The 'Inventory Service' is experiencing extreme latency and intermittent timeouts. How should an API Gateway with a 'Circuit Breaker' pattern react to protect the rest of the system?

1. A It should automatically increase the request timeout to 60 seconds to give the service more 'breathing room'.
2. B It should retry every failed request 10 times across different instances to ensure eventual success.
3. C It should 'Open' the circuit: immediately returning a 503 error or a cached fallback response without even attempting to call the failing service.
4. D It should restart the Inventory Service containers automatically using a shell script.