How to safely scale down number of nodes


I’ve recently removed a few applications from one of my clusters and now I want to scale it down because it doesn’t need that many nodes any more.

I tried using kubectl to see what pods are running on each node and delete going to rancher and delete that node. But that didn’t work because rancher would auto replace the node. I tried disabling auto-replace and that didn’t work either.

My question is if I click on the ‘-’ button on my node pool, what will happen? Does rancher randomly pick a node and delete it or it follow some sort of scheduling rules? I don’t want to use drain or cordon either because all my pods are single replica.