My Cloud Run service switchyard-golden-match in australia-southeast1 is Ready, and its startup TCP probe succeeds on port 8080. Both service URLs return Google’s 404 for /healthz from Cloud Shell and my laptop browser. The deployed source defines GET /healthz. Billing is enabled, the project is ACTIVE, ingress is all, the default URL is enabled, and public invocation is allowed. My latest Cloud Logging query returned no matching request logs.
What read-only check could identify why Google’s frontend returns 404 before the request appears in the service logs? I have not changed IAM or redeployed.