When are changes to server_type applied?

[marvinscharle]

For testing purposes, I recently created a cluster with

module "kube-hetzner" {
  providers = {
    hcloud = hcloud
  }
  
  ... 
  
  autoscaler_nodepools = [
    {
      name = "default"
      server_type = "cx42",
      ...
    }
  ]
}

This worked fine. Then I tried to change the server_type of the default nodepool to "cx32". As seen in the terraform log, the autoscaler_nodepool was replaced:

  # module.kube-hetzner.null_resource.configure_autoscaler[0] must be replaced
-/+ resource "null_resource" "configure_autoscaler" {
      ~ id       = "420145175733131415" -> (known after apply)
      ~ triggers = { # forces replacement

However, the former "cx42" server is still present and no new "cx32" server has been created.

Is this a bug or what is the intended behavior for this?

---

Edit: After a few more tests, it seems like the autoscaler replaces the existing node of the "wrong" type once a new node has been created (and scaled down again).