Tutorial 28: gRPC Microservices with Tonic 🌐

Generate a Tonic/Protobuf starting point using cargo rullst make:grpc. The command does not start a gRPC server, add every build dependency, or define production transport/authentication policy for the application.

🛠️ Step 1: Generate a gRPC Service

cargo rullst make:grpc UserService

This generates:

• proto/user_service.proto (Protobuf definition)
• src/grpc/user_service.rs (Tonic service implementation)

Names ending in Service remain a single service suffix: UserService produces the user_service_server::UserService trait rather than UserServiceService.

💻 Step 2: Implement the gRPC Handler

In src/grpc/user_service.rs, after the generated application’s build.rs has compiled proto/user_service.proto and made the Tonic dependencies available:

use tonic::{Request, Response, Status};

pub mod proto {
    tonic::include_proto!("user_service");
}

use proto::user_service_server::UserService;
use proto::{HelloRequest, HelloResponse};

#[derive(Debug, Default)]
pub struct UserServiceImpl;

#[tonic::async_trait]
impl UserService for UserServiceImpl {
    async fn say_hello(
        &self,
        request: Request<HelloRequest>,
    ) -> Result<Response<HelloResponse>, Status> {
        let reply = HelloResponse {
            message: format!("Hello {} from Rullst gRPC!", request.into_inner().name),
        };
        Ok(Response::new(reply))
    }
}

💡 Key Takeaways

• The generator creates a starting point for tonic and prost; inspect the generated service and add/review the required build.rs, dependencies, reflection/health endpoints, and server bootstrap before deployment.
• Apply TLS or mTLS, authentication, per-method authorization, deadlines, message-size/concurrency limits, and proxy policy explicitly.
• Network, serialization, handler, and proxy latency must be measured in the target environment. Rullst does not claim a universal latency bound.