[iooi]

I run k8s "in my house", which is probably different than what you had in mind, but it might still be useful for others.

I use ESXi and run a separate master node and a separate worker node. There's around ~12 services running in the worker node, mostly things related to media. I maintain a workbook for how I bring up a new node and how I upgrade the cluster, which are the normal operations I've had to do in the past. For example, in the past I had too little storage allocated to the worker node and it was easier to bring up a new one than to edit the existing one.

I use dynamic volumes that use NFS on my NAS for any data that needs to be persisted across pod restarts. This works surprisingly well. I use nfs-client-provisioner with helm.

I also use a combination of MetalLB, an nginx ingress controller, and a BIND service so I can point the DNS on my laptop to the BIND server and I can access all my services using DNS instead of IPs.