### Terraform Version ```shell Terraform v1.14.0 on darwin_arm64 + provider registry.terraform.io/hashicorp/local v2.6.1 ``` ### Terraform Configuration Files ```terraform resource "local_file" "foo" { content = "foo!bar" filename = "/tmp/foo.bar" lifecycle { action_trigger { events = [after_create, after_update] actions = [action.local_command.bar] } } } action "local_command" "bar" { config { command = "cat" arguments = ["/tmp/foo.bar"] } } ``` ### Debug Output https://gist.github.com/jeremmfr/87c348af5c77d22e5e10c8133d99e821 ### Expected Behavior When using `lifecycle.action_trigger` with `after_create` or `after_update` events, the triggered action should execute **after** the resource has been fully created/updated and its side effects (like file creation) are completed. In the example above, the `local_command` action should run after the file `/tmp/foo.bar` has been successfully written to disk, allowing the `cat` command to read its contents. ### Actual Behavior The action is triggered before the resource creation is completed. The `local_command` action fails because the file `/tmp/foo.bar` does not exist yet when the action attempts to read it. This suggests that `after_create` and `after_update` events are firing before the resource's side effects are finalized, despite the "after_" prefix implying completion. ### Steps to Reproduce 1. Create a Terraform configuration with the above code 2. `terraform init` 3. `terraform apply` 4. Observe that the action fails with an error indicating the file does not exist ### Additional Context This timing issue affects the reliability of `action_trigger` when actions depend on the actual completion of resource operations, not just the API call completion. ### References None ### Generative AI / LLM assisted development? No