Installation + Quickstart#

Before installing be sure to read through the setup documentation to ensure your environment is ready to receive traces.

Installation#

Install with pip:

pip install ddtrace

Important

pip version 18 and above is required to install the library.

Quickstart#

Important

Using gevent? Read our gevent documentation.

Using Gunicorn? Read the Gunicorn documentation.

Using uWSGI? Read our uWSGI documentation.

Tracing#

Getting started for tracing is as easy as prefixing your python entry-point command with ddtrace-run.

For example if you start your application with python app.py then run (with your desired settings in place of the example environment variables):

DD_SERVICE=app DD_ENV=dev DD_VERSION=0.1 ddtrace-run python app.py

For more advanced usage of ddtrace-run refer to the documentation here.

To verify the environment configuration for your application run the command ddtrace-run --info. This will print out info useful for debugging to make sure your environment variable configurations are being picked up correctly and that the tracer will be able to connect to the Datadog agent with them. Note: --info Only reflects configurations made via environment variables, not those made in code.

When ddtrace-run cannot be used, a similar start-up behavior can be achieved with the import of ddtrace.auto. This should normally be imported as the first thing during the application start-up.

If ddtrace-run isn’t suitable for your application, then ddtrace.patch_all() can be used to configure the tracer:

from ddtrace import config, patch_all

config.env = "dev"      # the environment the application is in
config.service = "app"  # name of your application
config.version = "0.1"  # version of your application
patch_all()

Note

We recommend the use of ddtrace-run when possible. If you are importing ddtrace.auto as a programmatic replacement for ddtrace, then note that integrations will take their configuration from the environment variables. A call to ddtrace.patch_all() cannot be used to disable an integration at this point.

Service names also need to be configured for libraries that query other services (requests, grpc, database libraries, etc). Check out the integration documentation for each to set them up.

For additional configuration see the configuration documentation.

To learn how to manually instrument check out the basic usage documentation.

Profiling#

Profiling can also be auto enabled with ddtrace-run by providing the DD_PROFILING_ENABLED environment variable:

DD_PROFILING_ENABLED=true ddtrace-run python app.py

If ddtrace-run isn’t suitable for your application then ddtrace.profiling.auto can be used:

import ddtrace.profiling.auto

Configuration#

Almost all configuration of ddtrace can be done via environment variable. See the full list in Configuration.

OpenTracing#

ddtrace also provides an OpenTracing API to the Datadog tracer so that you can use the Datadog tracer in your OpenTracing-compatible applications.

Installation#

Include OpenTracing with ddtrace:

$ pip install ddtrace[opentracing]

To include the OpenTracing dependency in your project with ddtrace, ensure you have the following in setup.py:

install_requires=[
    "ddtrace[opentracing]",
],

Configuration#

The OpenTracing convention for initializing a tracer is to define an initialization method that will configure and instantiate a new tracer and overwrite the global opentracing.tracer reference.

Typically this method looks something like:

from ddtrace.opentracer import Tracer, set_global_tracer

def init_tracer(service_name):
    """
    Initialize a new Datadog opentracer and set it as the
    global tracer.

    This overwrites the opentracing.tracer reference.
    """
    config = {
      'agent_hostname': 'localhost',
      'agent_port': 8126,
    }
    tracer = Tracer(service_name, config=config)
    set_global_tracer(tracer)
    return tracer

For more advanced usage of OpenTracing in ddtrace refer to the documentation here.