# How I Would Learn DevOps From Scratch in 2026: 9 Free Sites, in Order

By Bobby Iliev, 2026-10-29. https://bobbyiliev.com/post/how-i-would-learn-devops-from-scratch-in-2026/

## Introduction

Hey! I'm Bobby, a DevOps engineer and the author of a few free, open-source ebooks on Bash, Git and Docker.

People often ask me where to start with DevOps. Many "best sites to learn DevOps" lists give you a long list of links in no real order, and a lot of them are video courses. Videos are fine, but you do not learn DevOps by watching someone else type. You learn it by typing the commands yourself, breaking things and fixing them.

So here is what I would do if I started again in 2026: 9 free sites, in the order I would use them. For each one you get what it is good for, one weak point, and one thing to do there this week.

Each site has a free plan that is enough to learn on. I checked all of them and their free plans in October 2026. Free plans change, so check again before you plan around one.

Go through the stages in order. Stage 2 is also the practice you keep coming back to at every later stage.

## The plan at a glance

| Stage | Site | Free plan | First thing to do |
| --- | --- | --- | --- |
| 1. The map | roadmap.sh | Free | Pick one beginner project |
| 2. A terminal with no setup | DevOps Daily | Free | The Linux terminal tutorial |
| 3. Linux and the shell | OverTheWire Bandit | Free | Levels 0 to 10 |
| 4. Git | Learn Git Branching | Free, open source | The Introduction Sequence |
| 5. Containers | iximiuz Labs | 1 hour of playtime a day | Run a container and look inside it |
| 6. CI/CD | GitHub Skills | Free | hello-github-actions |
| 7. Kubernetes | Killercoda | Free scenarios, 1-hour sessions | Deploy, break, fix |
| 8. Terraform | HashiCorp tutorials | Free | The Docker "Get Started" track |
| 9. Troubleshooting | SadServers | 24 of the 33 easy scenarios | "Saint John" |

## 1. The map: roadmap.sh

[roadmap.sh](https://roadmap.sh/devops) is a free, community-made map of what DevOps covers, from the shell to monitoring.

**Why first:** you need to see the shape of the field before you start, so that each new tool has a place to go.

**Weak point:** it is a map, not a course. The full DevOps roadmap is long, and if you try to tick every box before you build anything, you will stall. Use it to see where you are, not as a checklist.

**This week:** open the [DevOps projects page](https://roadmap.sh/devops/projects). It has 26 project ideas, 11 of them for beginners. Pick [Server Performance Stats](https://roadmap.sh/projects/server-stats), a small script that reports CPU, memory and disk use. You will write it after stage 3.

## 2. A terminal with no setup: DevOps Daily

[DevOps Daily](https://devops-daily.com/) has more than 60 free browser games and simulators, so you can type your first commands before you install anything. There are terminal simulators for [Linux](https://devops-daily.com/games/linux-terminal), [Docker](https://devops-daily.com/games/docker-terminal-simulator), [Kubernetes](https://devops-daily.com/games/kubernetes-terminal-simulator) and [Terraform](https://devops-daily.com/games/terraform-terminal-simulator), plus more than 100 [interview questions](https://devops-daily.com/interview-questions) with worked answers, and [quizzes](https://devops-daily.com/quizzes).

**Why here:** the first steps are the easiest place to give up, and a terminal in the browser removes the setup. After that, keep it open next to every later stage: there is a simulator for most of them, and ten minutes a day, or a few interview questions on the topic you just finished, keeps it fresh.

**Weak point:** the terminals are simulated, not real machines. They teach the commands and the flow, but not every edge case. Use them for quick practice, and the real environments in the next stages for depth.

**This week:** do the six lessons of the [Linux terminal tutorial](https://devops-daily.com/games/linux-terminal) (`pwd`, `ls`, `cd`, `cat`, `grep`, `chmod`, pipes). Then add [On-Call: The Daily Incident](https://devops-daily.com/games/on-call) to your day: a different simulated production incident each day, where you find the root cause from logs, metrics and `kubectl` output. Once you have done stage 4, try the [challenges](https://devops-daily.com/challenges): you fix something in your fork, open a pull request, and CI grades it.

## 3. Linux and the shell: OverTheWire Bandit

Most servers and containers you will work with run Linux, and a lot of DevOps work happens in a terminal. [Bandit](https://overthewire.org/wargames/bandit/) is a free game that you play over SSH. Each level hides the password for the next one, and you find it with real commands: `ls`, `cat`, `find`, `grep`, `sort`, `ssh` and more.

```bash
ssh bandit0@bandit.labs.overthewire.org -p 2220
```

The password for the first level is on the [Level 0 page](https://overthewire.org/wargames/bandit/bandit0.html).

**Why here:** it makes you read `man` pages and search for answers, which is a big part of the job.

**Weak point:** Bandit is a security wargame. The later levels go into topics you will not need on day one. Each level page gives you hints and commands to read about, not a step-by-step solution, so you have to work things out yourself. That is the point, but it can be slow.

**This week:** finish levels 0 to 10. Then write the Server Performance Stats script from stage 1. If you want a guide next to you, my free [Introduction to Bash Scripting](https://bobbyiliev.com/ebooks/introduction-to-bash-scripting/) ebook covers the basics.

## 4. Git: Learn Git Branching

[Learn Git Branching](https://learngitbranching.js.org/) teaches Git with a picture of your commit graph that changes as you type. It runs in the browser, and it is free and open source.

**Why here:** every later stage keeps its work in Git: your pipelines, your Kubernetes manifests and your Terraform code.

**Weak point:** it simulates Git. There is no real repository, and it does not cover pull requests on GitHub or GitLab. You get those on a real repository in stage 6.

**This week:** do the Introduction Sequence (commits, branches, merge and rebase), then the remote levels (push, pull and fetch). My free [Introduction to Git and GitHub](https://bobbyiliev.com/ebooks/introduction-to-git-and-github/) ebook has the commands in one place.

## 5. Containers: iximiuz Labs

[iximiuz Labs](https://labs.iximiuz.com/) gives you real Linux machines in the browser, with playgrounds for Docker, Kubernetes and networking. Its tutorials explain what a container is underneath (namespaces, cgroups and image layers), not only how to type `docker run`.

**Why here:** once you know Linux, a container makes sense as normal Linux processes with their own isolated view of the system. That one idea makes containers much easier to debug.

**Weak point:** the free plan gives you one playground at a time and 1 hour of total playtime a day. The challenges and the official courses are in the paid plan.

**This week:** start the [Docker playground](https://labs.iximiuz.com/playgrounds/docker) and run a container. Keep it running, find its process on the host with `ps -ef`, then look at its namespaces with `sudo ls -l /proc/<pid>/ns` (use the PID you found). If the Docker commands are new to you, my free [Introduction to Docker](https://bobbyiliev.com/ebooks/introduction-to-docker/) ebook starts from zero.

## 6. CI/CD: GitHub Skills

[GitHub Skills](https://learn.github.com/skills) courses run in your own repository. You copy an exercise into your account, and Mona, GitHub's Octocat, gives you each step and checks your work as you push. So you learn GitHub Actions with real workflow runs, not slides.

**Why here:** CI/CD is where Git, the shell and containers come together. A pipeline runs steps you already know, like build and test, when something happens in the repository, such as a push or a pull request.

**Weak point:** these CI/CD exercises teach GitHub Actions only. The ideas carry over to GitLab CI and the others, but the syntax does not. Keep the repository public: GitHub Actions is free for public repositories on the standard runners.

**This week:** do [hello-github-actions](https://github.com/skills/hello-github-actions), then [test-with-actions](https://github.com/skills/test-with-actions). After that, [publish-docker-images](https://github.com/skills/publish-docker-images) connects this stage to stage 5.

## 7. Kubernetes: Killercoda

[Killercoda](https://killercoda.com/) runs Kubernetes and Linux environments in the browser. You get a terminal on a real cluster in a few seconds, with nothing to install.

**Why here:** Kubernetes is much easier after containers and CI/CD. You already know what runs inside a Pod, and you know why deployments and rollouts matter.

**Weak point:** a free session lasts one hour, and the environment is gone when it ends, so keep your YAML files in your Git repository. The CKA and CKAD exam courses are paid, but the free scenarios and playgrounds are enough to learn on.

**This week:** open the [Kubernetes playground](https://killercoda.com/playgrounds/scenario/kubernetes). Create a Deployment and expose it with a Service. Then break it on purpose: set the Deployment's image to a tag that does not exist and watch the new Pod. Fix it from what `kubectl describe pod <pod-name>` and `kubectl get events` tell you.

## 8. Infrastructure as code: HashiCorp's Terraform tutorials

The official [Terraform tutorials](https://developer.hashicorp.com/terraform/tutorials) are free and well maintained.

**Why here:** after you have set things up by hand, Terraform shows you how to describe them in code and let a tool make the changes for you.

**Weak point:** many of the tracks need a cloud account, and a cloud account can cost you money if you forget to destroy what you made. Some tutorials also point you to HCP Terraform, HashiCorp's hosted service. You do not need it to learn.

**This week:** do the [Docker "Get Started" track](https://developer.hashicorp.com/terraform/tutorials/docker-get-started). It has seven short tutorials, runs against Docker on your own machine and needs no cloud account. You build, change and destroy a container with Terraform.

## 9. Troubleshooting: SadServers

[SadServers](https://sadservers.com/) gives you a broken Linux server and a task: find out what is wrong and fix it. It is good practice for the day something breaks for real.

**Why here:** by now you know the tools. Troubleshooting is the skill that puts them together, and it is what many interviews test.

**Weak point:** the free plan covers a selection of the scenarios. When I checked, 24 of the 33 easy ones were free. The paid plan has more than 125 scenarios and gives you more time on each one.

**This week:** try "Saint John" from the [scenario list](https://sadservers.com/scenarios). A program keeps writing to `/var/log/bad.log`, and you must find it and stop it without deleting the file. Then try "Saskatoon": find the IP address with the most requests in an access log.

## What I would skip

- **Paying for a bootcamp before you try the free route.** Do stages 1 to 5 first. If you still want a course after that, you will know what you need from it.
- **Collecting certificates before you build anything.** A small public repository with a pipeline, a Dockerfile and some Terraform tells an interviewer more than a badge.
- **Learning three clouds at once.** Pick one cloud provider, learn its basics, and set a billing alert on day one.
- **Tutorial hopping.** Finish one thing before you start the next. The order above is there so you do not have to decide.

## One project that ties it all together

After stage 9, try this as your first portfolio project. Use a sample app if you do not want to write one:

1. A small web app in a Git repository.
2. A Dockerfile for it.
3. A GitHub Actions workflow that tests it and builds the image on every push.
4. Terraform that creates the place where it runs.
5. A Kubernetes Deployment, or a single server if Kubernetes is too much for now.
6. A README that explains how to run it and what you would improve next.

roadmap.sh has more [project ideas](https://roadmap.sh/devops/projects) at every level if you want a different one.

## Conclusion

DevOps has a lot of tools, but the order matters more than any one tool: a map and a browser terminal first, then Linux, Git, containers, pipelines, Kubernetes and Terraform, and then troubleshooting. Every site on this list lets you learn by doing, for free.

If a free site helped you more than any of these, let me know and I will take a look.

Happy learning! 🚀