5.2. Meta-Arguments

Preparation

Create a new directory for this exercise:

mkdir -p $LAB_ROOT/advanced/meta_arguments
cd $LAB_ROOT/advanced/meta_arguments

Optional: Create empty files:

touch main.tf

Step 5.2.1: Missing dependency

Sometimes Terraform can not imply the dependency between resources explicitly. For such cases, a dependency is added to one or multiple resources or data sources. Consider the following snippets.

Create a new file named main.tf and add the following content:

resource "local_file" "foobar_txt" {
  content  = "4thelulz"
  filename = "foobar.txt"
}

data "local_file" "reference" {
  filename = "foobar.txt"
}

Now run:

terraform init
terraform apply

This will print the following error:

╷
│ Error: open foobar.txt: no such file or directory
│
│   with data.local_file.reference,
│   on main.tf line 5, in data "local_file" "foobar_txt":
│    5: data "local_file" "reference" {
│
╵

Explanation

The data source local_file.reference is refreshed at the execution of terraform apply. However at this stage, the file does not exist yet and Terraform fails.

Step 5.2.2: Explicit dependency

Change the resource local_file.reference as followed:

data "local_file" "reference" {
  filename = "foobar.txt"

  depends_on = [local_file.foobar_txt]
}

Now run:

terraform init
terraform apply

Terraform will skip trying to refresh (access) local_file.reference because of the explicit dependency on the resource local_file.foobar_txt which does not yet exist.

Step 5.2.3: Ignoring external changes

We set the file content to be 4thelulz. Now lets change it and run apply again:

echo 4real > foobar.txt
terraform apply

Terraform will restore the file foobar.txt to the configuration defined in the code. All good!

But sometimes we don’t want that behaviour - we want to ignore the content.
Luckily Terraform offers another meta-argument for this purpose.

Change the data local_file.foobar_txt as followed:

resource "local_file" "foobar_txt" {
  content  = "4thelulz"
  filename = "foobar.txt"

  lifecycle {
    ignore_changes = [content]
  }
}

Now run:

terraform apply

And Terraform will happily ignore the content = "4thelulz".

Explanation

This is particularly useful in cloud engineering to set initial values for tags or secrets and expect an external system or user to override or extend the value.