Missing allow_stopping_for_update into google_compute_instance_template

[Maxime-SIQUIER]

Terraform Version

Terraform v0.10.7
+ provider-google-version 1.16.0

Affected Resource(s)

• google_compute_instance_template

Terraform Configuration Files

resource "google_compute_instance_template" "openshift_node" {
  name_prefix               = "${var.stack_name}-ose3-${var.cluster_name}-${var.node_name_suffix}"
  region                    = "${var.gce_region}"
  machine_type              = "${var.node_flavor_name}"
  min_cpu_platform          = "${var.node_min_cpu_platform}"

  scheduling {
    automatic_restart = "${lookup(var.preemptible_auto_restart, var.node_is_preemptible)}"
    preemptible       = "${lookup(var.true_false, var.node_is_preemptible)}"
  }

  disk {
    disk_type    = "${var.node_disk_type}"
    source_image = "${var.node_image_id}"
    boot         = true
    disk_size_gb = "${var.node_disk_size}"
  }

  network_interface {
    # Providing both ID's (network together with subnetwork) is not supported.
    # Giving subnetwork a higher precedence.
    network = "${var.subnetwork_id == "" ? var.network_id : ""}"

    subnetwork         = "${var.subnetwork_id}"
    subnetwork_project = "${var.subnetwork_project}"
  }

  metadata {
    "stackname"      = "${var.stack_name}"
    "host-type"      = "node"
    "env-host-type"  = "${var.stack_name}-openshift-${var.node_subtype}"
    "sub-host-type"  = "compute"
    "is-preemptible" = "${lookup(var.yes_no, var.node_is_preemptible)}"
    "clusterid"      = "${var.stack_name}-${var.cluster_name}"
    "consul_ready"   = "${var.consul_ready}"
  }

  # This section might be updated soon on openshift-ansible
  # Tags are used by openshift-ansible on GCE (not byo mode)
  tags = [
    "${var.cluster_tag}",
    "${local.nodetype_id}",
    "host-type-node",
    "sub-host-type-compute",
    "${var.stack_name}-${var.cluster_name}-${local.role}-${var.node_subtype_name}",
    "${var.stack_name}",
  ]

  metadata {
    startup-script = "${data.template_file.openshift_node_user_data.rendered}"
  }

  service_account {
    email  = "${var.service_account_email}"
    scopes = ["useraccounts-ro", "compute-rw", "storage-rw", "logging-write", "monitoring-write", "service-management", "service-control"]
  }

  # 1 if number of nodes > 0, otherwise 0
  count = "${var.node_count > 0 ? 1 : 0}"

  lifecycle {
    create_before_destroy = true
  }

  allow_stopping_for_update = true
}

Debug Output

N/A

Panic Output

N/A

Expected Behavior

I should be able to set allow_stopping_for_update into google_compute_instance_template

Actual Behavior

I am not able to set allow_stopping_for_update into google_compute_instance_template and it generate the follwing error when I perform a terraform validate :

Errors:

  * module.ext-a.module.minions-compute.google_compute_instance_template.openshift_node: : invalid or unknown key: allow_stopping_for_update

Steps to Reproduce

Please list the steps required to reproduce the issue, for example:

1. terraform validate

Important Factoids

N/A

References

I think that the bug has been introduced with the following PR : #1005

[paddycarver]

Hi @Maxime-SIQUIER, just to check, have you ever been able to use this field with google_compute_instance_template? #1005 only affects google_compute_instance, as far as I can tell.

And because this isn't an API feature, but a Terraform behaviour toggle, I'm not sure what you're asking for here. Instance Templates can't be stopped. Can you talk about a workflow you'd like enabled that currently isn't possible?

[Maxime-SIQUIER]

Hi @paddycarver,

First of all thanks for the quick answer.

No it has never worked for sure but I expected it to work when I started to think about to implement it following the management of the min CPU platform.

My problem is that I manage all my instance thanks to the template feature, and since I have added the min_cpu_platform, I would like to ensure that user understand that an update of this variable will lead to a reboot of the VM by forcing them to set it at true. As from my understanding but maybe I am wrong, with #1005 we are removing the ForceNew on min_cpu_platform to just a stop/update/reboot if we set the allow_stopping_for_update to true but with the google_compute_instance_template if I update my min_cpu_platform it will lead to a forceNew and I will not be able to benefit of the what I will call soft-update. Am i right ? And does my explanation is clear enough ? I am a little bit quite new on this... ^^"

[paddycarver]

Hi @Maxime-SIQUIER, how are you turning instance templates into instances? Managed instance groups? compute_instance_from_template? #1005 seems unrelated, and is just where this feature was introduced for google_compute_instance, but that doesn't interact with templates at all. Instance templates can't be updated--it's an API limitation--so any change to an instance template must always be ForceNew. There's no such thing as an update to an instance template, soft or not. There's only replacing it.

[Maxime-SIQUIER]

Understood and agreed.

[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!