Many people I talk to with a history in system administration or engineering want to learn about DevOps but don’t know how to get started. This post lays out some steps to create a simple website and follow many of the same tools and workflows I’ve used over the years to get things done. If you’re able to get through these steps with the help of a mentor, you’ll be well on your way to be able to be proficient in the world of DevOps.

I recommend you follow the phases below in order. Don’t skip any steps. And if you get to a point where you’re thinking “I don’t know anything about git!” or whatever, have no fear. Stop, spend some time learning git, and then keep going. You can do this project without any previous technology experience, assuming you have someone help you look in the right places. And that’s the final thing: use this as a guide to get you on your way but find a technical friend to sponsor your learning and interact with you. That will be huge.

We’ll deliver the project in phases:

Phase 1: Simple Website

In order to work well with DevOps automation and tooling, you need to know the basics about source control and editing code. This first phase gives you that foundation by creating a basic website on GitHub.

1. Create an account on https://github.com/
2. Create a repository named website on GitHub
3. Clone your website repository locally using your terminal (Terminal on a Mac or PowerShell on windows).
4. Create a branch called feature_index
5. Add a file index.html to the branch that says Hello, World! in the text. You’ll want to make the change with Visual Studio Code.
6. Check that in
7. Create a pull request for your branch to be merged into master and assign the pull request to your mentor. If you don’t have a mentor, assign it to me, mhedgpeth! 🙂
8. Merge your pull request into master after it’s approved

Resources for Phase 1

• Try Git in 15 minutes
• Getting Started with Visual Studio Code
• Codecademy Learn HTML

Phase 2: Simple Webserver

Now that you have a website, you want to now create a server to host that website. This is a machine that will provide other machines with the contents of your index.html file from Phase 1. This machine will be a Linux machine hosted on Microsoft Azure Cloud with nginx running on it.

1. Create an account on Azure and activate your free trial
2. Create an Ubuntu virtual machine on Azure
3. SSH to that machine. If you’re on Windows use Matt Wrock’s guide to getting an ssh client.
4. Set up nginx on the machine
5. Clone your git repository to /var/www/html on the server
6. Using the public ip assigned to your Ubuntu server, access the website (i.e. http://[your-ip]). Your website should show up.
7. Have a friend on another computer do this. They should see the website too. Magical!

What We’re Learning

My friend Nathan Harvey has said to me that you can’t automate that which you do not understand. If you’re going to do the DevOps but can’t do it manually, then you’re going too fast! So we’re learning the basics here of setting up a machine.

Resources for Phase 2

• Setting Up Linux Virtual Machine on Azure
• Nginx Tutorial
• Nano Tutorial for editing text files in an ssh session

Phase 3: Deploy a Change

Now it’s time to deploy a change! Let’s do this the old-fashioned way the first time around.

1. Create a new branch on your repository named feature_myname.
2. Update index.html to say Hello, Michael!
3. Create a pull request, get it reviewed, and merge it
4. On your webserver, update your index.html file from GitHub.
5. Get on Twitter and send me a message @michaelhedgpeth with a link. Show me that I didn’t waste my time! 🙂

What We’re Learning

This is how a majority of people have deployed things for a long time. The manual way. You can’t realize what problems different tools solve before you do it manually. I’m purposely not giving you the exact steps to follow because I want you to find those steps on your own or with a mentor. Google it! This is how you learn.

Phase 4: Create a Chef Cookbook to Automate Machine Setup

Now think about what your life would be like if you had to do the above steps thousands of times on hundreds of servers. Your life would not be fun and in addition to that you would have hundreds of servers that were all a little bit different because, over time, you weren’t consistent. This is why you use something like Chef.

This is a big technical jump from the previous phases. Take your time here. If you’ve never done this stuff you might spend a couple of weeks on it. You’ll have questions. Get on the Chef Community Slack and post them on the #general channel. Don’t be afraid; we all were here and support you!

1. On GitHub create a repository called my_website.
2. Clone that repository locally and switch to a branch called feature_nginx
3. Set up nginx using the package resource
4. Use Test Kitchen to make sure your cookbook installs the package. Use kitchen with a Policyfile. There will be a Policyfile.rb in the cookbook that defines your run list if you’re doing this right.
5. Write an Inspec Test that ensures the package is installed
6. Create a PR and get it merged
7. Create another branch called feature_website
8. Using the git resource clone your repository on GitHub
9. Write another inspec test to make sure that the website is served when you go to http://localhost and that the contents contain Hello, Michael!
10. Make sure that kitchen test works
11. Create a Pull Request and Merge into master

Resources for Phase 4

• Learn Chef Rally. If you’re new to Chef, spend a lot of time here. Like days. Get through the basics and what I write above will make a lot more sense
• My Policyfiles Post. My description of that feature. Do yourself a favor and use it. It will simplify your life.
• Annie’s InSpec Tutorials. Still the best place on the internet to learn InSpec.

Phase 5: Deploy Your Chef Cookbook to Azure

Now it’s time to use your Chef Cookbook as a way to not have to manually deploy and update your machine. After you do this step, you’ll be able to consistently create as many machines as you want, with very little effort!

1. Create a new virtual machine on Azure using the portal
2. Create an account and organization on manage.chef.io. If you did the learn Chef rally above, you should already have this set up.
3. Push your cookbook policy to the chef server (see my blog post for the exact command).
4. Bootstrap your node with the Chef Server. This will run the policy on that node, setting up nginx and everything!
5. Hit the public IP associated with the node and see that it serves your website perfectly! Magic!
6. Now deploy a new change to your website and call it Hello, Automated World!. Tell me about it (@michaelhedgpeth on Twitter). I’ll give you a high five for getting this far!

What We’re Learning

Hopefully, you see the benefits of automating the setup of the machine. From now on, everything is consistent and just works. Your drama for making changes goes down. And you can do this as many times as it’s called for!

Phase 6: Automate the Creation of VMs in Azure with Terraform

Now that we have a Virtual Machine that we can easily set up, you might be tempted to quit. But there’s something lingering there: even though you automated what was inside the virtual machine, you still have to manually set up the machine itself. This is easy enough when you have one machine, but what if you have hundreds? That’s where Terraform comes in:

1. Create a new repository in GitHub called website_provision
2. In that repo, create a terraform script that creates a virtual machine within a network with an external ip address
3. Make your Terraform script bootstrap the VM with your Chef Server and assign it to your policy
4. Watch with awe how Terraform allows you to create and destroy your entire stack, all the way from the machines themselves to what’s on the machines (with Chef).

What We’re Learning

The environment within which your application runs is one of the most complicated aspects of the application. Thus, you should invest in automating that and keeping yourself away from the user interface. Terraform is a great tool for this.

Phase 6 Resources

• Terraform Resources
• Azure Terraform Provider
• Azure Terraform Examples

Phase 7: Workflow Automation with GitHub Actions

Now that we have an automated process that will deploy our stuff, we want to make our workflow easy to execute with GitHub Actions.

1. Create a rakefile for your my_website cookbook using my blog post as a guide.
2. Add a GitHub Actions build script
3. Create a branch called broken and add code that will break your cookbook
4. Check it into the branch, and watch GitHub Actions tell you it’s broken!

What We’re Learning

We’re automatically providing the accountability people need to know that their software still works. It’s better to learn that as you’re making the changes rather than when they get to production. So GitHub Actions helps us see how, when we change software over time, that software still meets our expectations. It reinforces the best practices for your team: you should run test kitchen before checking in changes. The safety imposed by GitHub Actions will keep you on the straight and narrow path!

Resources for Learning

• GitHub Actions for Chef Cookbooks

Extra Credit

I think if you do the above project, you’ll be well on your way to getting things working. Here are some other ideas, for extra credit:

• With Hashicorp Vault store a secret and have your Chef cookbook (using the vault gem) read the secret and write it to your web page
• Create a GitHub Actions pipeline that will deploy your Terraform templates
• Instead of using Chef, use Docker to deploy your application

Conclusion

This project will give you the context to know the basics of automating infrastructure. With this basis under your belt, you’ll be able to make great progress in whatever situation you find yourself.