So here’s my take. I’m not a devops guy professionally. I started my homelab with docker. The problem was the number things I was hosting kept growing and I was worried of loading the machine. I had a few other machines lying around that I decided to pull into a k3s. I somehow love it. My entire home lab is now stored with IaC and lives in a GitHub with CI/CD. Any changes I make to the repo are automatically deployed to the cluster. If I need to Takedown a machine I don’t need to worry about loss of service. I also use velero for backups. If things go wrong a few commands and my entire cluster is fully restored from backups. Now I can easily agree that kubernetes is overkill for a homelab. But I feel it offers some convenience in terms of administration. For docker I still had to deploy everything by portainer which I hadn’t found a way to automate. Backup and restore was not fully automated. You could backup the data but you had to manually redeploy your apps and then restore data to it. At least this was what I could implement. With kubernetes everything is fully in code and controlled by the GitHub repo. Granted the learning curve is steep. Took me 3 months to fully port my system to k3s. Also for general apps check out https://bjw-s.github.io/helm-charts/docs/ You can use that chart to make a helm chart for any app that can be deployed via docker compose. So I just create my own helm charts for apps that only habe docker instructions and deploy it.

TLDR; learning curve is steep but there are a few gains in terms of IaC administration and ability to leverage multiple machines