I’m Lucas Käldström from Finland. I speak Swedish as my mother tongue in the same manner as 300,000 others in my country do. I just turned 17 and am attending my second year in general upper secondary school. In my spare time, I play soccer, program, go to the gym and read a good book.
I’ve always been interested in Math, and am quite good at it as well. So when I was about 13, I started to become interested in programming. I found it interesting because I could command the computer to do nearly anything. I’ve always loved creating things and it was fascinating to see that every change you make can make a difference in a good way. I started creating small programs with VB.NET and C# and about a year later switched to Node.js and web programming (HTML, CSS and JS). At this point I started to feel the power of open source and what it could do. Also, I made myself a Github account in order to be ready if I found a project to contribute to.
In the beginning of May 2015, I first noticed Kubernetes. I got so excited that I could use something Google has designed free of charge! Unfortunately, I did not have any normal Linux hardware I could use at the time. However, I had two Raspberry Pis that I had been tinkering with a little bit. My Bash skills were practically non-existent and most of the time I was scared of typing something into the command line; however, I realized that Raspberry Pi in fact is the ultimate tool to use when teaching Kubernetes to someone with little cloud computing experience as the cluster becomes really practical. You literally get the “hands-on” experience that is so valuable. This later became the main theme for a 163-page master’s thesis paper Martin Jensen and Kasper Nissen wrote. Likewise, Ray Tsang has been travelling a lot with his Raspberry Pi cluster as well, but now I am getting ahead of myself.
After a lot of hacking in May, I got Kubernetes running on my Raspberry Pi, but it was quite pointless as Docker on ARM had a bug that made it impossible to run any containers via Kubernetes. I continued to improve my scripts during the summer, while not playing soccer or doing something else, like swimming! In August of 2015, I tried the same programming with the v1.0.1 Kubernetes release, and I got it working! That was a truly amazing feeling. I quickly started to expand the context to make it more generic, reproducible and faster. In mid-September, I had it working well, and noticed that a Glasgow University group had done the same thing; both of us working beside each other without knowing the other one.
I knew my work could help others, so I quickly published the source I had to the world; it was the right thing to do. I wanted to help more people run, test and learn from a Kubernetes cluster running on small Raspberry Pis… as well as other devices. This project is known as Kubernetes on ARM. After that, I continued to make lots of improvements on it, with feedback from others suddenly! One of the best moments was when someone reported the first issue and showed interest in helping to improve the project. I was part of the open source community!
I wanted to make my work even more widespread and bring it to the core. And so I did. In November-December I started making myself more familiar with the source code, the contributing process, etc.
On December 14, 2015, I got my first Pull Request merged. What a great feeling! I admit it was really small (a removal of 6 chars from a Makefile), but it was a big step personally to realize that the Kubernetes maintainers wanted my contributions. From January-March, I focused on getting the Kubernetes source code to cross-compile ARM binaries and release them automatically. Kubernetes v1.2.0 was the first release that shipped with “official” Google-built ARM binaries.
I then started to focus on getting an official deployment method multiarch-ready. I chose docker-multinode. The result of that work made Kubernetes v1.3.0 ship ARM and ARM 64-bit binaries and corresponding hyperkube images, which made it possible to run a cluster on different architectures with the same documented commands.
In April 2016, I was added to the Kubernetes organization. I couldn’t believe it! I was one out of about 170 at the time. One week later, I became a Kubernetes Maintainer, with a big M! It was totally crazy! I was 16 and got write permission to several repositories! But with great permissions comes great responsibility, and I always have been looking to the projects’ best when reviewing something for instance. In fact, maintainership should be a very important but boring task.
That same month, I noticed that a project called minikube was added to the Kubernetes organization. The repo was empty, just a markdown file, nothing more. I noticed Dan Lorenc started committing to the repo and I thought it would be cool to improve my Go skills by starting a project from scratch, so I started working with him. I became a maintainer for minikube as well, and am still the 6th-highest contributor to that project by commits.
I continued to contribute to Kubernetes during the summer and I worked on minikube until release v0.5.0 was out. Then I switched focus to improving the Kubernetes deployment experience. sig-cluster-lifecycle added me to their group and it turned out to be a great fit for me and my interests. Subsequent Kubernetes work includes:
- Wrote a proposal (a design goal) for Kubernetes about how multi-platform clusters should be supported, set up and managed. It got merged in August and is one of the guidelines for how Kubernetes should be improved.
- Making Kubernetes easier to install in the sig-cluster-lifecycle group. We have made a tool called kubeadm that makes the setup process much easier than before. This is a crucial effort in my opinion, since right now it is hard for new users to know what deployment solution to use and how they relate to each other. kubeadm will be a building block all higher-level (turn-key) deployment tools can use. This way we are hoping we can somewhat standardize and simplify the Kubernetes deployment story. I’ve also written a document about what we’d like to do in time for v1.6, please check it out if you’re interested.
All in all, I’ve learned a lot thanks to being able to contribute to Kubernetes. And it has been a lot of fun to be able to actually make a difference, which now has led to that I have more than 150 commits to the Kubernetes organization in total. My Kubernetes journey started with a Raspberry Pi, and we don’t know how it ends.
One thing to remember is that I’ve never taken any computing or programming classes. This has been my spare-time hobby. I’ve learned practically everything I can just by doing and participating in the community, which really shows us the power of the internet combined with the will to create something and make a difference.
I am not monetarily paid for my work on the Kubernetes project, but I have gained knowledge, experience, better English communication skills, respect, trust and so on. That is worth more than money right now. Also, I got about five full-time job opportunities during the summer from large worldwide companies without even asking for a job or listing myself as job-searching.
What I’ve really enjoyed while coding on Github is my partial anonymity. My full name has always been there, my email, my location, etc., but I haven’t written that I’m just 17 (or 16 at the time), so people on Github haven’t judged me for being a minor or for not having been to University or for the fact that I’ve never taken a computing class or worked for a big tech company. The community has accepted me for who I am, and respected me.
That’s the power of open source in my opinion. Regardless of who you are when you’re away from the keyboard, you are allowed to join the party and have fun with others like-minded and make a difference together. Diversity is the true strength of open source, and I think diversity scholarships like the one the Cloud Native Computing Foundation provided me to attend CloudNativeCon/KubeCon 2016 in Seattle are a powerful way to make the community even stronger. Thanks!