Skip to content

A realworld app implementation using nx, nest.js & graphql using a micro-services approach

Notifications You must be signed in to change notification settings

sefininio/microservices-realworld-app

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

39 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Microservices Realworld Example App

A realworld app micro-services implementation, using the following tech stack:

  • NX.dev Monorepo
  • Nest.js for BE services
  • Mongoose using @nestjs/mongoose module for persistence
  • GraphQL using @nestjs/graphql module for gateway GraphQL API
  • Auth with passport-jwt using @nestjs/jwt and @nestjs/passport modules
  • Global config using @nestjs/config module
  • Swagger OpenAPI route documentation using @nestjs/swagger module
  • TypeScript
  • MongoDB - for service persistence layer running in a docker container
  • Redis - Queue implementation over Bull and Redis as a service message bus using @nestjs/bull

Application Bootstrap

  • Run yarn predeploy to build services and create docker containers
  • Run docker-compose up to start containers
  • Create the mongo db: SSH into the mongo container:
docker exec -it realworld-mongo sh

Once SSHed into it:

mongo
use realworld
quit()
  • Use PostMan exported collection realworld.postman_collection.json for calling APIs
  • Use Apollo Studio and point the sandbox to http://localhost:3330/graphql for executing GraphQL queries and Mutations.
  • Run nx serve frontend if you would like to use the (not complete) frontend application.

The feature micro services are:

  • feature-user is responsible for /api/user functionality and endpoints.
  • feature-tag is responsible for /api/tags functionality and endpoints.
  • feature-profile is responsible for /api/profie functionality and endpoints.
  • feature-article is responsible for /api/articles functionality and endpoints.

All feature services expose OpenApi Swagger definitions under the /api route.

The GraphQL Gateway

  • realworld-examples-app is the GraphQL gateway for the above services. Code-First GraphQL autogenerated schema is saved under schema.graphql.

The playground is accessible via Apollo Studio, point the sandbox to http://localhost:3330/graphql.

For example, getting the current logged-in user details, along with her feed and the list of articles she wrote (with comments for each article returned in feed/articles if they exist):

fragment userFragment on User {
  _id
  email
  username
}

fragment articleFragment on Article {
  _id
  body
  description
  favoritesCount
  tagList
  title
  comments {
    _id
    body
    author {
      ...userFragment
    }
  }
  author {
    ...userFragment
  }
}

query {
  me {
    ...userFragment
    feed {
      ...articleFragment
    }
    articles {
      ...articleFragment
    }
  }
}

For using the GraphQL gateway with authenticaed routes:

  • First call the login mutation with username (email) and password -> get the access_token.
  • Add {"token": <JWT token>} to the headers.

See data-access lib for predefined Queries, Mutations and Fragments and their respective autogenerated Types and React hooks.

The libs

Feature services and the GraphQL gateway use the following libs:

  • auth is responsible for user login and issueing the JWT token, as well as exporting the JwtAuthGuard used on protected routes in all feature services.
  • models exposes DTOs and ENUMs used by feature services.
  • shared exposes reusable services used by feature services.
  • data-access exposes auto-generated types and react hooks based on schema types and queries/mutations. Generated code is the result of running yarn codegen or yarn codegen:watch. Codegen config is codegen.yaml.

Apps and Libs dependency graph

dep-graph

Micro services message bus

We use Redis as the queue persistence layer and Bull & @nestjs/bull for the queue implementation.

The purpose of this is for feature services to be able to notify something happened in the system, and other feature services can react and complete the flow.

One flow that uses this is:

  • User creates an article, with tagList that includes some new tags and some existing tags
  • feature-article service publishes a message with the tagList on the EvaluateTags topic to the Tags queue and saves the new article.
  • feature-tag implements a consumer that processes messages under the EvaluateTags topic on the Tags queue and creates the new tags in DB.

Another exaple is:

  • User is created or updated
  • feature-user service publishes userUpdated event on the Users queue and saves the new/updated user.
  • feature-profile services implements a UserConsumer consumer that processes messages under userUpdated topic on the Users queue and creates/updates the relevant profile in DB.

See Real World App spec here.

About

A realworld app implementation using nx, nest.js & graphql using a micro-services approach

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published