raspberry-pi-dramble VS k3s

https://www.pidramble.com - but you do need to power the router. 5th port is an uplink

Don't bother. Just put them on an open stack and blow a quiet Noctua fan sideways through the stack. I used to have a four-pi4 cluster ala https://www.pidramble.com and one fan blowing back to front was easily good enough. I also removed the fans from the official PoE HATs since they sounded like rocket engines no matter what I did.

There's no rack, it's simply on the corner of my desk. I haven't made photos of it but it looks pretty much like Jeff Geerling's Dramble: https://www.pidramble.com/

Had a pi 'dramble' just like this one and it worked fine. I ran k3s at the time on it. Since it was a lab experimentation thing, I didn't throw more hardware at it, but I could have likely gotten similar results with a bunch of Vagrant/VirtualBox VMs on one NUC.

Automate the Pi Dramble website I run in my basement with Drupal Pi.

How is latency and stuff? Any slowness that you wouldn't expect from the software that may come from running it on a rpi4? I've been thinking of setting up a few services, but I kinda want to do something like the Pi Dramble, wasn't even aware that a single pi would have the power to run all of these services. Though I'm not actually surprised it does.

A couple years ago when I thought I'd have to get into kubernetes for a living, I built a 4x4GB pi4 cluster with the exact parts list from Jeff Geerling's "dramble" setup (link) and did a bunch of experimentation with k3s as a lightweight cluster solution. Worked pretty amazingly for things like "spin me up 10 more of those things" and for moving things around to upgrade the SD card on the 3rd pi while keeping the services running on the other pi. But I don't need to do kubernetes for a living so the cluster is just sitting here collecting dust powered-down basically.

I see this kind of question asked a lot when people are posting their Kubernetes Pi clusters. The rpi 4 quad core with 4-8GB RAM gives a lot of potential for the price in a kuberenetes setup I feel. I found the pi dramble project interesting, and even more the turing pi using rpi compute modules. Are NUCs really a cheaper alternative for a kubernetes cluster, every time I start try to come up with something I end up with a far more expensive setup and fewer cores. Any hints on viable setups, e.g. replacing 4x POE powered RPI 4's.?

My setup is exactly the same as the top photo on https://www.pidramble.com just with the stack'o'pi zip-tied to the switch so I can move it around without the things sliding around.

For self-hosting I've found https://k3s.io to be really good from the SUSE people. Works on basically any Linux distro and makes self-hosting k8s not miserable.

Yes it’s going to depend on which k8s distribution you’re using. We have work in-progress for k3s to natively support nix-snapshotter: https://github.com/k3s-io/k3s/pull/9319

For other distributions, nix-snapshotter works with official containerd releases so it’s just a matter of toml configuration and a systemd unit for nix-snapshotter.

We run Kubernetes outside of NixOS, but yes the NixOS modules provided by the nix-snapshotter certainly make it simple.

K3S: is a lightweight distribution of Kubernetes that is designed for resource-constrained environments. It is an excellent option for running Kubernetes on a virtual machine or cloud server.

I recently purchased a used Lenovo M900 Think Centre (i7 with 32GB RAM) from eBay to expand my mini-homelab, which was just a single Synology DS218+ plugged into my ISP's router (yuck!). Since I've been spending a big chunk of time at work playing around with Kubernetes, I figured that I'd put my skills to the test and run a k3s node on the new server. While I was familiar with k3s before starting this project, I'd never actually run it before, opting for tools like kind (and minikube before that) to run small test clusters for my local development work.

Kubernetes - k3s distribution

I’ve created a local cluster with K3S and installing Windmill could not be simpler with just one chart to configure, which already has sane defaults to get started. For this demo we will also configure workers to passthrough environment variables to our scripts so that they have access to the Kubernetes API server for later.

You should be familiar with Kubernetes and have set up a Kubernetes cluster. I recommend k3s.

K3s' go.mod[0] is insane.

[0] https://github.com/k3s-io/k3s/blob/master/go.mod