Overview

Date: 2026-05-11

Time: 15:50

I'll create a summary entry for this GCP teardown example.

Overview

This is a GCP resource teardown script that demonstrates how to use FTL2 to delete cloud infrastructure in the correct dependency order. It serves as the complement to examplegcpprovision.py, showing the full lifecycle pattern of provisioning and deprovisioning GCP resources using FTL2's google.cloud.* modules.

Usage Patterns

The script uses the standard FTL2 async automation pattern:

async with automation(secret_bindings=bindings, verbose=True) as ftl:
    await ftl.google.cloud.gcp_compute_instance(
        name="ftl2-web01",
        zone="us-central1-a",
        state="absent",
    )

Resources are deleted by calling the same modules used for provisioning but with state="absent". Deletion order is the reverse of creation order to respect dependencies: instance → firewalls → subnet → network.

API and Configuration

Environment variables (bound via secret_bindings):

• GCP_PROJECT — GCP project ID
• GCPAUTHKIND — authentication type (e.g. serviceaccount)
• GCPSERVICEACCOUNT_FILE — path to service account JSON key

Secret bindings pattern: "google.cloud.*" wildcard binds all three env vars to every module in the google.cloud namespace, so individual module calls don't need to repeat auth parameters.

Automation options: verbose=True enables detailed output during execution.

Key Behaviors

• Reverse dependency ordering is manual — the user must sequence deletions so that dependents are removed before their dependencies (e.g., instances before the network they live on).
• state="absent" is the universal pattern for resource deletion across FTL2 modules. The same module handles both creation and removal.
• ftl.failed is checked at the end to determine if any module call failed during the teardown. This is a property on the automation context, not per-call.
• Minimal parameters are needed for deletion — only enough to identify the resource (name, and zone/region where applicable).

Relationships

• Companion script: examplegcpprovision.py creates the resources this script tears down.
• Core import: ftl2.automation.automation — the async context manager entry point for all FTL2 scripts.
• Modules used: gcpcomputeinstance, gcpcomputefirewall, gcpcomputesubnetwork, gcpcomputenetwork — all under the google.cloud namespace.
• External services: Google Cloud Platform Compute Engine API, authenticated via service account credentials.