teardown

Teardown — Remove AWS Resources

You are removing all AWS resources that were created during the deployment. This is irreversible.

Prerequisites

• AWS resources were previously deployed
• AWS CLI is configured and working

What To Do

Phase 1: Plan

Show the user exactly what will be deleted:

1. List all deployed stacks:

   cd infrastructure && npx cdk list
   

2. For each stack, explain what gets deleted:

   • "Lambda functions — your API will stop working"
   • "API Gateway — API endpoints will return errors"
   • "DynamoDB table — your data will be..."
     • If RemovalPolicy.RETAIN: "...preserved (you can delete it manually later)"
     • If RemovalPolicy.DESTROY: "...permanently deleted"
   • "S3 bucket — your frontend files will be deleted"
   • "CloudFront distribution — your site URL will stop working"
   • "Cognito User Pool — all user accounts will be..."
     • If RemovalPolicy.RETAIN: "...preserved"
     • If RemovalPolicy.DESTROY: "...permanently deleted"

3. Highlight data implications:

   • "DynamoDB tables with RETAIN policy will survive and need manual cleanup"
   • "All other resources will be permanently deleted"
   • "This cannot be undone"

Phase 2: Confirm

Get explicit confirmation:

"This will permanently delete the following AWS resources:"

• [list each resource]
• "Any data in DynamoDB tables with DESTROY policy will be lost"
• "Your site will go offline immediately"

"Are you absolutely sure you want to proceed? Type 'yes' to confirm."

Wait for explicit confirmation. Do NOT proceed without it.

Phase 3: Execute

1. Empty S3 buckets first (CDK can't delete non-empty buckets):

   aws s3 rm s3://[bucket-name] --recursive
   

   Explain: "Emptying the S3 bucket before deleting the stack"

2. Destroy all stacks:

   cd infrastructure && npx cdk destroy --all --force
   

   Stream progress and explain what's being removed

3. Handle errors:

   • If destruction fails, explain the error and suggest fixes
   • Common issues: non-empty S3 buckets, resources in use, permission errors
   • Some resources may need manual deletion

Phase 4: Post-Teardown

1. Verify deletion:

   cd infrastructure && npx cdk list
   

   Should return empty or error (no stacks)

2. Check for retained resources:

   • DynamoDB tables with RETAIN policy
   • CloudWatch log groups (CDK often doesn't delete these)
   • CDK bootstrap resources (usually fine to keep)

3. Guide manual cleanup if needed:

   • "These resources were retained and may need manual cleanup:"
   • Provide AWS Console URLs or CLI commands for each
   • Explain how to delete them if the user wants to

4. Clean up local files:

   • Ask if user wants to remove .migration/outputs.json
   • Update .migration/plan.md to reflect teardown

5. Final message:

   • "All resources have been removed. Your AWS account should not incur any further charges from this deployment."
   • "The CDK bootstrap resources are still in your account — they cost virtually nothing and can be reused for future deployments."
   • "Your code is still here locally — you can redeploy anytime"

Important Notes

Safety

• ALWAYS confirm before destroying
• List every resource that will be deleted
• Highlight data loss implications
• Check for retained resources after deletion

Common Issues

• S3 buckets must be emptied before the stack can be deleted
• CloudFront distributions take 5-15 minutes to delete
• If a stack is in ROLLBACK_COMPLETE state, it needs aws cloudformation delete-stack
• Log groups may remain — they cost nothing but can be cleaned up

What's NOT Deleted

• CDK bootstrap resources (CDKToolkit stack) — keep these for future deployments
• CloudWatch log groups — minimal cost, can be manually deleted
• Any resources with RemovalPolicy.RETAIN