Extension chaostracing

Version 0.1.2
Repository https://github.com/chaostoolkit-incubator/chaostoolkit-opentracing

Build Status Python versions

This project is an extension for the Chaos Toolkit for OpenTracing.

Here is an example of what it could look like with the Jaeger backend.



This package requires Python 3.5+

To be used from your experiment, this package must be installed in the Python environment where chaostoolkit already lives.

$ pip install -U chaostoolkit-opentracing


Currently, this extension only provides control support to send traces to your provider during the execution of the experiment. It does not yet expose any probes or actions per-se.

To use this control, add the following section to your experiment, at the top-level:

    "configuration": {
        "tracing_provider": "jaeger",
        "tracing_host": "",
        "tracing_port": 6831
    "controls": [
            "name": "opentracing",
            "provider": {
                "type": "python",
                "module": "chaostracing.control"

This will automatically create a Jaeger client to emit traces onto the address

Use from other extensions

You may also access the tracer from other extensions as follows:

import opentracing

def some_function(...):

As not all Open Tracing providers support yet to fetch the active span from the tracer (Open Tracing 2 specification), we attach the following attributes to the tracer instance:

tracer.experiment_span  # span during the lifetime of the experiment
tracer.hypothesis_span  # span during the lifetime of the hypothesis
tracer.method_span  # span during the lifetime of the method
tracer.rollback_span  # span during the lifetime of the rollback
tracer.activity_span  # span during the lifetime of an activity

For instance, assuming you have an extension that makes a HTTP call you want to trace specifically, you could do this from your extension’s code:

import opentracing
import requests

def my_activity(...):
    headers = {}

    tracer = opentracing.tracer
    parent_span = tracer.activity_span
    span = tracer.start_span("my-inner-span", child_of=parent_span)
    span.set_tag('http.url', url)
    span.set_tag('span.kind', 'client')
    span.tracer.inject(span, 'http_headers', headers)

    r = requests.get(url, headers=headers)

    span.set_tag('http.status_code', r.status_code)

Because the opentracing exposes a noop tracer when non has been initialized, it should be safe to have that code in your extensions without having to determine if the extension has been enabled in the experiment.

Open Tracing Provider Support

For now, only the Jaeger tracer is supported but other backends will be added as need be in the future.

Jaeger tracer

To install the necessary dependencies for the Jaeger tracer, please run:

$ pip install chaostoolkit-opentracing[jaeger]

Unfortunately, the Jaeger client does not yet support Open Tracing 2.0.


To run the tests for the project execute the following:

$ pytest


If you wish to contribute more functions to this package, you are more than welcome to do so. Please, fork this project, make your changes following the usual PEP 8 code style, sprinkling with tests and submit a PR for review.

The Chaos Toolkit projects require all contributors must sign a Developer Certificate of Origin on each commit they would like to merge into the master branch of the repository. Please, make sure you can abide by the rules of the DCO before submitting a PR.

Exported Controls

This package exports controls covering the following phases of the execution of an experiment:

Level Before After
Experiment True True
Steady-state Hypothesis True True
Method True True
Rollback True True
Activities True True

To use this control module, please add the following section to your experiment:

  "name": "chaostracing",
  "provider": {
    "type": "python",
    "module": "chaostracing.control"
name: chaostracing
  module: chaostracing.control
  type: python

This block may also be enabled at any other level (steady-state hypothesis or activity) to focus only on that level.

When enabled at the experiment level, by default, all sub-levels are also applied unless you set the automatic properties to false.