Rancher creates local cluster automatically with public IP

Hi,

First time setting up Rancher here so any help would be greatly appreciated. My setup is Proxmox > Ubuntu Server > Docker > Rancher > and would like to run Kubernetes and Plex.

When I enter Rancher for the first time by going to the IP address of my ubuntu server I noticed that there is already a cluster created called ‘local’. Then when I look into ‘Nodes’ there is already one created called ‘local-node’ with my public IP.

  1. Why does this happen?
  2. Why can’t I delete these cluster and nodes?
  3. shouldn’t the IP address be my private IP for my ubuntu server?

TIA for any help and I have attached some screenshots below. Cheers!

Rancher is a collection of controllers and custom resources and other glue that run in, and only in, a Kubernetes cluster.

If do the HA install with helm, you are directly installing those into the target cluster. That cluster then shows up inside Rancher as local .

If you do the docker run ... rancher/rancher , there’s no Kubernetes cluster lying around to run in, so the container…contains it’s own installation of k3s running inside it, to provide the local cluster Rancher runs on.

You can’t delete it because that’s where all of Rancher is. As the admin you need access to the local cluster to manage or debug various things. You do not need to (and generally should not) give access to it to other users.

Kubernetes picks up the node IP and it is what it is because that’s what came back. But for the docker container only the ports you bind (which is normally just -p 80:80 -p 443:443 are actually reachable from the outside.

p.s. Posting the same thing in multiple places simultaneously just creates more work for the same people that watch pretty much all of them.

Can we used this cluster (local) for another purpose? Ex. Deploy some image into.

It is not recommended if you have more than one cluster to manage. Giving away too much access to the local cluster can give someone the equivalent of admin on all the other clusters registered with that Rancher (since their configuration is stored in local), and your workloads could compete with Rancher for resources. But if you only wanted one cluster in the first place, you can.