Cloud Build in us-central1 accepts a manual public-pool build with custom service account, exact Cloud Storage source generation, own regional GCS logs bucket and own Artifact Registry output. Two240-second builds timed out before the first Docker step. One600-second final attempt failed after340.520580 seconds with native FAILURE, Docker step QUEUED, resolved source provenance, no step/pull timing, no logs/images, and no statusDetail/failureInfo/warnings. Audit confirms cloudbuild.builds.create granted then status code9 without a message. Default machine/disk, no private pool, nonzero public CPU quota. All retries stopped. What read-only diagnostic can reveal the exact failed precondition before changing IAM/resources or retrying? No credentials, customer data or private resource identifiers included.