This repository contains ddtrace, Datadog's APM client Python package. ddtrace contains APIs to automatically or manually trace and profile Python applications.

These features power Distributed Tracing, Continuous Profiling, Error Tracking, Continuous Integration Visibility, Deployment Tracking, Code Hotspots and more.

To get started, check out the setup documentation.

For advanced usage and configuration, check out the API documentation.

Confused about the terminology of APM? Take a look at the APM Glossary.

See the contributing docs first.

The tracer library uses formatting/linting tools including black, flake8, and mypy. While these are run in each CI pipeline for pull requests, they are automated to run when you call git commit as pre-commit hooks to catch any formatting errors before you commit. To initialize the pre-commit hook script to run in your development branch, run the following command:

$ hooks/autohook.sh install

The test suite requires many backing services such as PostgreSQL, MySQL, Redis and more. We use docker and docker-compose to run the services in our CI and for development. To run the test matrix, please install docker and docker-compose using the instructions provided by your platform. Then launch them through:

$ docker-compose up -d

1. Clone the repository locally: git clone https://github.com/DataDog/dd-trace-py
2. The tests for this project run on various versions of Python. We recommend using a Python version management tool, such as pyenv, to utilize multiple versions of Python. How to install Pyenv
3. Install the relevant versions of Python in Pyenv: pyenv install 2.7.18 3.5.10 3.6.15 3.7.13 3.8.13 3.9.13 3.10.5
4. Make those versions available globally: pyenv global 2.7.18 3.5.10 3.6.15 3.7.13 3.8.13 3.9.13 3.10.5

Once your docker-compose environment is running, you can use the shell script to execute tests within a Docker image. You can start the container with a bash shell:

$ scripts/ddtest

You can now run tests as you would do in your local environment. We use tox as well as riot, a new tool that we developed for addressing our specific needs with an ever growing matrix of tests. You can list the tests managed by each:

$ tox -l
$ riot list

You can run multiple tests by using regular expressions:

$ scripts/run-tox-scenario '^futures_contrib-'
$ riot run psycopg

1. Install riot: pip install riot.
2. Create the base virtual environments: riot -v generate.
3. You can list the available test suites with riot list.
4. Certain tests might require running service containers in order to emulate the necessary testing environment. You can spin up individual containers with docker-compose up -d <SERVICE_NAME>, where <SERVICE_NAME> should match a service specified in the docker-compose.yml file.
5. Run a test suite: riot -v run <RUN_FLAGS> <TEST_SUITE_NAME>.
   1. Optionally, use the -s and -x flags: -s prevents riot from reinstalling the dev package; -x forces an exit after the first failed test suite. To limit the tests to a particular version of Python, use the -p flag: riot -v run -p <PYTHON_VERSION>.

The run command uses regex syntax, which in some cases will cause multiple test suites to run. Use the following syntax to ensure only an individual suite runs: ^<TEST_SUITE_NAME>$ where ^ signifies the start of a string and $ signifies the end of a string. For example, use riot -v run -s -x ^redis$ to run only the redis suite.

The APM test agent can emulate the APM endpoints of the Datadog agent. Spin up the testagent container along with any other service container:

$ docker-compose up -d testagent <SERVICE_CONTAINER>

Run the test agent as a proxy in your tests:

$ DD_TRACE_AGENT_URL=http://localhost:9126/ riot -v run <RUN_FLAGS> --pass-env <TEST_SUITE_NAME>

--pass-env injects the environment variables of the current shell session into the command. Here's an example command for running the redis test suite along with the test agent, limited to tests for Python 3.9:

$ DD_TRACE_AGENT_URL=http://localhost:9126/ riot -v run -p 3.9 -s -x --pass-env '^redis$'

Read more about the APM test agent: https://github.com/datadog/dd-apm-test-agent#readme

We use CircleCI 2.0 for our continuous integration.

The CI tests are configured through config.yml.

The CI tests can be run locally using the circleci CLI. More information about the CLI can be found at https://circleci.com/docs/2.0/local-cli/.

After installing the circleci CLI, you can run jobs by name. For example:

$ circleci build --job django

This project follows semver and so bug fixes, breaking changes, new features, etc must be accompanied by a release note.

See the contributing docs for instructions on generating, writing, formatting, and styling release notes.