@Gapi Advanced Serverless Sequelize

@Nginx, @Rabbitmq, @Postgres, @Sequelize, @Docker, @Graphql

This is advanced example serverless project related with GAPI to check basic example project go to basic-example

Included @gapi/sequelize module

To start developing clone repository

git clone https://github.com/Stradivario/gapi-starter-serverless-sequelize

Better use command line utility(gapi) to install it type following command:

npm i -g @gapi/cli

Type the following command to create new project from scratch via CLI

gapi new my-project --serverless-sequelize

To start project for "development" type:

npm start

To deploy to staging or production type:

This will run Parcel and will Bundle your lambdas to single file inside dist folder example: src/user/main.ts will be bundled inside dist/user/main.js with main.map example2 src/auth/main.ts will be bundled inside dist/auth/main.js with main.map Then the lambdas will be deployed based on your serverless.yml configuration

staging

npm run deploy:serverless

production

npm run deploy:serverless:prod

Adding microservices inside src/app/app-proxy.module.ts

import { Module } from '@gapi/core';
import { CoreModule } from './core/core.module';
import { GapiMicroserviceModule } from '@gapi/microservices';

@Module({
    imports: [
        CoreModule,
        GapiMicroserviceModule.forRoot([
            {
                name: 'auth-microservice',
                link: `${process.env.AWS_CLOUD_LINK}/auth/graphql`
            },
            {
                name: 'user-microservice',
                link: `${process.env.AWS_CLOUD_LINK}/user/graphql`
            }
        ])
    ]
})
export class AppProxyModule { }

Environment AWS_CLOUD_LINK you can set inside gapi-cli.conf.yml for specific environment

    lambda-proxy-production:
      AWS_CLOUD_LINK: 'https://lpsm25eo1k.execute-api.us-east-2.amazonaws.com/prod'
    
    lambda-proxy-staging:
      AWS_CLOUD_LINK: 'https://aiey6imalh.execute-api.us-east-2.amazonaws.com/staging'

To start @gapi as a proxy server using 2 created lambda microservices

More info inside gapi-cli.conf.yml

It will run environment lambda-proxy-staging from gapi-cli.conf.yml

gapi proxy staging

It will run environment lambda-proxy-production from gapi-cli.conf.yml

gapi proxy prod

Serverless.yml configuration

service: gapi-sequelize-serverless
provider:
  name: aws
  runtime: nodejs8.10
  stage: staging
  profile: default
  region: us-east-2

functions:
  auth:
    handler: src/app/auth/main.handler
    events:
      - http:
          path: "/auth/{proxy+}"
          method: any
          cors: true
          integration: lambda-proxy
    memorySize: 128
    timeout: 10
  user:
    handler: src/app/user/main.handler
    events:
      - http:
          path: "/user/{proxy+}"
          method: any
          cors: true
          integration: lambda-proxy
    memorySize: 128
    timeout: 10

plugins:
  - serverless-offline

Testing shows that Gapi is using 85 MB lambda memory so i decided to limit them to 128 mb if you use it well it will treat you well :)

Important!!! To deploy to AWS with Serverless you need to compile project to ES6 commonjs module everything is preconfigurated inside tsconfig.json

Compiling project to ES6 commonjs

tsc

Simple as that :)

Bundle will be somewhere between 18-20 mb.The basic concept is that you can create many functions related with same project run them as SingleServer but when you deploy them every module is independent microservice.In the end you create PROXY server with Microservices which MERGES Both Microservices Lambdas as a Single Endpoint GraphQL

Njoy! :)

Testing

To start developing with testing GAPI uses JEST and gapi-cli is preconfigurated for your needs! :)

To run single test type:

gapi test

Testing watch mode

Note: You need to start server before running tests

Note: Everytime you make change to server it will restart server and execute tests

Note: To add more tests just create e2e.spec.ts or unit.spec.ts somewhere inside the application

Start the application

gapi start

Execute test with --watch argument

gapi test --watch

You will end up with something like this

Custom logic before testing ( for example creating MOCK users to database before testing)

Create file test.ts inside root/src/test.ts with this content

Everytime you run test with --before argument it will set environment variable BEFORE_HOOK

  if (process.env.BEFORE_HOOK) {
    // do something here
  }

Then execute tests with --before

gapi test --before

This command will start root/src/test.ts file and will wait for process.exit(0) so you can customize your before logic check this link for reference

Following commands will start RabbitMQ, PostgreSQL, API, NGINX as a services you need DOCKER for them

API will be served on https://localhost:80 and https://localhost:80/subscriptions

Docker

To build project with Docker type:

gapi app build

To start project with Docker type:

gapi app start

To stop project type:

gapi app stop

Workers

All workers will be mapped as Proxy and will be reverted to https://localhost:80 and https://localhost:80/subscriptions

So you don't have to worry about if some of your workers stopped responding

TODO: Create monitoring APP for all workers and main API

To start workers type:

gapi workers start

To stop workers type:

gapi workers stop

To add more workers

By default there are 4 workers with 4 processes with "exec_mode: cluster" of the original process inside single docker container

You can control Processes inside single docker container from "root/process.yml" file.

apps:
  - script   : './src/main.ts'
    name     : 'APP'
    exec_mode: 'cluster'
    instances: 4

To map new worker as a stream open root/nginx/config/private/default

upstream app_servers {
    server 182.10.0.3:9000; # Main process
    server 182.10.0.21:9000; # Worker 1
    server 182.10.0.22:9000; # Worker 2
    server 182.10.0.23:9000; # Worker 3
    server 182.10.0.24:9000; # Worker 4

    # Add more workers here
    # server 182.10.0.25:9000; # Worker 5
}

server {
    listen 80;
    server_name api.yourdomain.com;
    access_log api-yourdomain.access.log;

    location / {
       proxy_set_header Upgrade $http_upgrade;
       proxy_set_header Connection "upgrade";
       client_max_body_size 50M;
       proxy_set_header Host $http_host;
       proxy_set_header X-Real-IP $remote_addr;
       proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
       proxy_set_header X-Forwarded-Proto $scheme;
       proxy_set_header X-Frame-Options SAMEORIGIN;
       proxy_buffers 256 16k;
	     proxy_buffering off;
       proxy_buffer_size 16k;
       proxy_read_timeout 600s;
       proxy_pass http://app_servers;
    }

    location /subscriptions {
         # prevents 502 bad gateway error
        proxy_buffers 8 32k;
        proxy_buffer_size 64k;

        # redirect all HTTP traffic to localhost:9000;
        proxy_pass http://app_servers/subscriptions;
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header Host $http_host;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        #proxy_set_header X-NginX-Proxy true;

        # enables WS support
        proxy_http_version 1.1;
        proxy_set_header Upgrade $http_upgrade;
        proxy_set_header Connection "upgrade";
	      proxy_buffering off;
        proxy_read_timeout 999999999;

    }
    if ($scheme = http) {
       return 301 https://$server_name$request_uri;
    }
    listen 443;
    ssl on;
    ssl_certificate         /usr/share/certs/cert.pem;
    ssl_certificate_key     /usr/share/certs/cert.key;
}

When you add another worker it should be on different IP with same port 9000

Open root/gapi.conf.yml file you will find this file:

config:
# Application configuration
  app: 
    local:
      API_PORT: 9000
      API_CERT: ./cert.key
      NODE_ENV: development
      AMQP_HOST: 182.10.0.5
      AMQP_PORT: 5672
      DB_PORT: 5432
      DB_NAME: postgres
      DB_HOST: 182.10.0.4
      DB_USERNAME: dbuser
      DB_PASSWORD: dbuserpass
      GRAPHIQL: true
      ENDPOINT_TESTING: http://localhost:9000/graphql
      TOKEN_TESTING: eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJlbWFpbCI6ImtyaXN0aXFuLnRhY2hldkBnbWFpbC5jb20iLCJzY29wZSI6WyJBRE1JTiJdLCJpZCI6MSwiaWF0IjoxNTE2OTk2MzYxfQ.7ANr5VHrViD3NkCaDr0nSWYwk46UAEbOwB52pqye4AM
      GRAPHIQL_TOKEN: eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJlbWFpbCI6ImtyaXN0aXFuLnRhY2hldkBnbWFpbC5jb20iLCJpZCI6MSwic2NvcGUiOlsiQURNSU4iXSwiaWF0IjoxNTIwMjkxMzkyfQ.9hpIDPkSiGvjTmUEyg_R_izW-ra2RzzLbe3Uh3IFsZg
    prod:
      API_PORT: 9000
      API_CERT: ./cert.key
      NODE_ENV: production
      AMQP_HOST: 182.10.0.5
      AMQP_PORT: 5672
      DB_PORT: 5432
      DB_HOST: 182.10.0.4
      DB_USERNAME: dbuser
      DB_PASSWORD: dbuserpass
      DB_NAME: postgres
# Testing configuration for local(dev) or worker(running tests as a separate worker with separate database)
  test: 
    local: extends app/local
    worker:
      API_PORT: 9000
      API_CERT: ./cert.key
      NODE_ENV: production
      DB_PORT: 5432
      DB_HOST: 182.10.0.99
      AMQP_HOST: 182.10.0.5
      AMQP_PORT: 5672
      DB_USERNAME: dbuser
      DB_PASSWORD: dbuserpass
      DB_NAME: postgres
      ENDPOINT_TESTING: http://182.10.0.101:9000/graphql
      TOKEN_TESTING: eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJlbWFpbCI6ImtyaXN0aXFuLnRhY2hldkBnbWFpbC5jb20iLCJzY29wZSI6WyJBRE1JTiJdLCJpZCI6MSwiaWF0IjoxNTE2OTk2MzYxfQ.7ANr5VHrViD3NkCaDr0nSWYwk46UAEbOwB52pqye4AM
  schema:
    introspectionEndpoint: http://localhost:9000/graphql
    introspectionOutputFolder: ./src/app/core/api-introspection

commands:
  testing:
    stop:
      - docker rm -f gapi-api-prod-worker-tests-executor
      - docker rm -f gapi-api-prod-worker-tests-provider
    start:
      - gapi testing start-provider
      - sleep 10
      - gapi testing start-executor
      - echo Cleaning...
      - gapi testing stop
    start-executor:
      - docker run -d --network=gapiapiprod_gapi --ip=182.10.0.100 --name gapi-api-prod-worker-tests-executor gapi/api/prod
      - docker exec gapi-api-prod-worker-tests-provider npm -v
      - gapi test --worker --before
    start-provider: docker run -d --network=gapiapiprod_gapi --ip=182.10.0.101 --name gapi-api-prod-worker-tests-provider gapi/api/prod
  workers:
    start:
      - gapi workers start-1
      - gapi workers start-2
      - gapi workers start-3
      - gapi workers start-4
    stop:
      - docker rm -f gapi-api-prod-worker-1
      - docker rm -f gapi-api-prod-worker-2
      - docker rm -f gapi-api-prod-worker-3
      - docker rm -f gapi-api-prod-worker-4
    start-1: docker run -d --network=gapiapiprod_gapi --ip=182.10.0.21 --name gapi-api-prod-worker-1 gapi/api/prod
    start-2: docker run -d --network=gapiapiprod_gapi --ip=182.10.0.22 --name gapi-api-prod-worker-2 gapi/api/prod
    start-3: docker run -d --network=gapiapiprod_gapi --ip=182.10.0.23 --name gapi-api-prod-worker-3 gapi/api/prod
    start-4: docker run -d --network=gapiapiprod_gapi --ip=182.10.0.24 --name gapi-api-prod-worker-4 gapi/api/prod
    example-worker-with-port: docker run -d --network=gapiapiprod_gapi --ip=182.10.0.25 --name gapi-api-prod-worker-5 -p 9001:9000 gapi/api/prod
  app:
    start:
      - docker-compose -p gapi-api-prod up --force-recreate -d
      - gapi rabbitmq enable-dashboard
    stop:
      - gapi nginx stop
      - gapi api stop
      - gapi rabbitmq stop
      - gapi postgres stop
    build: docker build -t gapi/api/prod .
  api:
    stop: docker rm -f gapi-api-prod
  nginx:
    stop: docker rm -f gapi-api-nginx
  postgres:
    stop: docker rm -f gapi-api-postgres
  rabbitmq:
    stop: docker rm -f gapi-api-rabbitmq
    restart: docker restart gapi-api-rabbitmq
    enable-dashboard: docker exec gapi-api-rabbitmq rabbitmq-plugins enable rabbitmq_management

# You can define your custom commands for example 
# commands:
#   your-cli:
#     my-command: 'npm -v'
# This command can be executed as "gapi your-cli my-command"

Adding one more worker:

start-5: 'docker run -d --network=gapiapiprod_gapi --ip=182.10.0.25 --name gapi-api-prod-worker-5 -p 9005:9000 gapi/api/prod'

Then edit start task inside workers to start new worker 5

start: 'gapi workers start-1 && gapi workers start-2 && gapi workers start-3 && gapi workers start-4 & gapi workers start-5'

Thats' it!! Now you have 4 processes like CLUSTERS inside 1 docker container with ip 182.10.0.25 and external port(optional) 9005;

You can specify worker also without port because all workers are inside internal network called "gapiapiprod_gapi"

182.10.0.21/22/23/24/25:9000

start-5: 'docker run -d --network=gapiapiprod_gapi --ip=182.10.0.25 --name gapi-api-prod-worker-5 gapi/api/prod'

If you want to change port forwarding to another port you need to set just nginx configuration:

  nginx:
    image: sameersbn/nginx:1.10.1-5
    ports:
      - "81:80"
      - "443:443"

Now you can find your API served onto https://localhost:81/ and https://localhost:81/subscriptions

All workers don't care about that because they will be served and mapped from nginx to port 80.

You can check docker-compose file to configurate environment variables

version: '2'
services:

  nginx:
    image: sameersbn/nginx:1.10.1-5
    ports:
      - "81:80"
      - "444:443"
    volumes:
      - ./nginx/config:/etc/nginx
      - ./nginx/html:/usr/share/nginx/html/
      - ./nginx/certs:/usr/share/certs
    restart: always
    container_name: gapi-api-nginx
    networks:
      gapi:
        ipv4_address: 182.10.0.2

  api:
    image: gapi/api/prod:latest
    ports:
      - "9000"
    environment:
      - NODE_ENV=production
      - API_PORT=9000
      - GRAPHIQL_TOKEN=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJlbWFpbCI6InRhNGVuc3RvcmVAZ21haWwuY29tIiwic2NvcGUiOlsiQURNSU4iXSwiaWQiOjIsImlhdCI6MTUxMTk3NDkzNX0.M9PnW2IrVp4XGRvbzFrl0tx6vqs6oXItFK-wF5roneI
      - AMQP_HOST=182.10.0.5
      - AMQP_PORT=5672
      - CERT_PATH=

      # Production db config
      - DB_HOST=182.10.0.4
      - DB_PORT=5432
      - DB_NAME=postgres
      - DB_USERNAME=dbuser
      - DB_PASSWORD=dbuserpass

    restart: always
    mem_limit: 1000000000
    cpu_shares: 73
    container_name: gapi-api-prod
    depends_on:
      - nginx
      - rabbitMq
      - PostgreSQLDev
    networks:
      gapi:
        ipv4_address: 182.10.0.3

  PostgreSQLDev:
    image: sameersbn/postgresql:9.5-3
    ports:
      - "5432"
    environment:
      - DEBUG=false
      - TIMEZONE=Europe/Sofia
      - LOCALE=bg_BG.UTF-8

      - DB_USER=dbuser
      - DB_PASS=dbuserpass
      - DB_NAME=postgres
      - DB_TEMPLATE=

      - DB_EXTENSION=

      - REPLICATION_MODE=
      - REPLICATION_USER=
      - REPLICATION_PASS=
      - REPLICATION_SSLMODE=
    restart: always
    container_name: gapi-api-postgres
    networks:
      gapi:
        ipv4_address: 182.10.0.4

  PostgreSQLDevTesting:
    image: sameersbn/postgresql:9.5-3
    ports:
      - "5432"
    environment:
      - DEBUG=false
      - TIMEZONE=Europe/Sofia
      - LOCALE=bg_BG.UTF-8

      - DB_USER=dbuser
      - DB_PASS=dbuserpass
      - DB_NAME=postgres
      - DB_TEMPLATE=

      - DB_EXTENSION=

      - REPLICATION_MODE=
      - REPLICATION_USER=
      - REPLICATION_PASS=
      - REPLICATION_SSLMODE=
    restart: always
    container_name: gapi-api-postgres-testing
    networks:
      gapi:
        ipv4_address: 182.10.0.99

  rabbitMq:
    image: rabbitmq:3.7.2
    ports:
      - "15672:15672"
      - "5672:5672"
      - "5671:5671"
      - "4369:4369"
    restart: always
    container_name: gapi-api-rabbitmq
    networks:
      gapi:
        ipv4_address: 182.10.0.5

  pgadmin:
    image: thajeztah/pgadmin4
    ports:
      - "5050"
    volumes:
      - /usr/bin/:/usr/pg
      - /usr/database/:/usr/database
      - /tmp/:/tmp
    restart: always
    container_name: gapi-api-pg-admin
    networks:
      gapi:
        ipv4_address: 182.10.0.6

networks:
  gapi:
    driver: bridge
    ipam:
     config:
       - subnet: 182.10.0.0/16
         gateway: 182.10.0.1

After successfully started project you can open your browser to localhost:80 or 182.10.0.3:9000 the api will be served there

TODO: Better documentation...

Enjoy ! :)