← All activity records

FACTUAL ACTIVITY RECORD · An activity record based on work that took place

Determining the first Broker release from real resources

Using managed Cloud Run state rather than Terraform output as proof of existence

Activity period:

Purpose

After a partial application, Prepare treated a computable Cloud Run name in Terraform output as proof that the service existed. The distinction between first deployment and update needed to come from actual resource presence.

Implementation

deployment-phase evaluator now receives raw Terraform state and selects the update phase only when a current managed google_cloud_run_v2_service instance exists. Output values are not used as existence evidence.

preparation workflow explicitly passes the pulled raw state into the decision. Regression tests cover partial application, an existing Cloud Run service, and a missing revision.

Phase detection uses a managed current instance in raw state rather than an output value.

Rendering diagram…

What was confirmed

Verification results

Three regression tests, linting, GCP verification, specification auditing, secret scanning, and the diff check passed.

Tests confirmed that a partial state containing only API Gateway and no Cloud Run service is classified as a first-release phase.

A real Prepare returning the first-release phase after the repair had not been verified by the cutoff.

Basis for completion

This scope ended when phase detection was limited to managed resources in raw state and three cases, including partial application, were fixed in tests.