Clevertech boilerplate for projects based on Docker, Node.js and React
Clone or download
Hookkid Merge pull request #34 from hijarian/un-dockerizing
Added instructions on how to run the application without Docker.
Latest commit 26bc364 Jan 10, 2019

README.md

Clevertech Boilerplate

Clevertech provides an integrated technology stack that

  • 📚 Contains a huge amount of best practices already implemented for you
  • 🚀 Provides fast and powerful local application development. Including: hot-reloading, local utilities (npm scripts), dev tools (linters, git hooks), etc.
  • 🛠 Implements a solid workflow for building, testing and deploying applications
  • 👤 Secure and complete authentication functionality (including 2FA with either SMS or apps like Google Authenticator)
  • 🎨 Supports SASS/SCSS

Some of the best practices include:

  • 📦 Caching dependencies for faster builds
  • 🚀 Fast track for faster deploys
  • 🔒 Properly storage of secrets outside the code
  • 💊 Healthcheck implementations
  • ✒️ Linters and code prettifiers
  • 👤 Complete auth infrastructure based on JWT
  • 🏛 Source code architecture that scales

Some of the used technologies are:

  • Node.js for the backend.
  • React.js for the frontend with many other libraries already integrated such as redux, react-router, redux-saga, etc.
  • Docker for containers.
  • Docker Compose to run containers locally and during the build.
  • Jenkins to test, build and deploy the application.
  • Kubernetes to orchestrate the container deployment.
  • PostgreSQL or MySQL (see the create-boilerplate-app script) as database engines.

Containers offer big advantages in software development, quality assurance and software deployment -- namely consistency, reliability and scalability. In particular, scalability is implemented from the beginning, and the system is ready to grow as the application gets traction.

Creating new applications

To create a new application based on the Boilerplate, use this command:

npx github:clevertech/boilerplate#create-boilerplate-app my-new-app

The application will be initalized in the directory my-new-app.

This script can be found in the create-boilerplate-app branch. Follow the link for instructions on how to modify it.

Table of contents

generated with DocToc

Local Development

Local development tools

First of all, run yarn install in the root directory. This will install local development tools such as eslint and git precommits to keep the code formatted and without obvious errors.

Installation

Local prerequisites are minimal, please follow the installation instructions carefully. We support Linux and MacOS; Windows users can use Docker for Windows with several workarounds or use a Linux VM. Review your Docker Advanced settings and consider to assign more CPUs and more memory to the Docker process to boost performance.

Setting up the environment

Application configuration is done by using environment variables. For local development, inside each app directory there should be an .env.example file. Simply copy .env.example to .env and fill in your credentials as needed. The .env. file can be used to store sensitive / personal credentials without the risk of checking it into source control.

-By default CleverAuth is enabled and some env variables are required for it. Take a look to api/.env.example and fill the required values.

Running the application

$ docker/run

That's it! Your application is available at http://local.cleverbuild.biz:8080 (api) and http://local.cleverbuild.biz:3000 (React frontend). Both api and frontend support hot reloading.

On first execution, Docker must download the base container images, which might take a while. Subsequent executions will be faster, taking advantage of Docker caching and Yarn caching. See here for details about the caching mechanisms.

docker/run is the local development start script. This allows for changes made locally to restart the node application in the most efficient and cross-platform way. To configure the app restart, edit nodemon.json. docker/run uses docker-compose to start the application and is probably how you will want to do most of your development.

NOTE: Any change to package.json will require a full restart of the container: you should use CTRL+C to stop the running Docker instance and restart it to see your changes. To avoid that when doing simple changes (like adding a package), you can do something like:

docker-compose exec api yarn add $YOUR_PACKAGE$

Customizing Style

The boilerplate supports styling with SASS/SCSS. Just edit main.scss on frontend/src/styles and the boilerplate will convert it to css on-the-fly so you can take full advantage of all of SASS's features.

Tests

Running the tests

$ docker/test

Creating new tests

Unit tests

Just write regular tests using jest. Use describe() and it() to write new tests and test suites.

End 2 end tests

End to end tests are implemented using TestCafe. You can find an example in e2e. These tests need to be run having the containers up.

yarn run e2e

Connecting to the database

You'll find the password in the docker-compose.yml file.

From the command line

With the containers started just do:

yarn run db-client

Using Adminer

Just go to http://127.0.0.1:8081/ and fill the connection information based on what you'll find in docker-compose.yml.

Under the hood

How it works.

Application Health Check

Each application should expose two health checks.

Quick health check

The quick health checks is called by Kubernetes to assess the Liveness and Readiness of an application; it is called approximately every 10-30 seconds, so it can not involve any computation or resource access (database, cache, etc.) to not stress the infrastructure. For a web application, an endpoint returning HTTP 200 is usually the right choice.

URL: /healthz

Long health check

The long health checks is called by an external service to assess the status of the entire application stack. It is called approximately every 1 minute, so it can involve computation and light access to external resources. The long health check should verify the liveness of every dependent service and:

  • Return HTTP 200 if everything is good.
  • Return HTTP 500 and a human-readable error message if something is wrong.

Examples of controlled services:

  • Simple query to assess database liveness and connectivity (SELECT 1;)
  • Simple query to assess Redis connectivity
  • Simple API call to assess external REST API liveness and connectivity

A module should also check for a consistent internal state. For example: a email-sending worker should check for the sending queue to be less than a certain thresholds; a workerd consuming events in a queue should check for queue size, etc.

The long health check should be authenticated to avoid a potential DoS attack. The secret for the long health check is passed as an environment variable to the process (HEALTH_CHECK_SECRET).

URL: /healthz/long/${process.env.HEALTH_CHECK_SECRET}

Directory structure

This project is setup to have multiple services in the same repository, as such the structure of this is quite important. Each directory of this project is intended to be for a different service and as such, those child folders should contain all necessary components to build that service entirely.

It is important that each new service has a Dockerfile.

project
├─ docker-compose.yml
├─ README.md
├─ INSTALL.md
├─ api
│  ├─ Dockerfile
|  ├─ package.json
│  └─ src
│     ├─ file1.js
│     └─ file2.js
└─ frontend
   ├─ Dockerfile
   ├─ package.json
   └─ src
      ├─ file1.js
      └─ file2.js

Troubleshooting and useful Docker commands

Common issues that developers may encounter when executing this project and useful Docker commands.

Deploy

Branches development, staging and master are deployed after build to their server environments.

The deploy uses a fast-path mechanism. When a given application version is tested on the development site, the same container image is deployed on the staging/production site, avoiding to repeat the build and testing process when promoting a new version to staging/production. This allows to put a new version on production in 30 seconds and to avoid build inconsistencies.

The mechanism is triggered when a build is started for a branch which is identical to another branch that has already been built and deployed. So, for example, when you merge development to staging (without code modifications) the build phase is skipped, and the existing image is deployed. To take advantage of the fast path, do not merge development to staging while development is still building: it's faster and safer to wait for development build completion before merging.

As also Pull Requests are built, this mechanism works in some cases also for PRs. For example, if you merge a rebased PR to development, the fast path is used.

Implementation is in docker/fast_track script.