This message was deleted.

You may find https://docs.harvesterhci.io/v1.3/install/requirements#network-requirements useful

Thanks, that was alot of ports. I ended up installing harvester on my homelab with rancher. Then installing and managing kubernetes node on vps through rancher as harvester requires more than what vps can handle. My plan is to use my homelab for home apps and playground and the vps for publicly/production applications. Do you think that this is good practice?

I would not suggest deploying Harvester clusters across different networks (regarding the management network.) Two reasons come to my mind immediately: • It is better to have the etcd instances run in a stable network to prevent any issue imposed by network connectivity or performance • The harvester cluster VIP address maintained by kube-vip, which works in the ARP mode, requires that the nodes be in the same L2 network.

I see. Thanks. So it is a better thing to deploy Harvester on my homelab which I can access anytime with Rancher and control external clusters from my homelab Rancher?

If you have, let’s say, one Harvester cluster in your homelab, and other Harvester clusters externally, it’s easier to have a Rancher instance running on the VPS so all of your Harvester clusters (or any other K8s clusters) can access and register with the Rancher instance. As a result, you’ll get a single-pane-of-glass headquarter to manage all the clusters. If you deploy the Rancher instance in your homelab, you’ll need to expose the endpoint so other external clusters can access it.

Actually I did the connection using VPN. So there is no exposing of Homelab 😄 But thanks, I will dig deeper into my needs and see if I can host Harvester on external machine. The external machines are currently a bit weak to handle Harvester.