Overview

Observing an application in production means being able to answer what it is doing, how fast, and why it fails. Corvis does not ship an observability platform (tracing SDK, metrics, dashboard or APM provider integration). What it does offer are the basic pieces to build your own on top of:

Need	What Corvis has
Structured logs	ConsoleLogger::new().json(true) writes one JSON line per entry; LoggerService lets you send the logs wherever you want (see Logging)
Duration of each request	corvis::LoggingInterceptor records method, URL, status and milliseconds; or your own interceptor that measures and publishes the metric
Unhandled errors	A global exception filter sees every error before a response is sent; exception.is_internal() tells unexpected errors apart
Instance health	TerminusModule and HealthCheckService (see Health checks)

The other pages in this section describe features that do not exist in Corvis yet and, where applicable, how to cover the need with these pieces.

Example: a metrics interceptor

use std::time::Instant;

#[injectable]
pub struct MetricsInterceptor {
    #[init(Logger::new("Metrics"))]
    logger: Logger,
}

#[async_trait]
impl CorvisInterceptor for MetricsInterceptor {
    async fn intercept(&self, ctx: &ExecutionContext, next: CallHandler) -> HandlerResult {
        let started = Instant::now();
        let route = format!("{}.{}", ctx.get_class().name, ctx.get_handler().name);
        let result = next.handle().await;
        let status = match &result {
            Ok(r) => r.status().map(|s| s.as_u16()).unwrap_or(200),
            Err(e) => e.get_status(),
        };
        self.logger.log(format!("{route} {status} {}us", started.elapsed().as_micros()));
        result
    }
}

Register it as global with Provider::app_interceptor::<MetricsInterceptor>() in the root module. Instead of writing a log, you can publish the value to the metrics system you use.