: Destroy All Resources

cd infra
npx cdk destroy

Type y when prompted to confirm deletion.

Destruction takes ~10-15 minutes.

Step 3: Verify Cleanup

Check that resources are deleted:

# Check S3 buckets
aws s3 ls | grep documind

# Check ECS clusters
aws ecs list-clusters | grep documind

# Check RDS instances
aws rds describe-db-instances --query "DBInstances[*].DBInstanceIdentifier" | grep documind

# Check CloudFront distributions
aws cloudfront list-distributions --query "DistributionList.Items[*].DomainName"

# Check Secrets Manager
aws secretsmanager list-secrets --query "SecretList[*].Name" | grep documind

Step 4: Manual Cleanup (If Needed)

If any resources remain:

Delete S3 buckets:

aws s3 rb s3://documind-frontend-ACCOUNT-REGION --force
aws s3 rb s3://documind-documents-ACCOUNT-REGION --force

Delete Secrets:

aws secretsmanager delete-secret --secret-id documind/openai-api-key --force-delete-without-recovery
aws secretsmanager delete-secret --secret-id documind/db-credentials --force-delete-without-recovery

Delete CloudWatch Log Groups:

aws logs delete-log-group --log-group-name /ecs/documind-api

Step 5: Clean CDK Bootstrap (Optional)

If you don't plan to use CDK again in this account/region:

# Delete the CDKToolkit CloudFormation stack
aws cloudformation delete-stack --stack-name CDKToolkit

Cost Information

Monthly Costs (Approximate)

Service Configuration Cost
ECS Fargate 0.5 vCPU, 1GB RAM ~$15-20
RDS PostgreSQL db.t3.micro, 20GB ~$15-20
NAT Gateway Data processing ~$30-40
ALB Basic usage ~$20
CloudFront Low traffic ~$1-5
S3 Minimal storage ~$1-2
Secrets Manager 2 secrets ~$1
Total ~$85-115/month

Cost Optimization Tips

  1. NAT Gateway is the biggest cost - consider VPC endpoints for production
  2. Stop ECS service when not in use: aws ecs update-service --cluster documind-cluster --service documind-api --desired-count 0
  3. Use Fargate Spot for non-production workloads

Redeployment Checklist

When redeploying in the future:

  • AWS CLI configured (aws configure)
  • Docker running
  • CDK installed (npm install -g aws-cdk)
  • Run cd infra && npm install
  • Run npx cdk bootstrap (if new account/region)
  • Run npx cdk deploy
  • Configure OpenAI API key in Secrets Manager
  • Restart ECS service
  • Build and deploy frontend to S3
  • Invalidate CloudFront cache
  • Test application

Files Required for Redeployment

All infrastructure is captured in code:

infra/
├── bin/infra.ts          # CDK app entry point
├── lib/documind-stack.ts # All AWS resources defined here
├── package.json          # CDK dependencies
├── tsconfig.json         # TypeScript config
└── cdk.json              # CDK config

backend/
├── Dockerfile            # Container build instructions
├── pom.xml               # Java dependencies
└── src/                  # Application code

frontend/
├── package.json          # Frontend dependencies
└── src/                  # React application

No manual AWS console configuration is required - everything is Infrastructure as Code.