Tristan Dunn

Deploying Jekyll to a VPS

-

Part 3: Deploying Remotely to Digital Ocean

Continuing on part two of the series we’ll create a remote server, deploy to it, and set up a domain name to make it publicly accessible. We’ve already done most of the hard work in the previous parts.

Creating a Server

Create a DigitalOcean account if you don’t already have one. You may have to wait for account verification, but it normally doesn’t take long.

Creating a Droplet

Next we need to create a droplet, which is the DigitalOcean name for a server. You should see the green “Create Droplet” button in the top right of your dashboard. We’ll have to provide details about the server. (DigitalOcean also provides a more thorough guide to creating your first server.)

Droplet Hostname

This name will be the machine’s hostname, so we should use the website hostname we plan to use.

Select Size

The $5 per month size should be fine for a static site, but it’s up to you. If you’re planning to run any other services or background tasks then you may want to consider a larger size. Note that you can adjust the size in the future, but you may experience a bit of downtime depending on how you perform it.

Select Region

For the region you should pick whichever is closest to your target audience, unless you plan on using a CDN such as Fastly.

Available Settings

None of the available settings are necessary.

Select Image

The default machine image of Ubuntu 14.04 x64 is fine and matches our local Vagrant machine. If you’re using a different image locally, try to match it to ease testing of settings locally.

Add SSH Keys

And lastly you should add your public SSH key for root access instead of receiving a plain text password via e-mail. If you don’t already have an SSH key, you can follow this quick guide.

Verification

The server will take a minute or so to boot. After you can verify it’s running by SSHing into it. The IP address for the server is on the detail page you see. If you’re using the SSH option it will prompt you to verify authenticity of the host since it’s your first time connecting, which you can do by entering yes when prompted. If you’re using the password option you’ll need to provide the password e-mailed to you.

$ ssh root@IP_ADDRESS
SSHing into the remote server.

Configuring Chef

We’ll need a SSH host for the new server, which is simple compared to the Vagrant version. You can name it whatever you would like. The username will be root and the hostname is the IP address of your server. Note that you can use a domain name instead of an IP address once you configure the DNS.

Host example        # Name.
  User root         # Username.
  Hostname 1.2.3.4  # Your droplet IP address.
Adding a host to ~/.ssh/config for the remote server.

As for actual Chef configuration, all we need to do is duplicate the nodes/vagrant.json file and name it the same as your SSH host. So if we created an example host the file would be nodes/example.json.

Configuring Capistrano

We created the base configuration in part two, but you probably want to update the repository location, application name, and deployment location in the config/deploy.rb file. See the repo_url, deploy_to, and application variables.

To add a remote target we just need to specify the new remote server and which branch we would like to deploy in a config/deploy/remote.rb file. You can name the file whatever you would like, I just prefer seeing local and remote within the command when deploying to keep it clear.

# Define a web server, where "example" is the name our new SSH host and
# "deploy" is our server user created via Chef.
server "example", user: "deploy", roles: %w(web)

# Support deploying a specific branch, but default to the master branch.
#
# For example, to deploy the "css-fixes" branch:
#   BRANCH=css-fixes cap remote deploy
#
set :branch, ENV["BRANCH"] || "master"

# Optionally define custom configuration files, where the production version
# will overwrite the global version.
# set :configuration, "_config.yml,_config_production.yml"
Defining the host, branch name, and custom configuration in config/deploy/remote.rb.

Deploying to DigitalOcean

First we need to bootstrap Chef on the new server. Replace example with your SSH host created above.

$ bundle exec knife solo bootstrap example
Bootstrapping the remote server with Chef.

If everything went as planned you should see the 404 Not Found error when you visit the server IP address. And now we can deploy the Jekyll website to the remote server. If you named your Capistrano configuration file different than remote then replace it below.

$ cap remote deploy
Deploying to the remote server.

Once complete you should see your website when you visit the server IP address now. If you receive any errors ensure you changed the Capistrano variables mentioned before in the config/deploy.rb file.

Using a Domain Name

We can use DNSimple for creating and configuring a domain name. You can use any service you prefer though. And if you already have a domain somewhere else, you can also use DigitalOcean for DNS.

First we need to add a domain to DNSimple, and register or transfer the domain if necessary. If you already have a domain created elsewhere you’ll just need to update the nameservers to point to DNSimple.

Once added we need to add the DNS records. Select DNS from the left menu, then Manage Records under the Custom Records section. We’re going to add two A records. Select the Add Record dropdown, then the A option. For the first record leave the name blank, enter the server IP address for the address, and choose a TTL value. For the second provide the same values, but use www as the name.

Depending on how fast your DNS updates, which could take hours, you should see the deployed website on the domain name. You could also try using different DNS nameservers that update faster, such as Google’s Public DNS.

Summary

We now have all the components for deploying a Jekyll website to a remote server and be available to the public. See the jekyll-vps-server repository for the complete Chef source code, with the part-3 branch being specific to this article. The website source code is available in the jekyll-vps-website repository, with the part-3 branch being relevant.

In the next part we’ll add an asset pipeline and improve configuration for serving the assets. E-mail me if you have any tips, comments, or questions.