alphagov / rails-lambda-spike

Rails and AWS Lambda Spike

This git repository demonstrates a spike into running a simple Ruby on Rails application on AWS Lambda

AWS SAM is used to package and deploy the application. It is scripted in package-and-deploy.sh and uses template.yml as the CloudFormation template.

The application is deployed in the following arrangement:

API Gateway receives in coming requests and routes them to the RailsLambda Function.

RailsLambda uses lambda.rb to start the Rails application and handle events passed from the API Gateway. The #handle method translates the AWS Lambda event messages into the format required by Rack. It is based on the lambda.rb example from the sinatra demo, but some additional changes have been made to make Rails happy.

The Rails application communicates with a RDS Postgres database. This requires that the Lambda is deployed into the same VPC, and subnets, as the database. The Lambda also needs an IAM role that has permission to execute inside a VPC. All of this is configured using the CloudFormation template.

A second lambda function is also configured that can be used to run database schema migrations. When invoked the lambda function will run the db:create and db:migrate rake tasks. This behaviour is defined in lambda.rb with the #migration_handler method.