CCE Regression: "Wrong" version causes infinite create/destroy loop

[deas]

Terraform Version

Terraform v0.13.4

Affected Resource(s)

opentelekomcloud_cce_cluster_v3

Terraform Configuration Files

/*
export OS_IDENTITY_API_VERSION=3
export OS_AUTH_URL=https://iam.eu-de.otc.t-systems.com/v3
export OS_REGION_NAME=eu-de
export OS_AUTH_TYPE=token
export OS_PROJECT_NAME=eu-de
export OS_AUTH_TOKEN=your-token-here
export OS_DOMAIN_NAME=OTC-EU-DE-your-domain
*/
terraform {
  required_version = ">= 0.13"
  required_providers {
    opentelekomcloud = {
      source  = "opentelekomcloud/opentelekomcloud"
      version = "1.20.3"
    }
  }
}

resource "opentelekomcloud_cce_cluster_v3" "cluster" {
  name = "repro-loop"
  flavor_id = "cce.s1.small"
  cluster_version = "v1.15.11" # Loops
  # cluster_version = "v1.15.11-r1" # Works
  cluster_type = "VirtualMachine"
  description = "Dunno cluster by terraform"
  billing_mode = 0
  vpc_id = data.opentelekomcloud_vpc_subnet_v1.Subnet.vpc_id
  subnet_id = data.opentelekomcloud_vpc_subnet_v1.Subnet.id
  container_network_type = "overlay_l2"
  authentication_mode = "rbac"
  multi_az = false
  # 🐌⏳😴 - Likely a recent regression, destruction used be faster for me
  timeouts {
    create = "30m"
    delete = "30m"
  }
}

data "opentelekomcloud_vpc_subnet_v1" "Subnet" {
  name = "your-subnet-here"
}

Steps to Reproduce

Please list the steps required to reproduce the issue, for example:

1. terraform apply
2. terraform apply
3. terraform apply

Expected Behavior

Client not even accepting a version causing this behavior. You never ever want it.

Actual Behavior

Ending up in an infinite loop.

Apply complete! Resources: 1 added, 0 changed, 1 destroyed.

[outcatcher]

Fixed in the latest release (v1.21.2)