Guest Blog - Managing Infrastructure on Digital Ocean with Terraform, Mark Provan


As a general fan of the approach “automate all the things” I decided it was time to do a little more work on this blog. I wanted to be able to setup my blog, from the datacentre to browser in one command. I decided to go with Terraform as a method for doing this, as it looks like a really well-documented tool. Let's get started!


I use macOS, so I’ll be using brew, checkout Installing Terraform for guidance on your own system


Before we can create any resources, we need to configure our provider. You can use many providers in one script, but at the moment we just need Digital Ocean.

This tells Terraform how to authenticate with my Digital Ocean account. You can create an API token here. The first part of the above code defines a variable named do_token. By not assigning a value, we have two options. Firstly, we can create a terraform.tfvars file and this will be loaded at runtime. Secondly, we could leave it blank and Terraform will prompt us for the value. We’ll go with the first option. Create terraform.tfvars and add the following content. It’s best not to commit this file to version control, just to be safe.

Now that we have our token setup, we can create the setup for the digitalocean provider. At this point, we’ll need to run terraform init to download the Digital Ocean package for Terraform.

Creating a Server with SSH Keys

We’re now going to create a new server, with accompanying SSH keys so that we can log in to it if needed. To do this we can use the Digital Ocean provider we initialized above.

We’re doing a few things here. Firstly we’re creating two tags to use when provisioning the box, blog and personal. This is purely for me to be able to organize my server list in the Digital Ocean dashboard. This is where Terraform gets really cool. We can use ${} to access the ID given to the tag by Digital Ocean, to pass to the API when creating a new server. You can see we also do this for the SSH key ID. Being able to access information about other resources becomes super useful when we want to automate a few things that depend on each other. Speaking of dependencies, notice the depends_on array, which currently contains a reference to the SSH Key Resource. This means that Terraform won’t attempt to create this resource until the SSH Key has been setup.


Now that we have defined our server, let’s create a DNS entry on Cloudflare, to point my domain name to it.

Here we are creating a new provider instance for Cloudflare (remember to run terraform init to install the package) and defining two new variables that need to go into our variables file. Lastly, we create a new A record on my domain name, that points it to the IP address of the server we created. For reference, each resource in Terraform lists which variables are exposed in the documentation here’s a Digital Ocean Droplet for example.

Provisioning Software

Now that we have a server running, we want to actually do something with it. As I already have my blog setup to run using Docker, I want to upload my docker-compose.yml file and run it. Let’s modify our server resource to make use of provisioners, which let us run commands on it.

I’ve created two provisioners here, one of type file to handle copying my local docker-compose.yml file to the server and a remote-exec provisioner which executes commands on the server, in this case, docker-compose. I’ve also introduced a new variable for my local SSH key. It’s used in the connection block by Terraform to define how the server is accessed to run commands. I’d like to be able to set that as a global config, for all SSH connections, rather than define it twice, but I’m not sure how to do that at the moment. These provisioners are run in order, so my Docker file is uploaded first before it is run.

Now that we’ve got all this defined, we can simply run terraform apply and type yes to confirm the changes when prompted and we are up and running! Terraform is a really handy tool to have in your arsenal and I’m looking forward to using it more in future.

This blog on managing Infrastructure on Digital Ocean with Terraform was written by Mark Provan, Software Engineer in Voice + AI with the BBC. You can follow Mark on Twitter and read more of his blogs here

Mark works with the world's first public service voice assistant, as part of the Voice + AI team at the BBC. In a past life, Mark was a Ruby Developer working in E-Commerce. He has a passion for how teams work, most notably pair programming and Infrastructure as Code. Outside of work, he loves camping with his partner, cats and cycling

Mark recently spoke at September's JavaScript Exchange event on "Automating Infrastructure in TypeScript with AWS CDK" - you can view his full talk here

If you're interested in learning more about our monthly events, simply join your local DevOps Exchange or join the JavaScript Exchange community on Meetup.