Should the master node's hardware be the same as the worker node's hardware for a Kubernetes setup?

Question

I'm new to Kubernetes and am setting up a raspberry pi cluster at home and I want to put Kubernetes on it. I have 4 new raspberry pi 4s with 8gb each and one old raspberry pi 2+. There are several hardware differences between the two models but I think technically I can make the old pi2 the master node to manage the pi4s. I'm wondering if this is a good idea. I don't really understand what the consequences of this would be. For instance if I was to expand my little cluster to 8 or 16 pi4s in the future, would my one pi2 be overloaded in managing the workers? I'm really trying to understand the consequences of putting lower grade hardware in control of higher grade hardware in the master/worker node relationship in Kubernetes.

There are three main goals that I have for this hardware. I want to recreate an entire development environment, so some VMs that would host a testing environment, a staging environment, a dev environment, and then a small production environment for hosting some toy website, and then I want to be able to host some services for myself in Kubernetes like a nas storage service, a local github repo, an externally facing plex media server, stuff like that. I think I can use k8s to containerize all of this, but how will having a pi2 as master limit me? Thanks in advance.

Answer 1

Normally, kubernetes masters use more resources because they run a lot of things and checks by default. Mainly because etcd and apiserver . Etcd is the database that stores everything that happens in Kubernetes, and apiserver receives all api requests from inside and outside the cluster, check permissions, certificates and so on.

But this is not always truth, sometimes you can have node with a lot of heavy applications, consuming much more resources than masters .

There are always the recommend specs, the specific specs for our business logic and enterprise applications, and the "what we have" specs.

Because you can move pods between different machines, you can remove some weight from your master, no problem.