google_storage_bucket_object Destroys But Does Not Recreate

Problem

A Terraform plan for google_storage_bucket_object shows the resource will be replaced but after apply the resource does not exist at all.

Example Output

Plan shows resource replacement:

Terraform will perform the following actions:

  # google_storage_bucket_object.example must be replaced
  +/- resource "google_storage_bucket_object" "example" {

Apply completes without error:

Apply complete! Resources: 1 added, 0 changed, 1 destroyed.

But subsequent plan shows plan to create, and resource does not exist in Google Cloud Platform:

Terraform will perform the following actions:

  # google_storage_bucket_object.example will be created
  + resource "google_storage_bucket_object" "example" {

Cause

This is caused by a combination of create_before_destroy = true in the Terraform configuration and not changing the name argument on the google_storage_bucket_object resource.

Example Output

The plan shows that the name argument is not changing during the resource replacement operation:

 # google_storage_bucket_object.example must be replaced
+/- resource "google_storage_bucket_object" "example" {
    ...
    name = "example_object"
    ...
  ~ source = "image.png" -> "image2.png" # forces replacement

The apply shows that the resource is first created, then destroyed:

google_storage_bucket_object.example: Creating...
google_storage_bucket_object.example: Creation complete after 0s [id=hashi_example_bucket-example_object]
google_storage_bucket_object.example (deposed object f3bb18b6): Destroying... [id=hashi_example_bucket-example_object]
google_storage_bucket_object.example: Destruction complete after 0s

Explanation

This is due to a create_before_destroy in the configuration, either in the google_storage_bucket_object resource itself:

resource "google_storage_bucket_object" "example" {
  name = "example_object"
  source = "image2.png"
  bucket = google_storage_bucket.example.name
  lifecycle {
    create_before_destroy = true
  }
}

Or a create_before_destroy in a dependent resource, i.e. a resource that depends on google_storage_bucket_object implicitly or explicitly:

resource "null_resource" "example" {
  lifecycle {
    create_before_destroy = true
  }
  depends_on = [google_storage_bucket_object.example]
}

Per the create_before_destroy documentation, dependent resources must enforce this behavior on the resources they depend on to avoid dependency cycles. 

When used in combination with this resource, this causes an issue because the name is the identifier string for this resource in the Google Cloud Platform API, so the destroy request will destroy the recently overwritten resource. Other resources with this same API design can also face this issue.

Solution

Always change the name argument on the google_storage_bucket_object when using create_before_destroy = true, or do not use create_before_destroy = true with this resource or resources that depend on it.