### Terraform Version ```shell v1.11.3 ``` ### Terraform Configuration Files ```terraform ``` ### Debug Output ``` ``` ### Expected Behavior Running unit tests in a DevOps pipeline for a custom vm_windows module. My main configuration file contains a for_each loop with the config for 4 VMs, each with different values so the unit tests can test all cases. One of these values is the availability_set_id, to fill in this value a created availability_set is necessary to get the id from, I do this by creating an availability_set resource just above the for_each loop. The problem with this is that the pipeline ends up failing (even though all tests pass) because terraform test tries to destroy the availability_set before destroying the VMs (because it’s above the VMs in the main.tf file, which isn’t possible so terraform test throws an error and leaves the availability_set in state). Creating the availability_set after the VMs is also not an option since the VM config needs an ID to be valid. I solved this issue by adding a depends_on block to my main.tf file so the configuration recognizes that the AS needs to be created before, and destroyed after the VMs. ### Actual Behavior The problem here is that the **test time to run increases substantially**, from about 10 seconds per test, to 3-5min per test, making my entire pipeline take around 20-25mins! ### Steps to Reproduce 1. `terraform init` 2. create config files using a depends on function 3. create tests 4. `terraform test` ### Additional Context Some configuration code and extra context has been submitted through email to @crw due to my organizations policy on sharing internal code publicly. ### References _No response_ ### Generative AI / LLM assisted development? _No response_