Michele Nasti

Thoughts on what I learn

Handling configuration for multiple environments in NodeJS

Imagine you have a NodeJS app you're writing, and this app runs on several different environments:

  • on your developer PC, it should use some environment variables (e.g. database connection to localhost, port to use 3000...)
  • when you push your code you may want to run some continuous integration & deployiment on a test environment, so you have to configure new environment variables: database now points to 192.168.xx.yy, port is 8000...
  • Finally, on the production environment you use the official values for these configuration.

How do you handle this? There are several ways to accomplish this task, I wanted to replicate the simple, easy solution provided by Laravel in NodeJS.

How does laravel config works

In Laravel you create as many env files you want, for example:

  • .dev.env, that contains configuration for development environment

  • .test.env, for test environment

  • .prod.env for production

  • Then there's a last file called .env that contains a single variable:

    APP_ENV=dev (or test, or prod).

We're in NodeJS, how do we simulate this behaviour?

a Node solution

First, our configuration files are js files:

  • .development.js
  • .test.js
  • .production.js

Naming of files has changed a little bit, we'll see why in a while.

Let's see the content of the .development.js test file:

//.development.js 

module.exports = {
API_URL: 'api_url',
API_KEY: 'api_key',
SECRET_KEY: 'secret_key'
}

.production.js and .test.js will contain an object with the same keys but different values.

The environment variable NODE_ENV

Since Express became the most popular application server in Node, the NODE_ENV variable has became popular too. In Express, app.get('env') is used to retrieve the execution environment. Many popular platforms, like Heroku, set this variable to production and you can also change to your needs, too.

app.get('env') is pretty much implemented like this:

return process.env.NODE_ENV || 'development'

So you know that if the variable is not set, it is defaulted to development.

Let's write a solution that is not express-dependent.

Here's the configuration.js file:

//configuration.js 

const env = process.env.NODE_ENV || 'development'

module.exports = require(`../.${env}.js`)

Based on the NODE_ENV variable, we will pick up the right .js file.

And now let's test this. Run node index.js (should start with development environment):

//index.js

const configuration = require('./configuration')

console.log('configuration: ', configuration)
// =>
// configuration: { API_URL: 'api_url',
// API_KEY: 'api_key',
// SECRET_KEY: 'secret_key' }

How to start with other configurations?

Either set this in package.json:

 "scripts": {
"start": "node index.js",
"start:test": "NODE_ENV=test node index.js",
"start:prod": "NODE_ENV=production node index.js"
},

and run with npm run start:test; or set the NODE_ENV variable directly on the server (docker, heroku, aws... they all allow you to set env variables).

Some security concerns

This approach is very easy to use but you should not versionate .production.js file, as it will contain sensitive data. In your build process consider injecting this file from some other secure location.

An alternative could be to use environment variables only for the production file:

//.production.js 
module.exports = {
API_URL: process.env.API_URL || 'some_value' ,
API_KEY: process.env.API_KEY,
SECRET_KEY: process.env.SECRET_KEY,
...
}

The good thing about this approach is that you can even mix the two things or swap them altogether.

Happy coding with your configurations ;)