Goggle compute API beta->v1 change keeps triggering a TF diff with 1.13.0

[czka]

$ terraform version
Terraform v0.11.7
+ provider.google v1.13.0

terraform plan with provider.google v1.10.0 reports no changes on terraform plan. After upgrading to v1.13.0, terraform plan keeps reporting the change quoted below, even though I did run terraform apply on it in advance.

An execution plan has been generated and is shown below.
Resource actions are indicated with the following symbols:
  ~ update in-place

Terraform will perform the following actions:

  ~ google_compute_instance.machine1
      network_interface.#:                    "1" => "1"
      network_interface.0.address:            "10.118.234.33" => <computed>
      network_interface.0.name:               "nic0" => <computed>
      network_interface.0.network_ip:         "10.118.234.33" => <computed>
      network_interface.0.subnetwork:         
"https://www.googleapis.com/compute/beta/projects/project1/regions/us-east4/subnetworks/subnet1" => 
"https://www.googleapis.com/compute/v1/projects/project1/regions/us-east4/subnetworks/subnet1"
      network_interface.0.subnetwork_project: "project1" => <computed>


Plan: 0 to add, 1 to change, 0 to destroy.

[rosbo]

Hi, I cannot reproduce this issue.

I tried pinning the version to 1.10.0 and then upgrading to 1.13.0 and still see no diff.

In the output you posted, did you copy integrally the stuff below ~ google_compute_instance.machine1? Because there is no diff in the network_interface. When looking at it, one could think the diff comes from the subnetwork field but it does not. There is a function that suppress the diff between v1 and beta self link on the subnetwork field and since the subnetwork field is ForceNew, a diff would create a Create/Delete diff and not an in-place update. The output would look like that:

network_interface.0.subnetwork:         "https://www.googleapis.com/compute/beta/projects/xxx/regions/us-east1/subnetworks/default" => "https://www.googleapis.com/compute/v1/projects/xxx/regions/us-west1/subnetworks/default" (forces new resource)

Do you think you could share your debug logs please?

[rosbo]

This issue is also related to #988 where we have a perm diff when there should be no diff. We are investigating it.

[danawillow]

@czka, mind sharing your config? Are you using lifecycle.ignore_changes by any chance?

[czka]

@danawillow This particular node has ignore_changes = ["metadata_startup_script", "boot_disk.0.initialize_params.0.image", "boot_disk.0.device_name" only. Nothing else is being ignored all over the TF resources which are processed during the terraform plan. I can't share my config. It's top secret production stuff :).

@rosbo After switching from 1.10.0 to 1.13.0 I'm only getting a diff for google_compute_instance (a number of them; this one I'm quoting here is just one example). TF doesn't return diff for any other resources.

[rosbo]

@danawillow found a bug in Terraform Core which is causing ForceNew field with a DiffSuppressFunc to not behave properly when there is a 'suppressable' diff. See for more details: hashicorp/terraform#18209

While waiting for the Terraform Core to be fixed, we can cheat a bit by converting the subnetwork self_link to v1 before we store it in the state. I will open a PR soon

[rosbo]

I will keep this issue open until hashicorp/terraform#18209 is merged

[rosbo]

Actually, this specific issue with beta links is resolved with the latest fix merged. We will keep #988 open to track the real root cause of this issue. Closing

[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!