The graphql serverless backend

Table of Contents

1. Introduction
2. How to setup local dev environment
3. Architecture
4. Coding Practices

Introduction

This repo contains the starter code for the a serverless graphql backend.

How to setup local dev environment

Note: The steps mentioned below is for ubuntu. For MAC as well it should remain same as ubuntu. But for windows some things might change. I haven't tested for the windows yet.

Prerequisites

• Docker
• Docker Compose
• NodeJS

Steps

• First clone the repo
• After cloning run below command

yarn install

• Now run below command

docker volume create --name=local-development-pg-volume

OR

sudo docker volume create --name=local-development-pg-volume

• And now run below command

docker-compose up

OR

sudo docker-compose up

• Finally run below command to start the dev server

yarn dev

Graphql Playground

Open http://localhost:3000/dev/playground and you will be able to access the graphql playground.

Architecture

The backend is built as a serverless architecture. We are using a framework https://www.serverless.com/ It helps us deploy and upload our code easily to AWS Lambada. It utilizes AWS cloudformation to create required resources on AWS. The file serverless.yml in the root folder contains the

Tech Stack

• GraphQL
• Apollo
• AWS Lambada
• serverless.com
• NodeJS
• AWS RDS PostGrace

How the local dev environment working?

To run the the local development environment we are using a plugin called serverless-offline. This allows us to simulate a lambada function execution.

How the staging and prod environments are working?

AWS cloudformation deploy the required stacks. OAll of the code is stored in s3. The AWS Lambada then references that code.

When introducing new environment variables

• Make sure to add all the environment variables in the circle ci dashboard.
• Also add all of the environment variables in the serverless.yml file under the provider environment section.

Coding Practices

• For all db operations make a file with the same name as the table and do all of the operations in it. All those files currently are in core folder
• All of the table names should be singular. Just because "user" keyword is reserved in PostGres I have named our one table "users" to avoid any confusion with the served keyword. Make sure to name all other tables singular
• All of the data validation, authentication and autherization login should be in the graphql resolver.
• All of the data stored in the DB will be in snake case
• All requests and responses data to client will be in the snake case.
• Any unused variables should start with _ Like below.

async (_parent, _args, ctx, info) => {
      const user = ctx.assertAuthenticated();
      const results = await User.findById(ctx, user.id);
      return results[0];
    },