Attaching disks to OVA deploy

[kvandenhoute-eurocontrol]

Nutanix Cluster Information

Version pc.7.5.0.1
NCC Version: ncc-5.3.0

Terraform Version

registry.terraform.io/nutanix/nutanix v2.4.2

Affected Resource(s)

nutanix_ova_vm_deploy_v2

Terraform Configuration Files

resource "nutanix_ova_vm_deploy_v2" "vm" {
  ext_id   = local.ova.ext_id

  override_vm_config {
...
    dynamic "disks" {
      for_each = { for idx, disk in var.disks : idx => disk if disk.type == "ATTACHED" }
      content {
        disk_address {
          bus_type = "SCSI"
          index    = disks.key
        }
        backing_info {
          vm_disk {
            disk_size_bytes = disks.value.size_gb * 1024 * 1024 * 1024
            storage_container {
              ext_id = <id>
            }
          }
        }
      }
    }
...
}

Debug Output

The plan works:

  + resource "nutanix_ova_vm_deploy_v2" "vm" {
      + override_vm_config {
...
          + disks {
              + ext_id = (known after apply)

              + backing_info {
                  + vm_disk {
                      + disk_ext_id              = (known after apply)
                      + disk_size_bytes          = 1073741824000
                      + is_migration_in_progress = (known after apply)

                      + storage_container {
                          + ext_id = "bdee82a5-8340-40c5-99b3-19cfc180f84d"
                        }
                    }
                }

              + disk_address {
                  + bus_type = "SCSI"
                  + index    = 0
                }
            }
...
        }

but when applying it crashes, since the deployed OVA already has one (or more disks) attached (included in the ova):

│ Error: error waiting for disk creation (ZXJnb24=:3b4fe56f-9a0d-5f4b-9268-c7b9f0a87df7): error_detail: Failed to perform the operation on VM with UUID 'd611ff75-bcf4-45a4-4bcf-3d20579b896d' as the disk bus with type 'scsi' and index '0' is already in use., progress_message: 100
│ 
│   with module.dedicated_vms_0.module.nexus.nutanix_ova_vm_deploy_v2.vm,
│   on ../../modules/ahv-virtual-machine-v2/base_ova_virtual_machine/virtualmachine.tf line 1, in resource "nutanix_ova_vm_deploy_v2" "vm":
│    1: resource "nutanix_ova_vm_deploy_v2" "vm" {

If you change the disk_address to an empty spot:

...
    dynamic "disks" {
      for_each = { for idx, disk in var.disks : idx => disk if disk.type == "ATTACHED" }
      content {
        disk_address {
          bus_type = "SCSI"
          index    = disks.key + var.amount_of_disks_in_ova # in my case + 1
        }
...

you can succesfully apply and everything is succesful:

However, after applying the state is incorrect, because when applying again - without any changes to the terraform -, it wants to remove the disk:

          ~ disks {
                # (1 unchanged attribute hidden)

              ~ backing_info {
                  ~ vm_disk {
                      ~ disk_size_bytes          = 64424509440 -> 1073741824000
                        # (2 unchanged attributes hidden)

                      ~ storage_container {
                          ~ ext_id = "cbfaf36c-2f7b-4330-b695-26b7a6395f2d" -> "bdee82a5-8340-40c5-99b3-19cfc180f84d"
                        }
                    }
                }

              ~ disk_address {
                  ~ index    = 0 -> 1
                    # (1 unchanged attribute hidden)
                }
            }
          - disks {
              - ext_id = "46f7e56d-435f-4542-8ac7-33f0d9c25f7c" -> null

              - backing_info {
                  - vm_disk {
                      - disk_ext_id              = "46f7e56d-435f-4542-8ac7-33f0d9c25f7c" -> null
                      - disk_size_bytes          = 1073741824000 -> null
                      - is_migration_in_progress = false -> null

                      - storage_container {
                          - ext_id = "bdee82a5-8340-40c5-99b3-19cfc180f84d" -> null
                        }
                    }
                }

              - disk_address {
                  - bus_type = "SCSI" -> null
                  - index    = 1 -> null
                }
            }

applying indeed removes the disk:

and then the terraform crashes:

╷
│ Error: error waiting for disk (ZXJnb24=:b0d72552-781b-5fce-8533-7ae000ff49b4) to be updated: error_detail: Failed to perform the operation on the VM with UUID '97c9d3fd-6194-476d-404d-bc867f973ea9', due to an invalid argument with key 'Disk Index' and value '1'., progress_message: 100
│ 
│   with module.dedicated_vms_0.module.nexus.nutanix_ova_vm_deploy_v2.vm,
│   on ../../modules/ahv-virtual-machine-v2/base_ova_virtual_machine/virtualmachine.tf line 1, in resource "nutanix_ova_vm_deploy_v2" "vm":
│    1: resource "nutanix_ova_vm_deploy_v2" "vm" {
│ 
╵

on a third apply no changes are required but the disk is gone.

Expected Behavior

I can specify disk index 0, and it appends to the already existing disks,
or
I use disk index (amount_of_disks_in_ova) + 1

In any case I expect the disk to stay on there after a second apply

Actual Behavior

It removes my extra disk on a second apply

Steps to Reproduce

Try to attach a disk to a nutanix_ova_vm_deploy_v2, after it applied, re-apply

References

I previously created and implemented #985, where this problem did not happen. Is my terraform input incorrect, or is this a later introduced bug in the provider?