Terraform destroy fails when GCP disk used by multiple instances

[ghost]

This issue was originally opened by @honnibal as hashicorp/terraform#20963. It was migrated here as a result of the provider split. The original body of the issue is below.

---

Re Issue hashicorp/terraform#8667, PR hashicorp/terraform#14651

I have a readonly google_compute_disk that's attached to multiple instances. When I do terraform destroy, I get an Error detaching disk error, noting that an instance was not found. If I run delete a second time, it succeeds.

I'm using v0.11.8.

Further details:

• The disk is defined in the top-level main.tf
• The disk is used by an instance template within a managed instance group, within a submodule.
• In order to pass the disk into the submodule, I pass it by name (since I can't get a reference to the resource in the submodule)

[emilymye]

Providing config and debug logs would be helpful. My best guess is that since you are passing the disk by name instead of interpolating a value from the resource, Terraform cannot know it needs to delete the disks after it deletes the users of the disk. Usually, interpolation would have added an implicit depends_on.

Generally this is done by providing it via an input variable for the module. Is disk not an input variable for the module?

[honnibal]

I think you're right that this was a general misconfiguration on my part, thanks. I'll close this as I doubt it's a bug.

[danawillow]

Closing as per comment.

[ghost]

I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues.

If you feel this issue should be reopened, we encourage creating a new issue linking back to this one for added context. If you feel I made an error 🤖 🙉 , please reach out to my human friends 👉 [email protected]. Thanks!