Workbench Instance - Metadata forces recreate

kirksw commented 3 months ago

Community Note

Please vote on this issue by adding a 👍 reaction to the original issue to help the community and maintainers prioritize this request.
Please do not leave +1 or me too comments, they generate extra noise for issue followers and do not help prioritize the request.
If you are interested in working on this issue or have submitted a pull request, please leave a comment.
If an issue is assigned to a user, that user is claiming responsibility for the issue.
Customers working with a Google Technical Account Manager or Customer Engineer can ask them to reach out internally to expedite investigation and resolution of this issue.

Terraform Version & Provider Version(s)

Terraform v1.8.5 on aarch64-apple-darwin

provider registry.terraform.io/hashicorp/google v5.33.0

Affected Resource(s)

google_workbench_instance

Terraform Configuration

resource "google_workbench_instance" "default" {
  name     = "workbench-test"
  location = "europe-west1-b"

  gce_setup {
    machine_type = "n1-standard-4"

    service_accounts {
      email = google_service_account.default.email
    }

    metadata = {
      idle-timeout-seconds = "3600"
    }
}

Debug Output

if resource stopped

~ resource "google_workbench_instance" "default" {
  ~ desired_state        = "STOPPED" -> "ACTIVE"
    id                   = "projects/<redacted>/locations/europe-west1-b/instances/<redacted>"
    name                 = "workbench-test"
    # (12 unchanged attributes hidden)

  ~ gce_setup {
      ~ metadata             = {
          - "proxy-byoid-url"            = "<redacted>-dot-europe-west1.notebooks.byoid.googleusercontent.com" -> null

if running already

# module.poc_product_categorization[0].google_workbench_instance.default will be updated in-place
~ resource "google_workbench_instance" "default" {
    id                   = "projects/<redacted>/locations/europe-west1-b/instances/<redacted>"
    name                 = "<redacted>"
    # (13 unchanged attributes hidden)

  ~ gce_setup {
      ~ metadata             = {
          - "proxy-byoid-url"            = "<redacted>-dot-europe-west1.notebooks.byoid.googleusercontent.com" -> null
            # (16 unchanged elements hidden)

Expected Behavior

With no desired state defined if instance is active, instance should stay active, likewise if stopped should stay stopped
No update in-place should occur in relation to metadata.proxy-byoid-url

Actual Behavior

Starts all instances on every apply, updates metadata.proxy-byoid-url to null.

Steps to reproduce

terraform apply

Important Factoids

No response

References

No response

b/348498822

ggtisc commented 3 months ago

Confirmed issue!

After the resource was created when there aren't changes and ran a 2nd terraform apply it detects changes. This looks like a permadiff on proxy-byoid-url

kirksw commented 2 months ago

Confirmed issue!

After the resource was created when there aren't changes and ran a 2nd terraform apply it detects changes. This looks like a permadiff on proxy-byoid-url

I would say there are two issues, what you have referenced - but also the default desired state behavior, which will start or stop all the instances (no possibility to just stay in the current state)

bcreddy-gcp commented 2 months ago

The issue should be fixed with https://github.com/GoogleCloudPlatform/magic-modules/pull/10900

Verified by the following: Created a google_workbench_instance using the following config.

resource "google_workbench_instance" "default" {
  project = "<redacted>"
  name = "workbench-test"
  location = "us-west1-a"

  gce_setup {
    machine_type = "n1-standard-4"
    metadata = {
      idle-timeout-seconds = "3600"
    }
  }

}

terraform plan on the same config any number of times gave No changes. Your infrastructure matches the configuration..

Added desired_state = "ACTIVE" to the config. terraform plan on the same config any number of times gave No changes. Your infrastructure matches the configuration..

Updated desired_state = "STOPPED" in the config.

  ~ update in-place

Terraform will perform the following actions:

  # google_workbench_instance.default will be updated in-place
  ~ resource "google_workbench_instance" "default" {
      ~ desired_state        = "ACTIVE" -> "STOPPED"

terraform plan on the same config any number of times gave No changes. Your infrastructure matches the configuration..

Updated desired_state = "ACTIVE" in the config.

  ~ update in-place

Terraform will perform the following actions:

  # google_workbench_instance.default will be updated in-place
  ~ resource "google_workbench_instance" "default" {
      ~ desired_state        = "STOPPED" -> "ACTIVE"

terraform plan on the same config any number of times gave No changes. Your infrastructure matches the configuration..

hashicorp / terraform-provider-google