Why Do Autoscaler-managed Nodes Show as Created/Destroyed in Terraform State Updates?

[knuurr]

Hi all!

When running terraform plan or apply with autoscaler-managed nodes, Terraform still shows resources as being created or destroyed, even though these nodes are managed outside of Terraform and no actual resources are being modified. It seems like it's simply updating the state to match what's currently running on the cluster, yet the plan output still lists resources as being created or destroyed.

Here is an example from the plan output:

# module.kube-hetzner.null_resource.autoscaled_nodes_registries["node-pool-ccx43-nbg1-7fa3804a0a3cb68f"] will be created
  + resource "null_resource" "autoscaled_nodes_registries" {
      + id       = (known after apply)
      + triggers = {
          + "registries" = " "
        }
    }

# module.kube-hetzner.null_resource.autoscaled_nodes_registries["node-pool-ccx43-nbg1-e39578d50837e05"] will be destroyed
  # (because key ["node-pool-ccx43-nbg1-e39578d50837e05"] is not in for_each map)
  - resource "null_resource" "autoscaled_nodes_registries" {
      - id       = "3024xxxxxx" -> null
      - triggers = {
          - "registries" = " "
        } -> null
    }

Question:

• Why does Terraform show autoscaler-managed nodes as being created and destroyed, even though they aren't actually being provisioned or destroyed?
• Is there a way to suppress this behavior or prevent it from showing in the Terraform plan?
• Should I suppress it, or is this something that needs to be handled explicitly to maintain proper state management?

I’m assuming this is happening because Terraform is attempting to manage the state of these nodes, even though they are managed externally (by the autoscaler), but I’m not sure if this is expected behavior or if there’s a better approach.

Looking forward to hearing thoughts or recommendations on this.

Thanks in advance!