Express to AWS Lambda: Apex Edition (Part 2)

Terraform

Published in

Serverless Zone

5 min readJul 4, 2016

In Part 1 of this series we looked at how to get started with AWS Lambda and the Apex Framework. So far we have built, deployed and tested an AWS Lambda function. To finish converting our Express app from Part 1, we need to process HTTP requests. Therefore, the next step is to create an API Gateway using Terraform.

Example Express App

Our Express app from Part 1 had one GET endpoint. This endpoint accepted a single parameter :id.

app.get('/foo/:id', function(req, res) {
  var id = req.params.id;
  foo
    .get(id)
    .then(function(foo) {
      res.status(200).send(foo);
    })
    .catch(function (err) {
      res.status(500).send();
    });
});

Getting Started with Terraform

Terraform is a tool used to define infrastructure as code for various infrastructure providers. Apex has integrated Terraform into the framework through the apex infra command. This command wraps the terraform plan and terraform apply commands.

The infrastructure for the project is defined in the infrastructure directory.

infrastructure/
├── modules       // common modules
  ├── api_gateway
  │ └── main.tf
  ├── iam
  │ └── main.tf 
├── prod         // prod stage
│ └── main.tf

Terraform has created a custom syntax for configuration files. The syntax is very human readable and easy to understand. Terraform has implemented Variables and Modules to allow you to re-use parts of your configuration. While Terraform is easy to use, the difficult part is correctly configuring the resources you require.

API Gateway

AWS Lambda is a compute service. To process HTTP requests, we need to use another AWS service. API Gateway allows you to create HTTP endpoints as a service. The Gateway accepts requests and passes those requests to a back-end for processing. In this example, we need to configure an API Gateway resource to pass requests to Lambda and return the responses to the client.

The resource block is a core part of all Terraform configurations. A resource is a piece of infrastructure such as an EC2 instance, or in our case an API Gateway. For this example, I am going to focus on the resources needed to create an API Gateway endpoint. I suggest reading the Terraform Docs, if you want to learn more about the other block types.

Configuration
The first Terraform resource we need to define is an aws_api_gateway_rest_api. This resource defines the API Gateway. All the other Terraform resources will be children of this resource.

resource "aws_api_gateway_rest_api" "foo_api" {
 name        = "foo_api"
 description = “This is my api gateway”
}

The next step is to define our API Gateway resources using a aws_api_gateway_resource. The terminology becomes confusing here as we now have Terraform resources and API Gateway resources. I like to think of an API Gateway resource as part of a URL e.g. /foo.

In our example, we need to expose a GET endpoint at /foo/{id}.

# Resource /foo
resource "aws_api_gateway_resource" "foo_resource {
 rest_api_id = "${aws_api_gateway_rest_api.foo_api.id}"
 parent_id = "${aws_api_gateway_rest_api.foo_api.root_resource_id}"
 path_part = "foo"
}# Resource /foo/{id}
resource "aws_api_gateway_resource" "foo_id_resource" {
 rest_api_id = "${aws_api_gateway_rest_api.foo_api.id}"
 parent_id = "${aws_api_gateway_resource.foo_resource.id}"
 path_part = "{id}"
}

Each aws_api_gateway_resource references the parent resource and a path part. The foo resource’s parent is the root_resource of the API and the {id} resource’s parent is the foo resource.

Now that we have defined our API Gateway resources, we need to add a GET method to /foo/{id}.

# Method
resource "aws_api_gateway_method" "foo_get_endpoint_method" {
 rest_api_id = "${aws_api_gateway_rest_api.foo_api.id}"
 resource_id = "${aws_api_gateway_resource.foo_id_resource.id}"
 http_method = "GET"
 authorization = "NONE"
}

The next step is to define a back-end for our GET request. This is where we connect our Lambda function to our API Gateway.

# Integration
resource "aws_api_gateway_integration" "foo_get_endpoint_integration" {
 rest_api_id = "${aws_api_gateway_rest_api.foo_api.id}"
 resource_id = "${aws_api_gateway_resource.foo_id_resource.id}"
 type = "AWS" http_method = "${aws_api_gateway_method.foo_get_endpoint_method.http_method}"
 
 credentials = "${var.gateway_invoke_lambda_role_arn}" # Must be POST for invoking Lambda function
 integration_http_method = "POST" uri = "arn:aws:apigateway:${var.aws_region}:lambda:path/2015–03–31/functions/arn:aws:lambda:${var.aws_region}:${var.aws_account_id}:function:express-to-aws-lambda_foo/invocations" request_templates = { // Optional
   "application/json" = 
"${file("${path.module}/api_gateway_request_mapping.template")}"
   }
}

The important parts of the configuration are the uri and request_template. The uri points to the Amazon Resource Name (ARN) of our Lambda function. The request_templates property allows us to configure API Gateway request mappings. If you are new to API Gateway, I strongly suggest that you read the docs on request mappings.

Half Time
This is the halfway mark for our request. We have configured the API Gateway to accept a GET request and pass it to our Lambda function. There was a lot to configure to get to this point. I promise the next half is easier.

The next step is to pass the response from our Lambda function back to the client. To do so, we need to define an aws_api_gateway_integration_response and aws_api_gateway_method_response.

# Integration Response
resource 
 "aws_api_gateway_integration_response"   
 "foo_get_endpoint_integration_response" { rest_api_id = "${aws_api_gateway_rest_api.foo_api.id}"
 resource_id = "${aws_api_gateway_resource.foo_id_resource.id}"
 
 http_method = "${aws_api_gateway_method.foo_get_endpoint_method.http_method}"
 
 status_code = "${aws_api_gateway_method_response.foo_get_endpoint_method200.status_code}"
 
 response_parameters_in_json = <<PARAMS
 {
   "method.response.header.X-Header": "integration.response.body.json_path"
 }
PARAMS
}# Method 200 Response
resource
 "aws_api_gateway_method_response"
 "foo_get_endpoint_method200" { rest_api_id = "${aws_api_gateway_rest_api.foo_api.id}"
 resource_id = "${aws_api_gateway_resource.foo_id_resource.id}"
 
http_method = "${aws_api_gateway_method.foo_get_endpoint_method.http_method}" status_code = "200" response_models = {
   "application/json" = "Empty" // default model
 } response_parameters_in_json = <<PARAMS
  {
    "method.response.header.X-Header": true
  }
PARAMS}

Theses resources allow you to configure how data is mapped from the Lambda function back to the client. This configuration allows you to configure HTTP Headers and transform the Lambda results into different formats. I strongly suggest reading the API Gateway docs for more information.

Last but not least we have to configure the deployment.
The aws_api_gateway_deployment resource allows us to configure an API Gateway stage and stage variables.

resource "aws_api_gateway_deployment" "deployment" {
 depends_on=[
   "aws_api_gateway_integration.foo_get_endpoint_integration"
 ]
 rest_api_id = "${aws_api_gateway_rest_api.foo_api.id}"
 stage_name = "${var.api_stage}"
 variables = {
   "functionAlias” = “${var.api_stage}"
 }
}

This resource depends on other API Gateway methods. If no HTTP methods are defined, this resource will fail to deploy.

Wrapping up
There is a lot of configuration involved in setting up a single HTTP endpoint with Terraform. Each HTTP endpoint requires at least five Terraform resources. Furthermore, the HTTP method and integration resources cannot be separated into a module due to the aws_api_gateway_deployment dependencies (if it is possible let me know!). The end result is a lot of duplication due to boilerplate code. This is not a problem with the Apex framework, rather the API Gateway resources for Terraform.

What’s Next

So far we have created a basic Apex project with an API Gateway endpoint backed by Lambda. In the next part, we’ll look at using Terraform configure Authorization for API Gateway and a Lambda worker.

If you want to see more articles like this, follow Serverless Zone and myself on Medium or Twitter.

Myself and the team at A Cloud Guru are building a Serverless training system. If you need to get AWS certified or build Alexa skills sign up and start learning today.