Why Vagrant? What’s the point?
Having said that, I do stumble across the occasional “must-have” tool (depending on the project), like CodeKit, that truly does improve productivity with minimal impact on my current workflow.
Enter Vagrant – a self-contained, configurable virtual machine for loading and testing websites or applications. Having run MAMP without problems for years, I figured Vagrant would be cool but not useful enough to incorporate. But it is, and here’s why:
- It keeps the same server environment wherever the project goes
- It’s dynamic, and easy to configure to match whatever production server environment I need
- It’s fast to boot up, fast to shut down, and fast to erase (one command for each)
- It maintains its settings and packages outside the virtual machine, so I don’t waste space with several different Ubuntu environments set up for specific purposes.
You may be wondering why you would want to run a virtual machine for local server testing if you have *AMP(P) software already running. The thing is, each installation of Apache, MySQL, and PHP can differ in subtle but important ways. PHP versions may be different, Apache might not have the same configurations on different machines, and so on. And in many places, every developer and designer has a local server stack running on their machine to test and preview websites. This means that code can work on one machine, but not another. Code may work on all MAMP installs in the office, but fail on an Amazon server (I have had this happen before). Git takes care of maintaining the same code base across multiple machines, but doesn’t guarantee person A will see the same results as person B.
We could try to include a Virtual Machine image within each repo, but imagine the overhead in updating those files (averaging several gigabytes).
What if, instead, we could just push a flat configuration file for a server, have some software read it, build a virtual machine environment matched to a project, and boot it up? That’s where Vagrant comes in.
Now, I am by no means a server guru. I have difficulty setting up virtual hosts, remembering the correct commands to restart Apache, and have to keep a cheat sheet for everything I do over SSH. And yet, Vagrant was a breeze to set up. So let’s get into it. All it takes is some basic knowledge of the command line or terminal app, and you can pick that up as you go.
Head over to Vagrant’s Website and download a copy. Once you follow the installation procedures, nothing happens. Yet.
You also need to install virtualization software. I run VirtualBox for everything here. All you have to do is have it installed for Vagrant to work with it. This software basically creates, from system image files, a virtual computer running in it’s own shell, isolated from your host system. Vagrant’s virtual machine runs in the background, so you don’t actually “see” the virtual machine running.
After that, we are set to start booting up a Vagrant server.
Let’s create a project first. I’ll create a project folder called Vagrant for the hell of it. I’ll put it in my Development directory where all my other projects are. Vagrant automatically uses this directory as the root for your virtual machine user – in essence, a synced folder. We will see more of that after setup.
Once the project directory is created, I’ll navigate to it via command line / terminal:
Once I’m there, I will run the vagrant init function. This function takes one argument, where we must specify what type of server (OS, OS version) we want to initialize. Vagrant refers to this as a Box. Typically, this can be hashicorp/precise32, which is an Ubuntu (Linux) server (Box).
vagrant init hashicorp/precise32
It will grab the latest version of this OS and store it outside of the project folder – so you can have one root machine from which to build different virtual server evironments.
Now we run the boot up function:
And bam! The server is running. It may not look like it, but it’s there! To check, run
And it should tell you that the server is running. You can now SSH into it* using
*Note to Windows Users – you will need to install an ssh client to use this command. Git has one built in, but there are many others out there.
From here, you are using the Linux terminal to perform operations. To test that the folder you are currently in on the Vagrant server is the same as the project folder, run a simple touch command to create a file
Now, outside of the terminal window, navigate to that folder in Finder/Windows. You should see your index.html file sitting there.
But this Virtual Machine is just an operating system – it’s not running Apache, or MySQL, or PHP yet. We need to install and configure them. However, we don’t want to set it up for just this instance – we can use Vagrant’s configuration methods to make sure that whenever someone clones our source code, they can grab the configuration file for the virtual server and have the exact same environment we are running locally.
Configuring your Vagrant Virtual Machine
Vagrant configuration files are called Vagrantfile, and they are located in the project’s root directory. Editing them is simple using a text editor. Loading them into the Vagrant machine is called Provisioning, and provisioning only happens on the first run of a Vagrant project. To force a machine to re-provision itself, use the command
vagrant reload --provision
If it doesn’t reload properly.
To make a quick, LAMP-stack server, I first created a bash script called bootstrap.sh in the root project directory that checked a few things and installed a LAMP stack. I then called that script to be run as part of the provisioning configuration. This comes from a great article Getting Started With Vagrant on This Programming Thing.
sudo debconf-set-selections <<< 'mysql-server-5.5 mysql-server/root_password password rootpass'
sudo debconf-set-selections <<< 'mysql-server-5.5 mysql-server/root_password_again password rootpass'
sudo apt-get update
sudo apt-get -y install mysql-server-5.5 php5-mysql apache2 php5
if [ ! -f /var/log/databasesetup ];
echo "CREATE USER 'database_user'@'localhost' IDENTIFIED BY 'database_password'" | mysql -uroot -prootpass
echo "CREATE DATABASE vagrant_database" | mysql -uroot -prootpass
echo "GRANT ALL ON vagrant_database.* TO 'database_user'@'localhost'" | mysql -uroot -prootpass
echo "flush privileges" | mysql -uroot -prootpass
if [ -f /vagrant/data/initial.sql ];
mysql -uroot -prootpass vagrant_database < /vagrant/data/initial.sql
if [ ! -h /var/www ];
rm -rf /var/www
sudo ln -s /vagrant/public /var/www
sed -i '/AllowOverride None/c AllowOverride All' /etc/apache2/sites-available/default
service apache2 restart
You can copy this code directly into the bootstrap.sh file you use, just make sure of a few things:
- The database_user must be changed to a username of your choice
- The database_password must be changed to a secure password
- The vagrant_database must be named whatever you want
To make this script run at startup, I changed my Vagrantfile to look like this:
Vagrant.configure(2) do |config|
config.vm.box = "hashicorp/precise32"
config.vm.provision :shell, path: "bootstrap.sh"
config.vm.network "forwarded_port", guest: 80, host: 80
The last entry before end deals with port forwarding – basically, the port you want to preview the website from outside of the virtual machine. Many people opt for port 8080 for development purposes, but I prefer the standard port 80. It’s up to you.
Now that our provisioning and boot file are set up, head back to the project directory and run
vagrant reload --provision
Just to ensure it really reloads the provisioning files. You should see the Vagrant software outputting quite a bit of downloading and installation commands, and then boot up. Now you are set with a full LAMP stack in a Vagrant Virtual Machine!
From here, just download/clone/etc the latest version of WordPress, put the files in the root of your project, and run the installer.
For more information:
Github Repo of some great provisioning bash scripts