google_compute_region_backend_service : Every plan show a diff even after multiple apply

[noogg]

Terraform Version

Terraform v0.11.7
+ provider.google v1.10.0

Affected Resource(s)

• google_compute_region_backend_service
• google_compute_region_instance_group_manager

Terraform Configuration Files

resource "google_compute_health_check" "test" {
  name = "test"

  tcp_health_check {
    port = "3128"
  }
}

resource "google_compute_instance_template" "test" {
  name   = "test"
  region = "us-east1"

  machine_type = "n1-standard-1"

  disk {
    device_name  = "persistent-disk-0"
    source_image = "centos-cloud/centos-7"
    auto_delete  = true
    boot         = true
  }

  network_interface = {
    subnetwork = "default"
  }
}

resource "google_compute_region_instance_group_manager" "test" {
  name   = "test"
  region = "us-east1"

  base_instance_name = "test"
  instance_template  = "${google_compute_instance_template.test.self_link}"
  target_size        = 2

  auto_healing_policies {
    health_check      = "${google_compute_health_check.test.self_link}"
    initial_delay_sec = 120
  }
}

resource "google_compute_region_backend_service" "test" {
  name     = "test"
  region   = "us-east1"
  protocol = "TCP"

  backend {
    group = "${google_compute_region_instance_group_manager.test.instance_group}"
  }

  health_checks = ["${google_compute_health_check.test.self_link}"]
}

Debug Output

https://gist.github.com/noogg/317bfb95fe1fcbc604f7e6aee31d532f
https://gist.github.com/noogg/d2aba84f35086b9d5366d34d1106ffbe
https://gist.github.com/noogg/59069679333dbd90f53134a3c78309a4

Expected Behavior

After the first apply run successfully (and neither the infrastructure nor the configuration is modified) I shouldn't see any difference on subsequent plan execution.

Actual Behavior

After the first apply run successfully (and neither the infrastructure nor the configuration is modified) every plan execution show a difference and those even after multiple subsequent apply run successfully.

Steps to Reproduce

⚠️ (This issue start since version v1.8.0 of the google provider and happen only when auto_healing_policies is active on google_compute_region_instance_group_manager ) ⚠️

1. terraform apply
2. terraform plan

[paddycarver]

It looks like we're setting the backend group using a beta self_link in state when beta APIs are used.

We should probably just normalize it to a v1 self_link.

[ffilippopoulos]

+1 We are seeing the same behaviour when defining a google_compute_region_backend_service using a backend from a google_compute_region_instance_group_manager that has a rolling_update_policy.

@paddycarver did you guys managed to work around this?

[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!