Why I Migrated from k8s

At the time of this writing, my blog is still hosted with Vercel, however I think I will eventually move it to my VM. However, all the other sites and services I host have been migrated to my new VM (at the time of writing, only Nextcloud hasn’t been migrated, but it never made it into Kubernetes anyway). Why the switch? Well, I started using kubernetes after I took a job as a consultant doing SRE work. I wanted to have my personal infrastructure mirror the work I was doing professionally. Now that I am not doing that kind of work anymore, there is less motivation to keep a K8s cluster going. Additionally and primarily, I had a very hard time figuring out how to migrate Nextcloud to Kubernetes due to the amount of data our family Nextcloud instance already had and decisions I had made about the Nextcloud setup initially. So I opted to keep it on a VM and migrate everything else to K8s. Well, that costs a lot of money, I decided to cut costs and that meant either looking into moving Nextcloud to K8s or moving K8s back to a VM.

The decision was made for me when I went to update my resume and noticed that fluxcd was no longer automatically updating docker images when new ones were available. I looked into the problem for an hour or so but eventually just decided this was the writing on the wall and it was time to move back to a VM. I did the migration in a morning. It took about two hours, most of that two hours was spent trying to get my ansible+docker-compose stuff working. Technically that all worked just fine, but I could not get the dockerized Traefik proxy server to work. When I looked into alternatives I found that Caddy had matured a lot since I was last shopping for a reverse proxy. I tried the dockerized Caddy and it didn’t work, and then upon reflection I decided it might be a good idea to run my server on the OS level and then proxy to the container with ports.