This message was deleted.

autoscaling is really not intended for bare metal. It is designed for the cloud where you can get another VM dynamically. That being said you could set this up with Ranchers’ autoscaller. Here is a blog post on it : medium.com/@davide.ruti/autoscaling-made-easy-with-rancher… aka Rancher would orchestrate the autoscalling on a “downstream” cluster that was built on harvester. This would require an external Rancher instance to manage Harvester as a “cloud provider”. here are the docs for it : ranchermanager.docs.rancher.com/v2.10/…/install-cluster-autoscaler

So @thousands-advantage-10804 I have one physical server where i will install Harvestor and then create one RKE2 for installing Rancher and second RKE2 i will provision through Rancher which i want to demonstrate the autoscaling of two node pools one for linux and windows. Is this possible if i have a hypervisor like Harvestor ?

It should be. but what problem are you trying to solve? Is the physical node a big one? Or are you trying to demonstrate?

I want to demonstrate the worker node vms autoscaling. If the utilization is low it should scale down if high it should scale up considering the physical node has capacity

I have never set it up so I am not sure it will autoscale down.

Please Suggest @thousands-advantage-10804 How can we do this using SUSE recommended approach

I would follow the blog post. start with harvester and a single vm. on that vm put rke2 and rancher. Add harvester as a “cloud provider” in rancher. deploy a downstream cluster through rancher. Then setup the autoscaller. Here is a video outlining some automation

Harvestor can be done on the vmware as a vm right and can start to do this on demo environment right ?

yes and no. Harvester should replace vmware. nested virtualization is not ideal.

For Testing, On real environment we will do it bare metal

if you want to test harvester enable nested virtualization on vmware and spin up a LARGE vm for it

Yes @thousands-advantage-10804 We will do on a large VM

something like 48 cores and 128gb of ram. large

Yes sure

I definitely back up Andy's recommendation here. If you want it all self-contained, you can also install an add-on to your harvester cluster to host a rancher management server: docs.harvesterhci.io/v1.8/advanced/addons/rancher-vcluster. you can then "register" the harvester cluster with that rancher management server then use the Harvester cloud provider and follow that blog for autoscaling.

Thankyou @witty-honey-18052 Means a lot