Google ACE Deploying and Implementing a Cloud Solution Practice Question

Answer analysis

Option-by-option breakdown

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

• ✓

  The service is hitting the maximum number of concurrent requests per container instance (default 80) and needs more instances.

  Why this is correct

  A 503 from Cloud Run specifically signals that a request arrived but no container instance was available to accept it within the timeout window. Each instance can process only a fixed number of concurrent requests—the default concurrency is 80—so when all existing instances are saturated and the autoscaler cannot add new instances quickly enough (or the 'max instances' setting has been reached), the server returns Service Unavailable. The fix is to raise the max instances limit, lower the concurrency setting, or enable additional CPU to reduce per-instance bottleneck.

• ✗

  The region us-central1 does not support Cloud Run.

  Why it's wrong here

  Cloud Run is a fully managed serverless platform that operates in us-central1 and is one of the earliest supported regions. The service would have been deployed there without any regional restriction, and regional unavailability would produce a resource error at deploy time, not an HTTP 503 on an existing service. Therefore a regional limitation cannot be the cause.

• ✗

  The container image is not compatible with the managed platform.

  Why it's wrong here

  If a Cloud Run container image were incompatible with the managed platform, it would fail during startup or health checks, typically resulting in 500-level errors like 'Container failed to start' or a CrashLoopBackOff state. The deployment would never become ready to serve traffic, so it would not return intermittent 503 responses under load. Compatibility issues relate to the container listening on $PORT or the correct runtime, but those manifest as startup failures, not scaling symptoms.

• ✗

  The --allow-unauthenticated flag causes IAM permission errors.

  Why it's wrong here

  The --allow-unauthenticated flag simply configures the service's IAM policy to let the public invoke it; without it, requests lack identity and Cloud Run would return 401 or 403, not 503. This flag does not interfere with the service's ability to scale, so it has no bearing on concurrency-related availability. Thus it cannot be responsible for the occasional 503 responses you observe.

Go deeper

Related to this question