# Exporters

LLMS index: [llms.txt](/llms.txt)

---

将遥测数据发送到 [OpenTelemetry Collector](/docs/collector/)，以确保其被正确导出。
在生产环境中使用 Collector 是最佳实践。若要可视化你的遥测数据，可将其导出到后端系统，例如
[Jaeger](https://jaegertracing.io/)、[Zipkin](https://zipkin.io/)、
[Prometheus](https://prometheus.io/)，或某个[特定厂商的](/ecosystem/vendors/)后端。



## 可用的导出器 {#available-exporters}

镜像仓库中包含一份 [Rust 可用导出器的列表][reg]。





在所有导出器中，[OpenTelemetry 协议 (OTLP)][OTLP] 导出器是以 OpenTelemetry 数据模型为基础设计的，
能够无信息丢失地输出 OTel 数据。此外，许多处理遥测数据的工具都支持 OTLP
（例如 [Prometheus][]、[Jaeger][] 和大多数[厂商][vendors]），在你需要时为你提供高度的灵活性。
若要了解更多关于 OTLP 的信息，请参阅 [OTLP 规范][OTLP]。

[Jaeger]: /blog/2022/jaeger-native-otlp/
[OTLP]: /docs/specs/otlp/
[Prometheus]: https://prometheus.io/docs/prometheus/2.55/feature_flags/#otlp-receiver
[reg]: </ecosystem/registry/?component=exporter&language=rust>
[vendors]: /ecosystem/vendors/



本页面介绍了主要的 OpenTelemetry Rust 导出器以及如何进行配置。








## OTLP endpoint

To send trace data to a OTLP endpoint (like the [collector](/docs/collector) or
Jaeger) you'll want to use an exporter crate, such as
[opentelemetry-otlp](https://crates.io/crates/opentelemetry-otlp):

For example, you can update the [Getting Started](../getting-started/) dice
server by adding the new dependency:

```toml
[dependencies]
opentelemetry-otlp = { version = "0.28.0", features = ["grpc-tonic"] }
```

Next, update `init_tracer_provider` in `dice_server.rs` to configure the
exporter to point at an OTLP endpoint:

```rust
use std::convert::Infallible;
use std::net::SocketAddr;
use std::sync::OnceLock;

use http_body_util::Full;
use hyper::{Method, Request, Response, body::Bytes, server::conn::http1, service::service_fn};
use hyper_util::rt::TokioIo;
use opentelemetry::global::{self, BoxedTracer};
use opentelemetry::trace::{Span, SpanKind, Status, Tracer};
use opentelemetry_otlp::SpanExporter;
use opentelemetry_sdk::{Resource, propagation::TraceContextPropagator, trace::SdkTracerProvider};
use rand::Rng;
use tokio::net::TcpListener;

// ...

fn init_tracer_provider() -> SdkTracerProvider {
    let exporter = SpanExporter::builder()
        .with_tonic()
        .build()
        .expect("Failed to create span exporter");
    let provider = SdkTracerProvider::builder()
        .with_resource(Resource::builder().with_service_name("dice_server").build())
        .with_batch_exporter(exporter)
        .build();
    global::set_text_map_propagator(TraceContextPropagator::new());
    global::set_tracer_provider(provider.clone());
    provider
}
```

To try out the OTLP exporter quickly, you can run Jaeger in a docker container.
Jaeger natively supports OTLP, so you only need to expose the web UI (`16686`)
and the OTLP gRPC endpoint (`4317`):

```shell
docker run -d --rm --name jaeger \
  -p 16686:16686 \
  -p 4317:4317 \
  jaegertracing/all-in-one:latest
```

By default, the OTLP exporter sends data to `http://localhost:4317`, which
matches the OTLP gRPC endpoint exposed by Jaeger above, so no additional
endpoint configuration is needed.

Make requests on
[http://localhost:8080/rolldice](http://localhost:8080/rolldice), then view the
traces in Jaeger:

1. Open [http://localhost:16686](http://localhost:16686) and refresh.
2. Select `dice_server` from the **Service** dropdown.
3. Click **Find Traces**.

Click a trace to open the trace details view, which shows the span hierarchy and
timing as a Gantt chart.

When you're done, stop the Jaeger container:

```shell
docker stop jaeger
```