Root volume surviving »terraform destroy«

[wirepatch]

Consider a single server deployment:

...
provider "stackit" {
...
enable_beta_resources = true
}

resource "stackit_security_group" "standard-access" {
  name       = "my_security_group" ...
}

resource "stackit_security_group_rule" "ssh-access" {  ... } # allow ssh
resource "stackit_network" "intern" { ... }
resource "stackit_network_interface" "hello-interface" { ... }
resource "stackit_public_ip" "helloip" {...}

resource "stackit_server" "hello" {
  project_id        = "..."
  name              = "hello-server"
  availability_zone = "eu01-1"
  machine_type      = "t1.1"
  boot_volume       = {
        size              = 10
        performance_class = "storage_premium_perf0"
        source_type       = "image"
        source_id         = "d789a7b4-1efc-4148-be60-de97bc516b6b" # Debian 12 image
    }
}

resource "stackit_server_network_interface_attach" "attached_network_interface" { ... }

terraform apply will create the desired server but on terraform destroy the boot device won't be destroyed:

Name    | Status    | Server     | Availability zones| Size  |  
No name | Available | -          | eu01-1            | 10 GB |  

The web portal provides a Delete root volume when deleting server option not being present here. Besides the general expectation with regards to terraform destroy is cleaning up all resources previously being created.

BTW: It would be very helpful if the provider allowed for provisioning the STACKIT Server Agent like the web portal does. This might be possible by using cloud-init. But in case of a cloud init failure there is no way e.g. setting a password manually for accessing the host in question to debug the problem.

[joaopalet]

Hi @wirepatch,

Thanks for the suggestion!

After discussing with the STACKIT IaaS team, we will provide a field to enable the root volume to be deleted when destroying the server via Terraform.

This will only be supported for root volumes created from an image source type (like the one in your example). Since in the case the source type is volume, the server will use an existing volume and deleting it could cause inconsistencies in the Terraform configuration.

We will update you once this is addressed.

[Fyusel]

Hi @wirepatch,

the field (delete_on_termination) was implemented in order to delete the root volume as well when the server gets destroyed.

The flag is present in v0.40.0