# Exporters > Process and export your telemetry data --- 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 [Python 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]: [vendors]: /ecosystem/vendors/ 本页面介绍了主要的 OpenTelemetry Python 导出器以及如何进行配置。

注意

如果你使用了[零代码自动插桩]()，你可以参考[配置指南]()来了解如何设置导出器。

## OTLP ### Collector 设置 {#collector-setup}

注意

如果你已经配置好 OTLP Collector 或后端，可以跳过此部分，直接[设置应用的 OTLP 导出器依赖](#otlp-dependencies)。

为测试和验证你的 OTLP 导出器，你可以运行一个 Docker 容器形式的 Collector，将遥测数据直接输出到控制台。在一个空目录下创建名为 `collector-config.yaml` 的文件，并添加以下内容： ```yaml receivers: otlp: protocols: grpc: endpoint: 0.0.0.0:4317 http: endpoint: 0.0.0.0:4318 exporters: debug: verbosity: detailed service: pipelines: traces: receivers: [otlp] exporters: [debug] metrics: receivers: [otlp] exporters: [debug] logs: receivers: [otlp] exporters: [debug] ``` 然后运行以下命令，在 Docker 容器中启动 Collector： ```shell docker run -p 4317:4317 -p 4318:4318 --rm -v $(pwd)/collector-config.yaml:/etc/otelcol/config.yaml otel/opentelemetry-collector ``` 现在，这个 Collector 已能通过 OTLP 接收遥测数据。之后你可能需要配置 Collector，将遥测数据发送到你的可观测性后端。 {{__hugo_ctx/}} ## Dependencies {#otlp-dependencies} If you want to send telemetry data to an OTLP endpoint (like the [OpenTelemetry Collector](#collector-setup), [Jaeger](#jaeger) or [Prometheus](#prometheus)), you can choose between two different protocols to transport your data: - [HTTP/protobuf](https://pypi.org/project/opentelemetry-exporter-otlp-proto-http/) - [gRPC](https://pypi.org/project/opentelemetry-exporter-otlp-proto-grpc/) Start by installing the respective exporter packages as a dependency for your project:

pip install opentelemetry-exporter-otlp-proto-http

pip install opentelemetry-exporter-otlp-proto-grpc

## Usage Next, configure the exporter to point at an OTLP endpoint in your code.

from opentelemetry.sdk.resources import SERVICE_NAME, Resource

from opentelemetry import trace
from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
from opentelemetry.sdk.trace import TracerProvider
from opentelemetry.sdk.trace.export import BatchSpanProcessor

from opentelemetry import metrics
from opentelemetry.exporter.otlp.proto.http.metric_exporter import OTLPMetricExporter
from opentelemetry.sdk.metrics import MeterProvider
from opentelemetry.sdk.metrics.export import PeriodicExportingMetricReader

# Service name is required for most backends
resource = Resource.create(attributes={
    SERVICE_NAME: "your-service-name"
})

tracerProvider = TracerProvider(resource=resource)
processor = BatchSpanProcessor(OTLPSpanExporter(endpoint="<traces-endpoint>/v1/traces"))
tracerProvider.add_span_processor(processor)
trace.set_tracer_provider(tracerProvider)

reader = PeriodicExportingMetricReader(
    OTLPMetricExporter(endpoint="<traces-endpoint>/v1/metrics")
)
meterProvider = MeterProvider(resource=resource, metric_readers=[reader])
metrics.set_meter_provider(meterProvider)

from opentelemetry.sdk.resources import SERVICE_NAME, Resource

from opentelemetry import trace
from opentelemetry.exporter.otlp.proto.grpc.trace_exporter import OTLPSpanExporter
from opentelemetry.sdk.trace import TracerProvider
from opentelemetry.sdk.trace.export import BatchSpanProcessor

from opentelemetry import metrics
from opentelemetry.exporter.otlp.proto.grpc.metric_exporter import OTLPMetricExporter
from opentelemetry.sdk.metrics import MeterProvider
from opentelemetry.sdk.metrics.export import PeriodicExportingMetricReader

# Service name is required for most backends
resource = Resource.create(attributes={
    SERVICE_NAME: "your-service-name"
})

tracerProvider = TracerProvider(resource=resource)
processor = BatchSpanProcessor(OTLPSpanExporter(endpoint="your-endpoint-here"))
tracerProvider.add_span_processor(processor)
trace.set_tracer_provider(tracerProvider)

reader = PeriodicExportingMetricReader(
    OTLPMetricExporter(endpoint="localhost:5555")
)
meterProvider = MeterProvider(resource=resource, metric_readers=[reader])
metrics.set_meter_provider(meterProvider)

## Console To debug your instrumentation or see the values locally in development, you can use exporters writing telemetry data to the console (stdout). The `ConsoleSpanExporter` and `ConsoleMetricExporter` are included in the `opentelemetry-sdk` package. ```python from opentelemetry.sdk.resources import SERVICE_NAME, Resource from opentelemetry import trace from opentelemetry.sdk.trace import TracerProvider from opentelemetry.sdk.trace.export import BatchSpanProcessor, ConsoleSpanExporter from opentelemetry import metrics from opentelemetry.sdk.metrics import MeterProvider from opentelemetry.sdk.metrics.export import PeriodicExportingMetricReader, ConsoleMetricExporter # Service name is required for most backends, # and although it's not necessary for console export, # it's good to set service name anyways. resource = Resource.create(attributes={ SERVICE_NAME: "your-service-name" }) tracerProvider = TracerProvider(resource=resource) processor = BatchSpanProcessor(ConsoleSpanExporter()) tracerProvider.add_span_processor(processor) trace.set_tracer_provider(tracerProvider) reader = PeriodicExportingMetricReader(ConsoleMetricExporter()) meterProvider = MeterProvider(resource=resource, metric_readers=[reader]) metrics.set_meter_provider(meterProvider) ``` > [!NOTE] > > There are temporality presets for each instrumentation kind. These presets can > be set with the environment variable > `OTEL_EXPORTER_OTLP_METRICS_TEMPORALITY_PREFERENCE`, for example: > > ```sh > export OTEL_EXPORTER_OTLP_METRICS_TEMPORALITY_PREFERENCE="DELTA" > ``` > > The default value for `OTEL_EXPORTER_OTLP_METRICS_TEMPORALITY_PREFERENCE` is > `"CUMULATIVE"`. > > The available values and their corresponding settings for this environment > variable are: > > - `CUMULATIVE` > - `Counter`: `CUMULATIVE` > - `UpDownCounter`: `CUMULATIVE` > - `Histogram`: `CUMULATIVE` > - `ObservableCounter`: `CUMULATIVE` > - `ObservableUpDownCounter`: `CUMULATIVE` > - `ObservableGauge`: `CUMULATIVE` > - `DELTA` > - `Counter`: `DELTA` > - `UpDownCounter`: `CUMULATIVE` > - `Histogram`: `DELTA` > - `ObservableCounter`: `DELTA` > - `ObservableUpDownCounter`: `CUMULATIVE` > - `ObservableGauge`: `CUMULATIVE` > - `LOWMEMORY` > - `Counter`: `DELTA` > - `UpDownCounter`: `CUMULATIVE` > - `Histogram`: `DELTA` > - `ObservableCounter`: `CUMULATIVE` > - `ObservableUpDownCounter`: `CUMULATIVE` > - `ObservableGauge`: `CUMULATIVE` > > Setting `OTEL_EXPORTER_OTLP_METRICS_TEMPORALITY_PREFERENCE` to any other value > than `CUMULATIVE`, `DELTA` or `LOWMEMORY` will log a warning and set this > environment variable to `CUMULATIVE`. ## Jaeger ### 后端设置 {#jaeger-backend-setup} [Jaeger](https://www.jaegertracing.io/) 原生支持 OTLP，用于接收链路数据。你可以通过运行一个 Docker 容器来启动 Jaeger，其 UI 默认在端口 16686 上可访问，并在端口 4317 和 4318 上启用 OTLP： ```shell docker run --rm \ -e COLLECTOR_ZIPKIN_HOST_PORT=:9411 \ -p 16686:16686 \ -p 4317:4317 \ -p 4318:4318 \ -p 9411:9411 \ jaegertracing/all-in-one:latest ``` ### 使用方法 {#jaeger-usage} 现在，按照说明设置 [OTLP 导出器](#otlp-dependencies)。 {{__hugo_ctx/}} ## Prometheus 要将你的指标数据发送到 [Prometheus](https://prometheus.io/)，你可以选择[启用 Prometheus 的 OTLP 接收器](https://prometheus.io/docs/guides/opentelemetry/#enable-the-otlp-receiver)并且使用 [OTLP 导出器](#otlp)，或者使用 Prometheus 导出器，这是一种 `MetricReader`，他启动一个 HTTP 服务器，根据请求收集指标并将数据序列化为 Prometheus 文本格式。 ### 后端设置 {#prometheus-setup}

注意

如果你已经设置了 Prometheus 或兼容 Prometheus 的后端，可以跳过本节，直接为你的应用设置 [Prometheus](#prometheus-dependencies) 或者 [OTLP](#otlp-dependencies) 导出器依赖。

你可以按照以下步骤在 Docker 容器中运行 [Prometheus](https://prometheus.io)，并通过端口 9090 访问：创建一个名为 `prometheus.yml` 的文件，并将以下内容写入文件： ```yaml scrape_configs: - job_name: dice-service scrape_interval: 5s static_configs: - targets: [host.docker.internal:9464] ``` 使用以下命令在 Docker 容器中运行 Prometheus，UI 可通过端口 `9090` 访问： ```shell docker run --rm -v ${PWD}/prometheus.yml:/prometheus/prometheus.yml -p 9090:9090 prom/prometheus --web.enable-otlp-receiver ```

注意

当使用 Prometheus 的 OTLP 接收器时，确保在应用中设置 OTLP 端点为 `http://localhost:9090/api/v1/otlp`。并非所有的 Docker 环境都支持 `host.docker.internal`。在某些情况下，你可能需要将 `host.docker.internal` 替换为 `localhost` 或你机器的 IP 地址。

{{__hugo_ctx/}} ## Dependencies {#prometheus-dependencies} Install the [exporter package](https://pypi.org/project/opentelemetry-exporter-prometheus/) as a dependency for your application: ```sh pip install opentelemetry-exporter-prometheus ``` Update your OpenTelemetry configuration to use the exporter and to send data to your Prometheus backend: ```python from prometheus_client import start_http_server from opentelemetry import metrics from opentelemetry.exporter.prometheus import PrometheusMetricReader from opentelemetry.sdk.metrics import MeterProvider from opentelemetry.sdk.resources import SERVICE_NAME, Resource # Service name is required for most backends resource = Resource.create(attributes={ SERVICE_NAME: "your-service-name" }) # Start Prometheus client start_http_server(port=9464, addr="localhost") # Initialize PrometheusMetricReader which pulls metrics from the SDK # on-demand to respond to scrape requests reader = PrometheusMetricReader() provider = MeterProvider(resource=resource, metric_readers=[reader]) metrics.set_meter_provider(provider) ``` With the above you can access your metrics at . Prometheus or an OpenTelemetry Collector with the Prometheus receiver can scrape the metrics from this endpoint. ## Zipkin ### 后端设置 {#zipkin-setup}

提示

如果你已经设置了 Zipkin 或兼容 Zipkin 的后端，可以跳过本节并直接为你的应用设置 [Zipkin 导出器依赖](#zipkin-dependencies)。

你可以通过执行以下命令，在 Docker 容器中运行 [Zipkin](https://zipkin.io/)： ```shell docker run --rm -d -p 9411:9411 --name zipkin openzipkin/zipkin ``` {{__hugo_ctx/}} ## Dependencies {#zipkin-dependencies} To send your trace data to [Zipkin](https://zipkin.io/), you can choose between two different protocols to transport your data: - [HTTP/protobuf](https://pypi.org/project/opentelemetry-exporter-zipkin-proto-http/) - [Thrift](https://pypi.org/project/opentelemetry-exporter-zipkin-json/) Install the exporter package as a dependency for your application:

pip install opentelemetry-exporter-zipkin-proto-http

pip install opentelemetry-exporter-zipkin-json

Update your OpenTelemetry configuration to use the exporter and to send data to your Zipkin backend:

from opentelemetry import trace
from opentelemetry.exporter.zipkin.proto.http import ZipkinExporter
from opentelemetry.sdk.trace import TracerProvider
from opentelemetry.sdk.trace.export import BatchSpanProcessor
from opentelemetry.sdk.resources import SERVICE_NAME, Resource

resource = Resource.create(attributes={
    SERVICE_NAME: "your-service-name"
})

zipkin_exporter = ZipkinExporter(endpoint="http://localhost:9411/api/v2/spans")

provider = TracerProvider(resource=resource)
processor = BatchSpanProcessor(zipkin_exporter)
provider.add_span_processor(processor)
trace.set_tracer_provider(provider)

from opentelemetry import trace
from opentelemetry.exporter.zipkin.json import ZipkinExporter
from opentelemetry.sdk.trace import TracerProvider
from opentelemetry.sdk.trace.export import BatchSpanProcessor
from opentelemetry.sdk.resources import SERVICE_NAME, Resource

resource = Resource.create(attributes={
    SERVICE_NAME: "your-service-name"
})

zipkin_exporter = ZipkinExporter(endpoint="http://localhost:9411/api/v2/spans")

provider = TracerProvider(resource=resource)
processor = BatchSpanProcessor(zipkin_exporter)
provider.add_span_processor(processor)
trace.set_tracer_provider(provider)

## 自定义导出器 {#custom-exporters} 最后，你还可以编写自己的导出器。有关更多信息，请参见 [API 文档中的 SpanExporter 接口](https://opentelemetry-python.readthedocs.io/en/latest/sdk/trace.export.html#opentelemetry.sdk.trace.export.SpanExporter). ## 批量处理 Span 和日志记录 {#batching-span-and-log-records} OpenTelemetry SDK 提供了一组默认的 Span 和日志记录处理器，允许你选择按单条（simple）或按批量（batch）方式导出一个或多个 Span。推荐使用批量处理，但如果你不想批量处理 Span 或日志记录，可以使用 simple 处理器，方法如下： {{__hugo_ctx/}} ```python from opentelemetry.exporter.otlp.proto.grpc.trace_exporter import OTLPSpanExporter from opentelemetry.sdk.trace.export import SimpleSpanProcessor processor = SimpleSpanProcessor(OTLPSpanExporter(endpoint="your-endpoint-here")) ```