Skip to content

Latest commit

 

History

History
150 lines (128 loc) · 3.82 KB

README.md

File metadata and controls

150 lines (128 loc) · 3.82 KB

AWS Microservice package

This is a Node based lambda microservice package created by AWS-Architect.

Recent Changes

Visit the changelog.

Prerequisites

  • Install NodeJS (nodejs8.10 this is what lambda uses) & npm
    curl -sL https://deb.nodesource.com/setup_6.x | sudo -E bash -
    sudo apt-get install -y nodejs
  • Your user will need access to the following resources (or the continuously deployment user):
    • Development time resources (identical for deployment CI), example security policy
    • Service runtime resources (for testing only, not required, execute lambda, api gateway access, etc...)

Development

Development is templated using the make.js file. All the needed actions are present there. For ease, the AWS Architect to managed as a npm package. So all functionality is available directly from native nodejs, no having to write shell scripts just do some simple development.

  • Website is created from the content directory.
  • Lambda functions are created from the src/index.js source.
  • npm install: Install necessary dependencies.
  • npm run build or node make.js build: Builds and run unit tests.
  • sudo npm start: Runs the microservice locally, it inhabits the api and lambda functions using nodejs express.
  • npm run deploy: Deploys the package to AWS.

Building

  npm install
  npm run build

Running server locally

AWS Architect uses OpenAPI Factory to convert the src/index.js into a node server API used by node-express. This can be loaded, and the server can be started by running

   npm install
   npm run start

Deploying to AWS

Configure your make file with the necessary account information

  • Using the built in make.js file
	npm install
	npm run deploy
* Configure awsArchitect
	let packageMetadataFile = path.join(__dirname, 'package.json');
	let packageMetadata = require(packageMetadataFile);

	let apiOptions = {
		sourceDirectory: path.join(__dirname, 'src'),
		description: 'This is the description of the lambda function',
		regions: ['us-east-1']
	};
	let contentOptions = {
		bucket: 'WEBSITE_BUCKET_NAME',
		contentDirectory: path.join(__dirname, 'content')
	};
	let awsArchitect = new AwsArchitect(packageMetadata, apiOptions, contentOptions);

First time setup

AWS Architect uses some CF macros that need to be deployed to CloudFormation. These exist to make your stacks simplier. To deploy them, run these two commands. You'll want to replace the two variables with an S3 bucket and AWS credentials profile (unless you want to use the default).

  npm install -g aws-architect-cf-macros
  aws-architect-cf-macros deploy TMP_DEPLOYMENT_BUCKET --profile PROFILE_NAME

Setup

Permissions to invoke lambda functions

  • From CloudWatch Rules:
{
	"SourceAccount": { "Ref": "AWS::AccountId" },
	"SourceArn": {
		"Fn::Join": [
			"",
			[
				"arn:aws:events:",
				{ "Ref": "AWS::Region" },
				":",
				{ "Ref": "AWS::AccountId" },
				":rule",
				"/",
				{ "Ref": "serviceName" },
				"-*"
			]
		]
	}
}
  • From CloudWatch Logs:
{
	"SourceAccount": { "Ref": "AWS::AccountId" },
    "SourceArn": {
        "Fn::Join": [
			":",
            [
              "arn:aws:logs",
              { "Ref": "AWS::Region" },
              { "Ref": "AWS::AccountId" },
              "log-group",
              "*",
              "*"
            ]
        ]
	}
}
  • From SES:
{
	"SourceAccount": { "Ref": "AWS::AccountId" }
}
  • From API Gateway:
{
	"SourceAccount": { "Ref": "AWS::AccountId" },
	"SourceArn": {
		"Fn::Join": [
			"",
			[
				"arn:aws:execute-api:",
				{ "Ref": "AWS::Region" },
				":",
				{ "Ref": "AWS::AccountId" },
				":",
				{ "Ref": "ApiGateway" },
				"/*"
			]
		]
	}
}