尝试协调/刷新 terraform 基础设施漂移

Question

I have a Terraform GCE resource being created using the following module:

resource "google_compute_address" "gke_proxy1" {
  ...
}

resource "google_compute_disk" "gke_proxy1" {
  ...
}

resource "google_compute_instance" "gke_proxy1" {
  ...
}

However at some point, outside Terraform scripts, created the google_compute_address for a specific GCE already so when I try and apply the TF script, it tries to create the compute address which is not what I want. For eg using:

terraform plan -var "env=dev" -target module.gke-proxy-primary;

If I run the following to show real-world state:

 terraform state show module.gke-proxy-primary.google_compute_instance.gke_proxy1

Output shows the expected compute_address state address as expected:

network_interface {                                                                                                                                                             
     name               = "nic0"                                                                                                                                                 
     ...                                                                                                             
 }                                                                                                                                                                               
 network_interface {                                                                                                                                                             
     name               = "nic1"                                                                                                                                                 
     ...                                                                                                             
 }            

                                                                                                                                                               

So I guess I need rebuild the TF state from the real-world infra so I have tried the following since there's been a drift between state and real world: Running:

terraform refresh -var "env=dev" -target module.gke-proxy-primary;

Gives me:

module.gke-proxy-primary.google_compute_disk.gke_proxy1: Refreshing state... [id=projects/proj1/zones/europe-west2-a/disks/gke-proxy1]
module.gke-proxy-primary.google_compute_instance.gke_proxy1: Refreshing state... [id=projects/proj1/zones/europe-west2-a/instances/gke-proxy1]

So it's not refreshing the compute_address.gke_proxy1 remote bucket state for the resource as expected from real world. Any ideas what I'm doing wrong?

Answer 1

1. Import the missing dependency compute_address to state for the main resource gce using:

   terraform import -var "env=dev" -var "primary_cluster_suffix=1" module.gke-proxy-primary.google_compute_address.gke_proxy1 hsbc-10723777-safe-dev/europe-west2/gke-proxy1

2. Verify the resource is then in the local state:

   terraform state show module.gke-proxy-primary.google_compute_address.gke_proxy1 It should now show.

3. Run plan and it should reconcile as expected.