Terragrunt destroy dependency check including outside configuration

[jlepere-everlaw]

The purview of the feature added from https://github.com/gruntwork-io/terragrunt/issues/299 feels too broad. Do we really want to be checking every terragrunt.hcl file in the git repo when running destroy commands?

The issue I'm running into is that locals are being evaluated for every terragrunt.hcl file. For my configuration, this includes a run_cmd that calls a script to package terraform code into s3 and return the path to be used as the terraform source. It doesn't really make sense to do this for every terragrunt.hcl my organization has defined when trying to destroy a particular terragrunt module. Also, the run_cmds are failing because an environment variable the script depends on isn't set, for some reason, although it is set before calling terragrunt.

Would it be worth adding a way to skip this functionality by default? Maybe a configuration attribute to turn the functionality off or a cli option to turn it on if it's off by default?

[denis256]

Hi, can be an option to add feature flags to change the default behavior

[velkovb]

Facing a similar issue where a run_cmd command is executed for every project in a repo with a lot of projects. This really slows down the execution of the destroy command. Having a way to turn it off would be useful.

P.S. ended up doing terragrunt plan -destroy -out tf.plan; terragrunt apply tf.plan as a workaround.

[brokenintuition]

This is also an issue for us. In our case, we have separate directories for dev and prod environments and evaluating locals everywhere results in calls to an external service in the other environment which it doesn't have active credentials for.

We're using @velkovb 's workaround for now, but it would be good to be able to disable this behavior.

[BenjaminHerbert]

Another workaround:

terragrunt apply -destroy

Somehow, there seems to be a difference.. It's not only a convenience alias as terragrunt destroy --help claims:

terragrunt destroy --help
Usage: terraform [global options] destroy [options]

Destroy Terraform-managed infrastructure.

This command is a convenience alias for: terraform apply -destroy

[pteroctopus]

Another workaround:

terragrunt apply -destroy

Somehow, there seems to be a difference.. It's not only a convenience alias as terragrunt destroy --help claims:

terragrunt destroy --help Usage: terraform [global options] destroy [options] Destroy Terraform-managed infrastructure. This command is a convenience alias for: terraform apply -destroy

Hi,

We have the same problem after migration from terragrunt version v0.36.1 to v0.53.2.

And can confirm that using terragrunt apply -destroy instead of terragrunt destroy solves it.

Also terragrunt destroy should be made to behave like terragrunt apply -destroy because otherwise it causes problems like the ones already described.