Decommissioning - Additional disks are not removed if other method than TF is chosen

Terraform Providers Google
1

Hi,

I detailed my issue in the following GitHub issue : All additional disks are in allowed to have · Issue #32524 · hashicorp/terraform · GitHub

Terraform Version

[cs-cfme-ro@vmiadev01 dev1gcesim86rhel6431]$ terraform --version Terraform v1.3.7 on linux_amd64 + provider Terraform Registry v4.48.0 + provider Terraform Registry v4.48.0

Terraform Configuration Files

variables.tf.txt
variables.tfvars.txt
versions.tf.txt
disks.tf.txt
instances.tf.txt
outputs.tf.txt
startup-script-custom.txt

cannot put autodelete in TF for additional disks

Debug Output

All works fine, but I cannot have the additional disks as the boot_disk : auto_delete = var.auto_delete

Nothing in document to allow me to have additional disks with auto_delete = true

Expected Behavior

I want to have in resource “google_compute_disk” named “gce_disk” an additional option to allow auto_delete = true

Actual Behavior

The management wants us to move out from ansible to Terraform … Ansible playbook works fine for GCE but changes require from us to perform CLI or REST API calls

When I am doing a delete of my GCE instance by TF all works but by passing by GCP REST API calls it does not work fully : additional disks are not removed.

TF is not compliant with GCP usage here, I should have the possibility to put auto_deletation of the disks at the termination of the GCE instance. Even with Ansible module I can do it with the module gce_pd : community.google.gce_pd module – utilize GCE persistent disk resources — Ansible Documentation

Questions :

  • Why it is not possible with TF ?
  • How do the auto_deletation = true for the additional disks in resource “google_compute_disk” ?

Thanks in advance

Steps to Reproduce

alias apply=‘echo “terraform apply -var-file=“variables.tfvars” -auto-approve”; terraform apply -var-file=“variables.tfvars” -auto-approve’
alias destroy=‘echo “terraform destroy -var-file=“variables.tfvars” -auto-approve”; terraform destroy -var-file=“variables.tfvars” -auto-approve’
alias plan=‘echo “terraform plan -var-file=“variables.tfvars””; terraform plan -var-file=“variables.tfvars”’
alias show='echo "terraform show "; terraform show ’
alias tf=‘terraform’
alias gcurl=‘curl -H “Authorization: Bearer $(gcloud auth print-access-token)” -H “Content-Type: application/json” -f --request’

1. init
2. apply
3. call GCP REST API to delete the GCE instance

[cs-cfme-ro@vmiadev01 dev1gcesim86rhel6431]$ gcloud compute instances describe dev1gcesim86rhel6431
No zone specified. Using zone [europe-west3-c] for instance: [dev1gcesim86rhel6431].
canIpForward: false
cpuPlatform: Intel Broadwell
creationTimestamp: ‘2023-01-16T23:41:38.763-08:00’
deletionProtection: false
description: GCE Instance built from Image dbg-clearstream-sz-61064e0c/cs-sz-rhel86-image
disks:
- autoDelete: true
boot: true
deviceName: dev1gcesim86rhel6431-bootdisk
diskSizeGb: ‘65’
index: 0
interface: SCSI
kind: compute#attachedDisk
licenses:

kind: compute#metadata
name: dev1gcesim86rhel6431
networkInterfaces:

[cs-cfme-ro@vmiadev01 dev1gcesim86rhel6431]$ gcurl DELETE https://www.googleapis.com/compute/v1/projects/dbg-clearstream-simu-fe9d0e70/zones/europe-west3-c/instances/dev1gcesim86rhel6431
{
“kind”: “compute#operation”,
“id”: “1649327497074697796”,
“name”: “operation-1673943210806-5f27146744e8c-243b2c23-aeb2fc90”,
“zone”: “https://www.googleapis.com/compute/v1/projects/dbg-clearstream-simu-fe9d0e70/zones/europe-west3-c”,
“operationType”: “delete”,
“targetLink”: “https://www.googleapis.com/compute/v1/projects/dbg-clearstream-simu-fe9d0e70/zones/europe-west3-c/instances/dev1gcesim86rhel6431”,
“targetId”: “3824673040305483742”,
“status”: “RUNNING”,
“user”: “svc-cs-deployment@dbg-clearstream-simu-fe9d0e70.iam.gserviceaccount.com”,
“progress”: 0,
“insertTime”: “2023-01-17T00:13:31.328-08:00”,
“startTime”: “2023-01-17T00:13:31.351-08:00”,
“selfLink”: “https://www.googleapis.com/compute/v1/projects/dbg-clearstream-simu-fe9d0e70/zones/europe-west3-c/operations/operation-1673943210806-5f27146744e8c-243b2c23-aeb2fc90”
}

gcloud compute disks list --filter=“(users = null)”
NAME LOCATION LOCATION_SCOPE SIZE_GB TYPE STATUS
dev1gcesim86rhel6431-oradbms-data europe-west3-c zone 10 pd-ssd READY
dev1gcesim86rhel6431-sysvgsystem europe-west3-c zone 30 pd-standard READY
dev1gcesim86rhel6431-vgdocker europe-west3-c zone 10 pd-ssd READY

All additional disks remain and became orphan disks, because they are in auto_delete = false. Need a TF option to switch them as the boot_disk auto_delete = true

Additional Context

  1. Management wants to move out from Ansible to TF. Who sells it if I cannot do what works previously with Ansible modules ?
  2. TF should be compliant by doing alignement with GCP. Meaning TF behaviour = GCP behaviour and GCP as lead, because here TF follows one way and GCP follows in another way.

References

I saw someone having the same issue : google cloud platform - How to set auto-delete option for additional attached_disk in gcp instance uing terraform? - Stack Overflow