• Deploying a Node-Express-Mongoose App/API with Heroku & MLab

• Describe the difference between single-server and multi-server configurations for an application built in the MERN stack
• Install and use cors to allow for Cross-Origin Resource Sharing between our front-end application and back-end API
• Use npm run build to compile our front-end application into an optimized and deployable directory
• Describe the process of unifying our front-end and back-end code into a single project
• Set up our front-end development server to proxy requests to our back-end server
• Set up our back-end API to serve static build assets in production

To integrate React with a back-end framework (such as Express) we will need to make a few decisions about the desired architecture of our application. The first primary decision to make is where we want our front-end application to be served from. There are two primary options:

Multi-Server (Microservice) Configuration - our front-end application and back-end API will be completely separate. They will each have their own Git repositories and will be deployed independently to different servers. Our front-end will retrieve data from our back-end across separate domains.

• Pros
  • Separation of Concerns: Our application is more modular and our front-end and back-end developers can work on their parts of the application independently of each other.
  • Specialized Configuration: Since our front-end and API will be housed on separate servers, we can optimize server configuration for each one independently, allowing them to scale independently of each other and preventing them from competing over the same server resources.
• Cons
  • Since requests will be sent across separate domains, we will have to configure CORS for the browser to allow our front-end to retrieve data from our back-end

Single-Server Configuration - our front-end application will be housed in the same repository as our back-end code. They will be deployed together to a single server where our back-end will be responsible for serving up our front-end application in addition to our API data.

• Pros
  • Single Deployment: Since our front-end will be served up by our API, we only need to manage one deployment.
  • Unified Codebase: Since our front-end and back-end will be housed in the same Git repo, we will always have a more full picture of the application than if they were in separate repos.
  • No CORS: Since our front-end application will be making requests back to the same domain that served it up (its 'origin'), we do not need to configure CORS.
• Cons
  • We will have to configure our development environment to allow us to use the features of our Webpack server provided by react-scripts (automatic compiling, linting, hot reloading, etc.).
  • We will have to modify our server to serve up both json data for the API endpoints and the assets of our front-end which could require additional configuration and cause the two to compete over server resources.

Today, we will walk through setting up and deploying our application up first on separate servers, then examine how to combine them into one project and deploy onto a single server.

Today, we will be using the React Translator app that used the IBM Watson API to translate text and provide audio pronunciations. We will also be using a Mongoose / Express back-end to allow for users to save translations.

1. First, clone down the back-end, install dependencies, and open in atom:

   git clone [email protected]:ga-wdi-exercises/react-translator-api.git
   cd react-translator-api
   npm install
   atom .

2. Check to make sure mongod is running.

   Type in mongo and if it drops you into the mongo shell, you're good to go. Otherwise, open a separate tab in your terminal and run mongod. Leave this tab running for the rest of this exercise.

3. Run the seeds file to populate our MongoDB database with translation data.

   node db/seeds.js

4. Start the server

   npm start

   If you inspect package.json, you will see that this is an alias for node index.js

5. Navigate to localhost:3001/api/translations to see the json our back-end is serving

1. In a separate terminal window, clone down the React Translator app, checkout to the mern-starter branch, install dependencies, and open in atom:

   git clone [email protected]:ga-wdi-exercises/react-translator.git react-translator-front-end
   cd react-translator-front-end
   git checkout mern-starter
   npm install
   atom .

2. Start the development server:

   npm start

   If you inspect package.json, you will see that this is an alias for react-scripts start

3. Navigate to localhost:3000 to explore the application

Currently we are using this type of architecture. Our back-end is running on localhost:3001 while our front-end is running on localhost:3000. One way to say this is that these applications have different "origins". One issue with this is that our browser is not going to like requests from our front-end (served by localhost:3000) to our back-end on localhost:3001. In the application, navigate to the Saved Translations view and check the console. You should see:

XMLHttpRequest cannot load http://localhost:3001/api/translations. No 'Access-Control-Allow-Origin' header is present on the requested resource. Origin 'http://localhost:3000' is therefore not allowed access.

This is Chrome telling us that since our back end is running on a separate port ("origin") than our front end, our front end is not allowed to retrieve data from it. To fix this, we need to configure CORS (Cross-Origin Resource Sharing) on our back-end.

1. Stop your Express server, and install cors via npm:

   npm install --save cors

2. In index.js, require the cors module and integrate with Express:

   const cors = require('cors')

   app.use(cors())

3. Restart your Express server:

   npm start

Now when you navigate to localhost:3000/translations, you should see a list of the translations being retrieved from our API.

Note: The default cors configuration (above) will allow requests from any origin (which may or may not be ideal). To more precisely control access to our API, we would need to do little more configuration. Check out the cors documentation for more information on this process.

Now that our front end and back end are communicating properly, let's look at how we would deploy them.

As our back-end will need to be hosted on a server and host a database, we will want to deploy it to a hosting service like Heroku. We would go about this is no differently than we have previously:

heroku create react-translator-api
git push heroku master

Note: Heroku will require some additional configuration to host a MongoDB database. It will prompt you to set up an account with a cloud hosting service mLab. For details on this, check out the Heroku Documentation.

Currently, when we start our React application locally, we are using react-scripts. react-scripts is the black box that contains all of the major dependencies and configuration that we are using in development:

• Dependencies: Babel, Webpack, and ESLint and other dependencies that allow the use of JSX, hot reloading, compilation of our application, and other features.
• Configuration: Configuration files for the above dependencies that control (among other things) how our app is compiled in both development and production environment.
• Scripts: Commands that allow us to do things like start our Webpack server (react-scripts start) or create a compiled, minified version of our application (in vanilla JS) that can be deployed (react-scripts build).

The command that we will need to use for deploying our front-end is react-scripts build. create-react-app automatically aliases this for us in our package.json as npm run build. In the terminal, run this command to create a deployment-ready version of your app:

npm run build

This will create a directory called build in the root of your application. This directory contains all of your application's HTML, minified JS, and minified CSS in a deployment-ready format.

Now all you would need to do is to deploy the build directory to a static asset hosting service (such as GitHub Pages or surge.

If pushing to gh-pages:

git subtree push --prefix build origin master:gh-pages

Note that subtree allows us to only push a sub-directory of our repository instead of the entire app. More Information

If using surge:

npm install -g surge
surge

Surge is a CLI based npm package that lets you quickly deploy static front-end applications for free.

Now let's look at how we could consolidate our front-end and back-end into one project and deploy them together to one server. First, let's copy our React Translator app into our API repo:

From the root of the react-translator-api directory, run:

cp -r <path to the `react-translator` directory> ./client

Note that we are both copying the front-end repo to our root folder while simultaneously renaming it to client

Now, in a deployed production environment, we want our express server to serve up the static assets in client/build. In index.js, let's set up a route to do this:

In index.js

app.get('/', (req, res) => {
  res.sendFile(__dirname + '/client/build/index.html')
})

Now when we navigate to localhost:3001, we will see errors in the console saying that the browser cannot load our minified JS and CSS files. To fix this, we need to tell express where to look for static assets:

In index.js

app.use(express.static('client/build'))

Reload the page and we will see that our app is working as intended. However, before we deploy, we will need to update the paths of any requests in our front-end to be relative instead of absolute since we are now serving both the app and data from the same origin.

For example, in client/src/components/Translations/Translations.js, the url axios is querying should be changed from

axios.get('http://localhost:3001/api/translations')

to

axios.get('/api/translations')

Hint: use CMD + SHFT + F to find all instances of a url in a project

In order for this change to be reflected via localhost:3001, we would not have to run npm build again since our Express app is serving assets out of the client/build directory. This is cumbersome and not something we should have to do every time we make a change to our front-end code.

To use hot reloading and linting like we're used to, we would have to change into the client directory and run npm start to start up our Webpack development server on localhost:3000. This would be fine but it would break all of our relative paths that we just defined above for requests (the API is at localhost:3001/api/translations not localhost:3000/api/translations). How do we get around this?

One easy solution is to allow our Webpack server (on localhost:3000) to proxy requests for our API server (on localhost:3001). What this means is that if a request is made to localhost:3000 at a path that it doesn't recognize, it will simply forward that request to localhost:3001, and then relay the response back as well. In essence, our Webpack server will serve as a middleman between the browser and the API only during development so that we can use all of the bells and whistles that create-react-app provides to us.

In client/package.json, add this line:

"proxy": "http://localhost:3001/"

Then run npm start within client and navigate to localhost:3000 in your browser. Our data is being successfully retrieved!

We will still need to run npm run build inside of client whenever we want to update our static production code in build, but for quickly iterating in development, we now can quickly see our updates rendered.

Since we are only going to need to start up both our Express server and our Webpack server in development, we might want to set up a script to run both of these process concurrently:

Install the concurrently package via npm:

npm install --save-dev concurrently

In package.json, lets add a command to the scripts object:

"scripts": {
  "start": "node index.js",
  "test": "echo \"Error: no test specified\" && exit 1",
  "dev-start": "concurrently \"npm start\" \"cd client && npm start\""
},

Note that && allows us to chain commands in bash

Now, to start up both servers for development, all we have to do is run:

npm run dev-start

Microservice Solution

• Back-End: https://git.generalassemb.ly/ga-wdi-exercises/react-translator-api/tree/microservice-solution
• Front-End: https://github.com/ga-wdi-exercises/react-translator/tree/mern-starter

Single-Server Solution

• Back-End: https://git.generalassemb.ly/ga-wdi-exercises/react-translator-api/tree/single-server-solution