This post is part of the series 30 Days of React.

In this series, we're starting from the very basics and walk through everything you need to know to get started with React. If you've ever wanted to learn React, this is the place to start!

Deployment Introduction

Edit this page on Github

Today, we'll explore the different pieces involved in deploying our application so the world can use our application out in the wild.

With our app all tested up through this point, it's time to get it up and live for the world to see. The rest of this course will be dedicated to deploying our application into production.

Production deployment

When talking about deployment, we have a lot of different options:

  • Hosting
  • Deployment environment configuration
  • Continuous Integration (CI, for short)
  • Cost cycles, network bandwidth cost
  • Bundle size
  • and more

We'll look at the different hosting options we have for deploying our react app tomorrow and look at a few different methods we have for deploying our application up. Today we're going to focus on getting our app ready for deployment.

Ejection (from create-react-app)

First things first... we're going to need to handle some customization in our web application, so we'll need to run the npm run eject command in the root of our directory. This is a permanent action, which just means we'll be responsible for handling customizations to our app structure for now on (without the help of our handy create-react-app).

This is where I always say make a backup copy of your application. We cannot go back from ejecting, but we can revert to old code.

We can eject from the create-react-app structure by running the eject command provided by the generator:

npm run eject

After ejecting from the create-react-app structure, we'll see we get a lot of new files in our application root in the config/ and scripts/ directories. The npm run eject command created all of the files it uses internally and wrote them all out for us in our application.

The key method of the create-react-app generator is called webpack, which is a module bundler/builder.

Webpack basics

Webpack is a module bundler with a ginormous community of users, tons of plugins, is under active development, has a clever plugin system, is incredibly fast, supports hot-code reloading, and much much more.

Although we didn't really call it out before, we've been using webpack this entire time (under the guise of npm start). Without webpack, we wouldn't have have been able to just write import and expect our code to load. It works like that because webpack "sees" the import keyword and knows we need to have the code at the path accessible when the app is running.

Webpack takes care of hot-reloading for us, nearly automatically, can load and pack many types of files into bundles, and it can split code in a logical manner so as to support lazy-loading and shrink the initial download size for the user.

This is meaningful for us as our apps grow larger and more complex, it's important to know how to manipulate our build tools.

For example, when we want to deploy to different environments... which we'll get to shortly. First, a tiny introduction to webpack, what it is and how it works.

What it does with bundle.js

Looking into the generated files when we ran npm start before we ejected the app, we can see that it serves the browser two or more files. The first is the index.html and the bundle.js. The webpack server takes care of injecting the bundle.js into the index.html, even if we don't load our app in the index.html file.

The bundle.js file is a giant file that contains all the JavaScript code our app needs to run, including dependencies and our own files alike. Webpack has it's own method of packing files together, so it'll look kinda funny when looking at the raw source.

Webpack has performed some transformation on all the included JavaScript. Notably, it used Babel to transpile our ES6 code to an ES5-compatible format.

If you look at the comment header for app.js, it has a number, 254:

/* 254 */
/*!********************!*\
  !*** ./src/app.js ***!
  \********************/

The module itself is encapsulated inside of a function that looks like this:

function(module, exports, __webpack_require__) {
  // The chaotic `app.js` code here
}

Each module of our web app is encapsulated inside of a function with this signature. Webpack has given each of our app's modules this function container as well as a module ID (in the case of app.js, 254).

But "module" here is not limited to ES6 modules.

Remember how we "imported" the makeRoutes() function in app.js, like this:

import makeRoutes from './routes'

Here's what the variable declaration of makeRoutes looks like inside the chaos of the app.js Webpack module:

var _logo = __webpack_require__(/*! ./src/routes.js */ 255);

This looks quite strange, mostly due to the in-line comment that Webpack provides for debugging purposes. Removing that comment:

var _logo = __webpack_require__(255);

Instead of an import statement, we have plain old ES5 code.

Now, search for ./src/routes.js in this file.

/* 255 */
/*!**********************!*\
  !*** ./src/routes.js ***!
  \**********************/

Note that its module ID is 255, the same integer passed to __webpack_require__ above.

Webpack treats everything as a module, including image assets like logo.svg. We can get an idea of what's going on by picking out a path in the mess of the logo.svg module. Your path might be different, but it will look like this:

static/media/logo.5d5d9eef.svg

If you open a new browser tab and plug in this address (your address will be different... matching the name of the file webpack generated for you):

http://localhost:3000/static/media/logo.5d5d9eef.svg

You should get the React logo:

So Webpack created a Webpack module for logo.svg, one that refers to the path to the SVG on the Webpack development server. Because of this modular paradigm, it was able to intelligently compile a statement like this:

import makeRoutes from './routes'

Into this ES5 statement:

var _makeRoutes = __webpack_require__(255);

What about our CSS assets? Yep, everything is a module in Webpack. Search for the string ./src/app.css:

Webpack's index.html didn't include any references to CSS. That's because Webpack is including our CSS here via bundle.js. When our app loads, this cryptic Webpack module function dumps the contents of app.css into style tags on the page.

So we know what is happening: Webpack has rolled up every conceivable "module" for our app into bundle.js. You might be asking: Why?

The first motivation is universal to JavaScript bundlers. Webpack has converted all our ES6 modules into its own bespoke ES5-compatible module syntax. As we briefly touched on, it's wrapped all of our JavaScript modules in special functions. It provides a module ID system to enable one module to reference another.

Webpack, like other bundlers, consolidated all our JavaScript modules into a single file. It could keep JavaScript modules in separate files, but this requires some more configuration than create-react-app provides out of the box.

Webpack takes this module paradigm further than other bundlers, however. As we saw, it applies the same modular treatment to image assets, CSS, and npm packages (like React and ReactDOM). This modular paradigm unleashes a lot of power. We touch on aspects of that power throughout the rest of this chapter.

Complex, right?

It's okay if you don't understand that out of the box. Building and maintaining webpack is a complex project with lots of moving parts and it often takes even the most experienced developers a while to "get."

We'll walk through the different parts of our webpack configuration that we'll be working with. If it feels overwhelming, just stick with us on the basics here and the rest will follow.

With our newfound knowledge of the inner workings of Webpack, let's turn our attention back to our app. We'll make some modifications to our webpack build tool to support multiple environment configurations.

Environment configuration

When we're ready to deploy a new application, we have to think about a few things that we wouldn't have to focus on when developing our application.

For instance, let's say we are requesting data from an API server... when developing this application, it's likely that we are going to be running a development instance of the API server on our local machine (which would be accessible through localhost).

When we deploy our application, we'll want to be requesting data from an off-site host, most likely not in the same location from where the code is being sent, so localhost just won't do.

One way we can handle our configuration management is by using .env files. These .env files will contain different variables for our different environments, yet still provide a way for us to handle configuration in a sane way.

Usually, we'll keep one .env file in the root to contain a global config that can be overridden by configuration files on a per-environment basis.

Let's install an npm package to help us with this configuration setup called dotenv:

npm install --save-dev dotenv

The dotenv library helps us load environment variables into the ENV of our app in our environments.

It's usually a good idea to add .env to our .gitignore file, so we don't check in these settings.

Conventionally, it's a good idea to create an example version of the .env file and check that into the repository. For instance, for our application we can create a copy of the .env file called .env.example with the required variables.

Later, another developer (or us, months from now) can use the .env.example file as a template for what the .env file should look like.

These .env files can include variables as though they are unix-style variables. Let's create our global one with the APP_NAME set to 30days:

touch .env
echo "APP_NAME=30days" > .env

Let's navigate to the exploded config/ directory where we'll see all of our build tool written out for us. We won't look at all of these files, but to get an understanding of what are doing, we'll start looking in config/webpack.config.dev.js.

This file shows all of the webpack configuration used to build our app. It includes loaders, plugins, entry points, etc. For our current task, the line to look for is in the plugins list where we define the DefinePlugin():

module.exports = {
  // ...
  plugins: [
    // ...
    // Makes some environment variables available to the JS code, for example:
    // if (process.env.NODE_ENV === 'production') { ... }. See `./env.js`.
    // It is absolutely essential that NODE_ENV is set to production
    // during a production build.
    // Otherwise React will be compiled in the very slow development mode.
    new webpack.DefinePlugin(env.stringified),
    // ...
  ]
}

The webpack.DefinePlugin plugin takes an object with keys and values and finds all the places in our code where we use the key and it replaces it with the value.

For instance, if the env object there looks like:

{
  '__NODE_ENV__': 'development'
}

We can use the variable __NODE_ENV__ in our source and it will be replaced with 'development', i.e.:

class SomeComponent extends React.Component {
  render() {
    return (
      <div>Hello from {__NODE_ENV__}</div>
    )
  }
}

The result of the render() function would say "Hello from development".

To add our own variables to our app, we're going to use this env object and add our own definitions to it. Scrolling back up to the top of the file, we'll see that it's currently created and exported from the config/env.js file.

Looking at the config/env.js file, we can see that it takes all the variables in our environment and adds the NODE_ENV to the environment as well as any variables prefixed by REACT_APP_.

// ...
// Grab NODE_ENV and REACT_APP_* environment variables and prepare them to be
// injected into the application via DefinePlugin in Webpack configuration.
const REACT_APP = /^REACT_APP_/i;
// ...
function getClientEnvironment(publicUrl) {
  const raw = Object.keys(process.env)
    .filter(key => REACT_APP.test(key))
    .reduce(
      (env, key) => {
        env[key] = process.env[key];
        return env;
      },
      {
        // Useful for determining whether we’re running in production mode.
        // Most importantly, it switches React into the correct mode.
        NODE_ENV: process.env.NODE_ENV || "development",
        // Useful for resolving the correct path to static assets in `public`.
        // For example, <img src={process.env.PUBLIC_URL + '/img/logo.png'} />.
        // This should only be used as an escape hatch. Normally you would put
        // images into the `src` and `import` them in code to get their paths.
        PUBLIC_URL: publicUrl,
      }
    );
  // Stringify all values so we can feed into Webpack DefinePlugin
  const stringified = {
    "process.env": Object.keys(raw).reduce((env, key) => {
      env[key] = JSON.stringify(raw[key]);
      return env;
    }, {})
  };

  return { raw, stringified };
}

module.exports = getClientEnvironment;

We can skip all the complex part of that operation as we'll only need to modify the second argument to the reduce function, in other words, we'll update the object:

 {
  // Useful for determining whether we’re running in production mode.
  // Most importantly, it switches React into the correct mode.
  NODE_ENV: process.env.NODE_ENV || "development",
  // Useful for resolving the correct path to static assets in `public`.
  // For example, <img src={process.env.PUBLIC_URL + '/img/logo.png'} />.
  // This should only be used as an escape hatch. Normally you would put
  // images into the `src` and `import` them in code to get their paths.
  PUBLIC_URL: publicUrl,
}

This object is the initial object of the reduce function. The reduce function merges all of the variables prefixed by REACT_APP_ into this object, so we'll always have the process.env.NODE_ENV replaced in our source.

Essentially what we'll do is:

  1. Load our default .env file
  2. Load any environment .env file
  3. Merge these two variables together as well as any default variables (such as the NODE_ENV)
  4. We'll create a new object with all of our environment variables and sanitize each value.
  5. Update the initial object for the existing environment creator.

Let's get busy. In order to load the .env file, we'll need to import the dotenv package. We'll also import the path library from the standard node library and set up a few variables for paths.

Let's update the config/env.js file

var REACT_APP = /^REACT_APP_/i;
var NODE_ENV = process.env.NODE_ENV || 'development';

const path = require('path'),
      resolve = path.resolve,
      join = path.join;

const currentDir = resolve(__dirname);
const rootDir = join(currentDir, '..');

const dotenv = require('dotenv');

To load the global environment, we'll use the config() function exposed by the dotenv library and pass it the path of the .env file loaded in the root directory. We'll also use the same function to look for a file in the config/ directory with the name of NODE_ENV.config.env. Additionally, we don't want either one of these methods to error out, so we'll add the additional option of silent: true so that if the file is not found, no exception will be thrown.

// 1. Step one (loading the default .env file)
const globalDotEnv = dotenv.config({
  path: join(rootDir, '.env'),
  silent: true
});
// 2. Load the environment config
const envDotEnv = dotenv.config({
  path: join(currentDir, NODE_ENV + `.config.env`),
  silent: true
});

Next, let's concatenate all these variables together as well as include our NODE_ENV option in this object. The Object.assign() method creates a new object and merges each object from right to left. This way, the environment config variable

const allVars = Object.assign(
  {},
  {
    NODE_ENV: NODE_ENV
  },
  globalDotEnv.parsed,
  envDotEnv.parsed
);

With our current setup, the allVars variable will look like:

{
  'NODE_ENV': 'development',
  'APP_NAME': '30days'
}

Now we can add this allVars as an argument to the reduce function initial value called in the raw variable in the getClientEnvironment function. Let's update it to use this object:

function getClientEnvironment(publicUrl) {
  // ...
  const raw = Object.keys(process.env)
    .filter(key => REACT_APP.test(key))
    .reduce(
      (env, key) => {
        env[key] = process.env[key];
        return env;
      },
      {
        // Useful for determining whether we’re running in production mode.
        // Most importantly, it switches React into the correct mode.
        NODE_ENV: process.env.NODE_ENV || "development",
        // Useful for resolving the correct path to static assets in `public`.
        // For example, <img src={process.env.PUBLIC_URL + '/img/logo.png'} />.
        // This should only be used as an escape hatch. Normally you would put
        // images into the `src` and `import` them in code to get their paths.
        PUBLIC_URL: publicUrl,
        ...allVars
      }
    );
  // ...
}

Now, anywhere in our code we can use the variables we set in our .env files.

Since we are making a request to an off-site server in our app, let's use our new configuration options to update this host.

Let's say by default we want the TIME_SERVER to be set to http://localhost:3001, so that if we don't set the TIME_SERVER in an environment configuration, it will default to localhost. We can do this by adding the TIME_SERVER variable to the global .env file.

Let's update the .env file so that it includes this time server:

APP_NAME=30days
TIME_SERVER='http://localhost:3001'

Now, we've been developing in "development" with the server hosted on heroku. We can set our config/development.config.env file to set the TIME_SERVER variable, which will override the global one:

TIME_SERVER='https://fullstacktime.herokuapp.com'

Now, when we run npm start, any occurrences of process.env.TIME_SERVER will be replaced by which ever value takes precedence.

Let's update our src/redux/actionCreators.js module to use the new server, rather than the hardcoded one we used previously.

// ...
const host = process.env.TIME_SERVER;
export const fetchNewTime = (timezone = "pst", str = "now") => ({
  type: types.FETCH_NEW_TIME,
  payload: new Date().toString(),
  meta: {
    type: "api",
    url: host + "/" + timezone + "/" + str + ".json"
  }
});

Now, for our production deployment, we'll use the heroku app, so let's create a copy of the development.config.env file as production.config.env in the config/ directory:

cp config/development.config.env config/production.config.env

Custom middleware per-configuration environment

We used our custom logging redux middleware in our application. This is fantastic for working on our site in development, but we don't really want it to be active while in a production environment.

Let's update our middleware configuration to only use the logging middleware when we are in development, rather than in all environments. In our project's src/redux/configureStore.js file, we loaded our middleware by a simple array:

let middleware = [
  loggingMiddleware,
  apiMiddleware
];

export const configureStore = () => {
  // ...
  const store = createStore(rootReducer, initialState, applyMiddleware(...middleware));
  // ...
}

Now that we have the process.env.NODE_ENV available to us in our files, we can update the middleware array depending upon the environment we're running in. Let's update it to only add the logging if we are in the development environment:

// ...
let middleware = [apiMiddleware];
if ("development" === process.env.NODE_ENV) {
  middleware.unshift(loggingMiddleware);
}
// ...

Now when we run our application in development, we'll have the loggingMiddleware set, while in any other environment we've disabled it.

Today was a long one, but tomorrow is an exciting day as we'll get the app up and running on a remote server.

Great work today and see you tomorrow!

Learn React the right way

The up-to-date, in-depth, complete guide to React and friends.

Download the first chapter

Ari Lerner

Hi, I'm Ari. I'm an author of Fullstack React and ng-book and I've been teaching Web Development for a long time. I like to speak at conferences and eat spicy food. I technically got paid while I traveled the country as a professional comedian, but have come to terms with the fact that I am not funny.

Connect with Ari on Twitter at @auser.