Summary

Still a work in progress.

This repository is a sample with terraform + typescript on how to create a lambda function that receives HTTP requests from an API Gateway and connects safely with a database created with RDS and safely keeps its connection alive between function calls.

This repository does not worry about secrets privacy, since it doesn't version the terraform state.

This project does not use RDS Proxy, even though it is a great alternative for managing the connection pool for larger applications.

What has been done already

• Terraform creation of the following resources
  • API Gateway
  • Lambda function
  • S3 bucket to store the code of the function
• Automated build and deployment of lambda function on each terraform apply
• Manage VPC + private subnets for RDS
• The RDS instance itself
• Lambda environment variables (needed for database access)
  • Currently living inside .env file
  • Database hostname is generated after first terraform apply
• CI/CD to deploy on push to master

What still needs to be done

• Run migrations when lambda is deployed
• Create seed data for one table
• Create API to retrieve given seed data

How to use

Requirements

• node v12+
• yarn
• terraform
• A configured AWS profile with admin access
• Docker

Running locally

$ cp .env.example .env
$ docker-compose up --build

Deploying to AWS

$ touch infrastructure/terraform.tfvars
# Fill in the dbname, dbuser and dbpassword variables
$ terraform apply --auto-approve
# Get database endpoint from output
$ cp .env.example .env
# Fill in the data for .env file (database hostname will be generated after first apply and will be in the output)
$ terraform apply --auto-approve # To update the lambda environment variables

Consuming API

With the terraform apply, you'll get the invoke url output. Use that url to consume the API with

$ curl <invoke_url>/api/health

It should output OK