Google ACE Practice Question: A Cloud Run service has been running for weeks

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Cloud Monitoring Metrics Explorer — filter by request_count metric with error status

  Why it's wrong here

  Cloud Monitoring Metrics Explorer is designed for time-series aggregation, not log exploration. While the request_count metric (or a log-based metric derived from it) can show error-rate trends, it aggregates counts into buckets and cannot display individual request entries with URL paths, user agents, or detailed status code breakdowns needed to troubleshoot specific 5xx endpoints. For per-request diagnostic details you must inspect the underlying Cloud Logging entries.

• ✓

  Cloud Logging Logs Explorer — filter by resource type 'cloud_run_revision'

  Why this is correct

  Cloud Run automatically streams structured request logs to Cloud Logging, and each entry includes an httpRequest object with the request URL, response status, latency, and protocol. In Logs Explorer, filter using resource.type="cloud_run_revision" and add a filter like httpRequest.status >= 500 to isolate failed requests; you can then expand each log entry to see the exact endpoint, caller IP, and response details. This gives you the forensic, per-request view needed to identify which endpoints are returning errors.

• ✗

  Cloud Trace — filter by 5xx response code

  Why it's wrong here

  Cloud Trace exists to analyze latency and distributed call paths across services; it records spans only when your application is instrumented with the Cloud Trace SDK or uses a supported framework, and services without propagation won't be represented. Even when instrumentation is present, filtering by 5xx response code in Trace shows trace metadata, not the full request log payload (headers, body, URL path details) that Cloud Logging captures, so you cannot rely on it for complete error diagnostics.

• ✗

  Cloud Debugger — set a snapshot at the error handler

  Why it's wrong here

  Cloud Debugger snapshots capture the call stack and local variables of a single live execution when code hits a breakpoint, and they expire quickly — they do not provide a historical log of past requests. Moreover, Cloud Run's serverless instances are ephemeral and snapshots require attaching to a running instance at the exact moment the error handler executes, which is not a viable way to review prior 5xx responses. It answers "what is the state now?" not "what requests failed and why?".

Go deeper

Related to this question