This message was deleted.

3000 clusters on a raspberry pi station? that's hella lot of RPi's you'd need to support the basic memory management i suppose. Not something I can test for ya with two RPi's.

There is a table here that you can review for requirements (for rancher)... https://rancher.com/docs/rancher/v2.6/en/installation/requirements/#k3s-kubernetes

No, each raspberry pi has one cluster in it. I got 3000 raspberry pis.

I have no idea, but k3s is lightweight and good for the arm architecture. But I have no idea why you would want 3.000 clusters, rather something like 30 clusters with each 100 nodes.

@kind-nightfall-56861 All the 3000 raspberry pis are located across country in 3000 different fields. I’m not sure about the network latency. I can work with hosting master on AWS and having all the nodes on raspberry pis. But I’m worried about network latency. The requirement is to deploy services on raspberry pis using gitops approach and collect/monitor heath stats.. any suggestions on how to handle this situation would be helpful 🙂

I started with k3s about a month ago, so I might be the wrong person to ask, haha. I'm no expert. I think multiple nodes on one cluster is primarily for load balancing, which doesn't really sound like something you'd need, guessing that with "fields" you actually mean fields, so you'd work in something like agriculture or something, where any form if latency is a big no-no, and you'd have to make sure that EACH pod runs a certain pod, which is something different than load balancing. I'm not sure if a single cluster offers the functionality you're searching for, haven't encountered that problem / question before.

Yep that’s true. I’ve gone through the same set of questions and decided to provision 3000 clusters lol. But I’ll wait for others to respond

Unless they're Pi 4s with more than 1GB RAM you may be limited with what you can do with them once K3S installed (and the version of K3S you can use)

Hi @witty-jelly-95845 Yes, they’re Pi 4s with more than 1GB RAM