Hi all, I'm deploying Harvester into my homelab because I want to start learning K8s more, and I have a mix of container and VM workloads. I'm starting off with a single node, but will add nodes as time goes on to eventually have at least 3. I intend on getting Rancher running to manage the cluster, and using the experimental baremetal flag so I can run containers on the baremetal alongside the VMs. My question is, where should I install Rancher? I don't have other nodes I can run it on, so it's going to run on the Harvester cluster one way another. I've seen two main methods in the docs: 1. Spin up a VM and install Rancher on k3s in the VM 2. Use the exprimental rancher-vcluster/rancher manager plugin. But both of those end up with having nested k8s clusters atop the already existing cluster in harvester, which seems like it just adds complexitiy and overhead to me. Couldn't I either: 1. Install Rancher onto the native k8s cluster running on Harvester (in a different namespace probably?) 2. Use the embedded rancher that's already on harvester in the cattle namespace. I've seen a few forum posts about doing it this way, but they didn't get a whole lot of response. What's the way I should be getting rancher running on this cluster so it can manage this cluster (and eventually other remote clusters)?

Harvester is a bare metal OS. Install it on the node itself. There are 2 options for Rancher. A. a vm with rke2, or k2s. B. Rancher running in a vcluster directly on Harvester : https://docs.harvesterhci.io/v1.5/advanced/addons/rancher-vcluster I made a video on all of this : I would play and test both ways. It really comes down to how you want to manage things.

Looks like I had a typo in my original post because my actual question where I should be installing Rancher onto my harvester cluster. Oops. So there's no real support/it's not recommended to deploy rancher as a workload directly onto the harvester underlying RKE2 cluster, using the manual helm install in the rancher docs? It should always be in a whole vcluster/VM, and thus nested kubernetes? And the other clusters I eventually run are a few offsite single node servers that currently have NixOS on them. If I like what I achieve with this cluster I'd fold them under rancher and spin up harvester on them; They're 99% just to have backups of data and don't do a whole lot regardless, and it's not an issue for me if Rancher goes down on my home cluster. Thanks for the vids for sure. Will definitely take a watch and think I've come across them before in my searches.