lsxBread / predictivehire-codetest

PredictiveHire Code Test

This demo project is containing a GraphQL Server + REST API gateway with back-end microservices all written using the NestJS Framework and TypeScript

Architecture Overview

The GraphQL server receives the requests from client side and fetch data by calling RESTful API, ultimately it will return aggregated data to client side.

The REST API acts as a gateway/proxy for the different microservices it exposes.

The controllers of the REST API make calls to the servers/microservices in the back-end through TCP.

The Nestjs microservices then handles the request to connect to databases or any other service it needs to serve requests.

All the components(GraphQL, API Gateway, microservies) are dockerized and for simplicity purpose, these containers are all in bridge network, so that they are visible to outside through its service name.

Diagram

A diagram of the architecture is shown below.

Deployment

Deployment is done with containers in mind. A Docker Compose file along with Dockerfiles for each project are given to run the whole thing on any machine. For production, it's always recommended to use Kubernetes for these kinds of microservices architecture to deploy in production.

Naming convention

The service folder is named by [service name]-[service default port]. E.g. api-gateway-3000

RESTful API Desgin

Auth controller

Method	API	Payload	Return	Authorization	GraphQL
POST	/login	{ "username": String, "password": String!, }	{ username: String userId: String companyId: String role: String accessToken: String }	false	mutation{ login(username:"leo", password:"123") { accessToken } }

Company controller

Method	API	Payload	Return	Authorization	GraphQL
POST	/company	{ "name": String, "address": String, }	{ _id: string; name: string; address: string; __v: number; }	false	mutation { createCompany(input:{name:"predictivehire", address:"ascot st"}) { _id } }
GET	/company/{id}		{ _id: string; name: string; address: string; __v: number; }	false	query{ company(id:"5fcf02c2b494590029b333a6") { name } }

User controller

Method	API	Payload	Return	Authorization	GraphQL
POST	/register	{ "name": String, "username": String "password": String "role": String "companyId": String }	{ _id: string; name: string; username: string; role: string; __v: number; }	false	mutation{ register(input:{name:"Bob Markle", username:"bob", password:"bob", role:"user", companyId:""}){ _id } }

Vacancy controller

Method	API	Payload	Return	Authorization	GraphQL
GET	/vacancy/{vacancyId}?company_id={companyId}		{id: String title: String description: String expireAt: String}	true	query{ vacancy(companyId:"", vacancyId:""){ title } }
POST	/vacancy?company_id={companyId}	{ "title": String "description": String "expireAt": ISOTimeString, }	{id: String title: String description: String expireAt: String}	true & Admin	mutation{ createVacancy(input:{title:"", description:"", expireAt:"", companyId:""}){ title } }
PUT	/vacancy/{vacanyId}?company_id={companyId}	{ "title": String "description": String "expireAt": ISOTimeString, }	{id: String title: String description: String expireAt: String}	true & Admin	mutation{ updateVacancy(input:{title:"", description:"", expireAt:""}, companyId:"", vacancyId:""){ title } }
DELETE	/vacancy/{vacanyId}?company_id={id}		{id: String title: String description: String expireAt: String}	true & Admin	mutation{ deleteVacancy(companyId:"", vacancyId:""){ title } }

How to Run

System Requirements - Linux/Mac should be better, I did not test Windows yet

• Node.js - v12 Required (You can run nvm use to install the desired node version which already been defined in .nvmrc)
• Docker - latest
• Docker Compose - latest

Production

Please check the .env.prod first to see which port you would like to use

1. On the Terminal, go into the project's root folder (cd /project/root/folder) and execute docker-compose -f docker-compose.yml --env-file .env.prod build . The docker will start building the image and it may take a while. After the images are build, execute docker-compose -f docker-compose.yml --env-file .env.prod up to run the docker containers.

2. Once all the containers are on, the GraphQL server will listen on http://localhost:4000 by default

Local Development

Please check the .env.dev first to see which port you would like to use

1. On the Terminal, go into the project's root folder (cd /project/root/folder) and execute yarn, then execute docker-compose -f docker-compose.dev.yml --env-file .env.dev build . The docker will start building the image and it may take a while. After the images are build, execute docker-compose -f docker-compose.dev.yml --env-file .env.dev up to run the docker containers.

2. Once all the containers are on, the GraphQL server will listen on http://localhost:4000 by default, and all the services are running in development --watch mode. So you can debug the code, and the services will be hot-reloaded.

Unit test

1. On the Terminal, go into the project's root folder (cd /project/root/folder) and execute yarn test

e2e test

1. On the Terminal, go into the project's root folder (cd /project/root/folder) and execute yarn test:e2e

Play with this demo App

Once the GraphQL server is on, you can use the playground to Query / Mutate some data.

In this app, I didn't save any default data in the mongoDB, so before starting playing, please execute the following script to insert ONE company and TWO users.

mutation {
  createCompany(input:{name:"Predictivehire", address:"15 Newto St"}) {
    _id
  }
}

please remember the _id, that's the companyId for user.

mutation{
  register(input:{name:"Bob Markle", username:"bob", password:"bob", role:"user", companyId:"<companyId>"}){
    _id
  }
}

mutation{
  register(input:{name:"Mark Smith", username:"mark", password:"mark", role:"admin", companyId:"<companyId>"}){
    _id
  }
}

Once abvoe steps are done, you can play with the app according to the RESTful API Desgin instruction.

Concerns

Due to time limit and simplicity purpose, here are some highlights I didn't do but have in mind:

1. .env.dev and .env.prod shouldn't be committed, the env variable value should be injected dynamically.
2. I only implement unit test and e2e test for a small partial of the code, but hopefully they can assist you to have a taste on how I coded
3. All the microservices connected through TCP in this demo, but in mose cases, the services will be distributed in different host / region, so using RabbitMQ or Kafka can help decouple the services and improve the system performance.
4. No load balance yet.
5. No service registry center yet.
6. No cache yet.