Make and Deploy a Serverless Application Into AWS lambda

At my job we needed a solution for writing, maintaining and deploying aws lambdas. The serverless framework is a nodejs framework used for making and deploying serverless applications such as AWS Lambdas.

So we selected the serverless application as our choice for these sole reasons:

• Easy to manage configuration environment via enviromental viariables.
• Easy to keep a record of the lambda settings and change history via git, so we can kill the person who did a mistake. (ok ok just kidding, no human has been killed ;) ... yet )
• Because it also is node.js framework we can use the normal variety of the frameworks used for unit and integration testing.
• Also for the reason above we also could manage and deploy dependencies as well using combination of nodejs tools and the ones provided from the serverless framework.
• Ce can have a single, easy to maintain, codebase with more than one aws lambdas without the need for duplicate code.

Install serverless

sudo -H npm i -g serverless

(For windows ommit the sudo -H part)

Our first lambda

If not we need to create our project folder and initialize an node.js project:

mkdir myFirstLambda
cd myFirstLambda
npm init
git add .
git commit -m "Our first project"

Then install serverless as dev-dependency, we need that because on colaborative projects it will install all the required tools to deploy and run the project:

npm install --save-dev serverless

And then run the following command to bootstrap our first lambda function:

serverless create --template aws-nodejs

With that command 2 files have been generated:

• handler.js Where contains our aws lambda handlers.
• serverless.yml where it contains all the deployment and running settings.

Then on handler.js change the function module.exports.hello with a respective name representing the functionality. For our purpoce we will keep it as is. We can run the lambda function locally via the command:

 sls invoke local --stage=dev --function hello

Which it will show the returning value of the function hello on handler.js. Also it is a good idea to place the command above as a start script into package.json at scripts section.

Deploy aws lambda

First of all we need to specify the lambda name. So we need to modify the serverless.yml accorditly in order to be able to specify the AWS lambda name. So we change the functions sections from:

functions:
  hello:
    handler: handler.hello

Into:

functions:
  hello:
    handler: handler.hello
    name: MyLambda
    description: "My First Lambda"
    timeout: 10
    memorySize: 512

With that we can list the deployed lambda as MyLambda as aws console, also as seen above we can specify and share lambda settings.

Furthermore is good idea to specify enviromental variables via at the environment: section with the following setting:

  environment: ${file(./.env.${self:provider.stage}.yml)}

With that we can use the stage for each deployment environment and each setting will be provided from .env files. Also upon deployment the .env files will be used in order to be able to specify the deployed lambda environmental variables as well.

Also is good idea to ship a template .env file named .env.yml.dist so each developer will need to do:

cp .env.yml.dist .env.dev.yml

And fill the appropriate settings. Also for production you need to do:

cp .env.yml.dist .env.prod.yml

Then exclude these files to be deployed except the on offered by the stage parameter (will seen bellow):

package:
  include:
    - .env.${self:provider.stage}.yml
  exclude:
    - .env.*.yml.dist
    - .env.*.yml

Then deploy with the command:

sls deploy --stage ^environment_type^ --region ^aws_region^

As seen the pattern followed is the: .env.^environment_type^.yml where the ^environment_type^ is the value provided from the --stage parameter at both sls invoke and sls deploy commands.

Also we could specify depending the environment the lambda name using these settings as well:

functions:
  hello:
    handler: handler.hello
    name: MyLambda-${self:provider.stage}
    description: "My First Lambda"
    timeout: 10
    memorySize: 512

Where the ${self:provider.stage} takes its value from the --stage parameter. Than applies where the ${self:provider.stage} is met at the serverless.yml file.