Don't refresh datasources during a destroy.

[clstokes]

If I reference an external resource in a datasource and later on do a terraform destroy and the datasource lookup fails, then my destroy will fail. If I instruct Terraform to not refresh via -refresh=false, my destroy will complete successfully.

Terraform Version

Terraform v0.9.8

Terraform Configuration Files

data "aws_subnet" "instance" {
  id = "${var.instance_subnet_id}"
}

Expected Behavior

I should be able to destroy successfully.

Actual Behavior

If the datasource returns a 404, my destroy fails.

Steps to Reproduce

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

1. Create a subnet externally.
2. terraform apply
3. Delete the subnet externally.
4. terraform destroy. This will fail.

[4dz]

This still occurs in terraform 0.11.7

[miguelaferreira]

Also happens with terraform 0.12

[kamranzafar]

Same thing is happening for ebs snapshot data source, if snapshot is deleted manually before terraform destroy. We are using terraform version 0.12.13

[vincentclee]

Similar issue with RDS aws_db_snapshot datasource. https://www.terraform.io/docs/providers/aws/d/db_snapshot.html

Code

data "aws_db_snapshot" "latest_prod_snapshot" {
  db_instance_identifier = "${aws_db_instance.prod.id}"
  most_recent            = true
}

# Use the latest production snapshot to create a dev instance.
resource "aws_db_instance" "dev" {
  instance_class      = "db.t3.micro"
  name                = "mydbdev"
  snapshot_identifier = "${data.aws_db_snapshot.latest_prod_snapshot.id}"

  lifecycle {
    ignore_changes = ["snapshot_identifier"]
  }
}

Error

Error: Error refreshing state: 1 error occurred:
 * data.aws_db_snapshot.latest_prod_snapshot: 1 error occurred:
 * data.aws_db_snapshot.latest_prod_snapshot: data.aws_db_snapshot.latest_prod_snapshot: DBSnapshotNotFound: DBSnapshot mydbdev-2019-12-31-08-13 not found.
 status code: 404, request id: 9e3af4f3-5b8f-4dd2-87e6-07c75f40c63d

[ayshamdmgit]

What's the solution?

[petelopez]

terraform destroy -refresh=false will stop the refresh and destroy. See here: ComputeCanada/magic_castle#40 (comment)

[jbardin]

The separate refresh walk has been removed, and resources are only refreshed as need during plan, so this should no longer be an issue in the upcoming 0.14 release.

Thanks!

[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 have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.