Using instrumentation libraries

当你开发应用时，可能会使用第三方库和框架来加快开发进度。如果你随后使用 OpenTelemetry 对应用进行插桩，你可能希望避免额外花时间为所用的第三方库和框架手动添加链路、日志和指标。

许多库和框架已经原生支持 OpenTelemetry，或者通过 OpenTelemetry 的插桩获得支持， 因此它们能够生成可导出到可观测性后端的遥测数据。

如果你正在为使用第三方库或框架的应用或服务进行插桩， 请按照以下说明学习如何为你的依赖项使用原生插桩库和插桩库。

使用原生插桩库

如果某个库默认就支持 OpenTelemetry，你只需在应用中添加并配置 OpenTelemetry SDK， 就可以获取该库发出的链路、指标和日志。

该库可能需要一些额外的插桩配置。请查阅该库的文档以了解更多信息。

• Elasticsearch Python Client

Use instrumentation libraries

If a library does not ship with native OpenTelemetry support, you can use instrumentation libraries to generate telemetry data for a library or framework.

For example, the instrumentation library for HTTPX automatically creates spans based on HTTP requests.

Setup

You can install each instrumentation library separately using pip. For example:

pip install opentelemetry-instrumentation-{instrumented-library}

In the previous example, {instrumented-library} is the name of the instrumentation.

To install a development version, clone or fork the opentelemetry-python-contrib repository and run the following command to do an editable installation:

pip install -e ./instrumentation/opentelemetry-instrumentation-{integration}

After installation, you will need to initialize the instrumentation library. Each library typically has its own way to initialize.

Example with HTTPX instrumentation

Here’s how you can instrument HTTP requests made using the httpx library.

First, install the instrumentation library using pip:

pip install opentelemetry-instrumentation-httpx

Next, use the instrumentor to automatically trace requests from all clients:

import httpx
from opentelemetry.instrumentation.httpx import HTTPXClientInstrumentor

url = "https://some.url/get"
HTTPXClientInstrumentor().instrument()

with httpx.Client() as client:
     response = client.get(url)

async with httpx.AsyncClient() as client:
     response = await client.get(url)

Turn off instrumentations

If needed, you can uninstrument specific clients or all clients using the uninstrument_client method. For example:

import httpx
from opentelemetry.instrumentation.httpx import HTTPXClientInstrumentor

HTTPXClientInstrumentor().instrument()
client = httpx.Client()

# Uninstrument a specific client
HTTPXClientInstrumentor.uninstrument_client(client)

# Uninstrument all clients
HTTPXClientInstrumentor().uninstrument()

Available instrumentation libraries

A full list of instrumentation libraries produced by OpenTelemetry is available from the opentelemetry-python-contrib repository.

You can also find more instrumentations available in the registry.

Next steps

After you have set up instrumentation libraries, you might want to add your own instrumentation to your code, to collect custom telemetry data.

You might also want to configure an appropriate exporter to export your telemetry data to one or more telemetry backends.

You can also check the Zero-code instrumentation for Python.