How many way to install Rancher?

As I’m beginner on Rancher. How many way to install Rancher?

As I research, Rancher has ways to install by

  • Way to pull docker image rancher/rancher – This is from Quick started guide that I have done on my lab. But it seems not stable since after I restarted server, sometime Rancher’s container is not work properly.
  • Install on Linux OS?
  • Install on Kubernetes? (I am wonder that this method install K3S on K8S, what is reason?)
  • Anything else?

I’m not sure that can everyone suggest pro&con for each installation methods.
Do I should focus to which installation methods that suitable for used Rancher on Production environment ?

I know of the docker image way and installing on Kubernetes using a Helm chart (though I used RKE2 instead of K3S).

The install direct on Linux was apparently discontinued with 2.6.

1 Like

Thank you.

This meant, in case that we choose install Rancher on Kubernetes. This is purpose for used Rancher to manage Kubernetes, right?

Yep. For a dev cluster I have as a sample that I want to follow high availability I end up with a lot of VM bloat. I’ve got the following:

  • rancher1vm-rancher3vm - three VMs that I installed RKE2 on and then used Helm to install Rancher
  • rancher-ui - DNS hostname with three A records pointing to rancher1vm-rancher3v and this is what I use in the browser and specify to Helm as hostname during install. This is also used for the server in the RKE2 install on nodes 2 & 3.
  • ctrlplane1vm-ctrlplane3vm - three VMs that I set to do etcd & control plane checkbox when deploying RKE2 from rancher
  • ctrlplane-ui - DNS hostname with three A records pointing to ctrlplane1vm-ctrlplane3vm and if I were doing kubectl commands direct I’d change the server line to point to this. I’d also use this in the RKE2 install if I were doing it manually, but for the most part this is unused and just for completeness.
  • worker1-5 - five VMs that I set to do worker checkbox when deploying RKE2 from rancher.

Then I’ve got some other stuff for DNS and some external load balancers and other things that’re more environment-specific. From Rancher I see two clusters, local and the one I named on the last 8 VMs.

Hope that helps. I found it odd/confusing at first too.