Event-Driven vs. Request/Response for Internal Microservice Communication?

Hey everyone, I'm working on a new service that needs to interact with several other internal microservices. I'm torn between an event-driven approach (using a message broker) and a more traditional request/response model (REST/gRPC). The data flow isn't super high-volume, but we do need good fault tolerance and scalability as we grow. Are there specific scenarios where one clearly outperforms the other for internal, synchronous-ish operations? What are your experiences with maintainability and debugging in heavily event-driven internal systems? Any war stories or best practices for making this decision would be super helpful.