automatic_restart false in scheduling/google_compute_instance_template forces new resource

[dgolja]

Hi there,

If in google_compute_instance_template you define scheduling block automatic_restart = false, terraform will always recreate the resource trying to update automatic_restart field to true.

Terraform Version

Terraform v0.9.11

Affected Resource(s)

• google_compute_instance_template

Terraform Configuration Files

...
resource "google_compute_instance_template" "instance_template" {
  name_prefix = "dgolja-instance-template-tf-bug-"

  region = "${var.region}"

  lifecycle {
    create_before_destroy = true
  }

  machine_type         = "n1-standard-1"
  can_ip_forward       = false

  scheduling {
    automatic_restart   = false
    on_host_maintenance = "MIGRATE"
  }

  network_interface {
    network = "default"
  }

  disk {
    source_image = "projects/ubuntu-os-cloud/global/images/ubuntu-1604-xenial-v20170619a"
    disk_type    = "pd-ssd"
    auto_delete  = true
    boot         = true
  }

  service_account {
    scopes = ["userinfo-email"]
  }
}
...

Debug Output

https://gist.github.com/dgolja/1c80d16db90d9bbea5fa1cce21d9f8f9

Expected Behavior

Since there is no change terraform shouldn't recreate the resource.

Actual Behavior

Keep updating the resource, however the new resource is the same.

Steps to Reproduce

1. terraform apply

[dgolja]

Workaround for tf v9.11 is to use a deprecated parameter automatic_restart and set it to false.

Also after digging into the code I noticed that deprecated automatic_restart value will always get overwritten if scheduling.automatic_restart is defined, so you can end up in really bizarre state. L514-L516 L532-L535

My recommendation would be to remove the automatic_restart parameter from the state file. This should fix the problem. The value of automatic_restart get already stored in the state file anyway under scheduling.automatic_restart

[selmanj]

Thanks for reporting, hopefully #248 fixes this for you.

[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!