This message was deleted.

The number of VMs on a single node would be constrained by the number of pods the underlying rke2 kubelet is configured to use, which I think is ~200. But there are also system services running as pods on each individual node, so probably a little less than 200.

@bored-painting-68221 thanks for your answer, because in last session i test, i only can create arround 23 VMs & 53 storage volume, and i check in harvester node, i found 102 pods running. so i think harvester cannot compete with nutanix/vmware, right ?

If you attach pvc to a VM, there will be an accompany hp-volume-xxx pod. When you attach another pvc to the same VM, a new hp-volume pod replaces the old hp-volume pod. So there might be maximum 3 pods for a VM (although the number is back to 2 soon) in runtime. The maximum allowable number of VMs on a node depends on system services pods, VM pods, and volume usage.

@prehistoric-balloon-31801 so, what you have simple way to me ? maybe sizer workload for harvester ? because we need know estimation detail for the workload consume. like this maybe:

I’m afraid I can’t have a way for this. I merely answer the question from the maximum pod limitation perspective. When considering CPU/memory, it’s also related to overcommit ratios. Is this a public estimation tool?

yes, its public estimation tool, you can access in here: https://wintelguy.com/vmcalc.pl because i still look, what the harvester can me deploy as second solution from nutanix/vmware, but for now, after i check harvester maybe only can use to best solution cloud-native with primary kubernetes workload. so harvester best-fit for support rancher as primary kubernetes solution (and harvester come for hybrid workload VMs) and not good if i running the traditional workload such as VMs concern in here.