Skip to content
Branch: master
Clone or download
Hamid Asaadi Copybara-Service
Hamid Asaadi and Copybara-Service Update external build metadata to state explicit support for Python 3.7.
The OSS issue is here: #8

PiperOrigin-RevId: 237494949
Change-Id: I7f0b17164baf6f51007ad523d44e0a25e10e8e89
Latest commit 89ce378 Mar 8, 2019
Type Name Latest commit message Commit time
Failed to load latest commit information.

Python Cloud Debugger Agent

Google Cloud Debugger for Python 2.7 and 3.7 (experimental support)


Cloud Debugger (also known as Stackdriver Debugger) lets you inspect the state of a running cloud application, at any code location, without stopping or slowing it down. It is not your traditional process debugger but rather an always on, whole app debugger taking snapshots from any instance of the app.

Cloud Debugger is safe for use with production apps or during development. The Python debugger agent only few milliseconds to the request latency when a debug snapshot is captured. In most cases, this is not noticeable to users. Furthermore, the Python debugger agent does not allow modification of application state in any way, and has close to zero impact on the app instances.

Cloud Debugger attaches to all instances of the app providing the ability to take debug snapshots and add logpoints. A snapshot captures the call-stack and variables from any one instance that executes the snapshot location. A logpoint writes a formatted message to the application log whenever any instance of the app executes the logpoint location.

The Python debugger agent is only supported on Linux at the moment. It was tested on Debian Linux, but it should work on other distributions as well.

Cloud Debugger consists of 3 primary components:

  1. The Python debugger agent (this repo implements one for CPython 2.7, and an experimental one for CPython 3.7).
  2. Cloud Debugger service storing and managing snapshots/logpoints. Explore the APIs using APIs Explorer.
  3. User interface, including a command line interface gcloud debug and a Web interface on Google Cloud Console. See the online help on how to use Google Cloud Console Debug page.

Getting Help

  1. StackOverflow:
  2. Send email to: Cloud Debugger Feedback
  3. Send Feedback from Google Cloud Console


The easiest way to install the Python Cloud Debugger is with PyPI:

pip install google-python-cloud-debugger

Alternatively, download the egg package from Releases and install the debugger agent with:

easy_install google_python_cloud_debugger-py2.7-linux-x86_64.egg

You can also build the agent from source code:

git clone
cd cloud-debug-python/src/
easy_install dist/google_python_cloud_debugger-*.egg

Note that the build script assumes some dependencies. To install these dependencies on Debian, run this command:

sudo apt-get -y -q --no-install-recommends install \
    curl ca-certificates gcc build-essential cmake \
    python python-dev libpython2.7 python-setuptools

Python 3

There is experimental support for Python 3.6 and Python 3.7. Python 3.0 to 3.5 are not supported, and newer versions have not been tested.

To build, the python3.7 and python3.7-dev packages are additionally needed. If Python 3.7 is not the default version of the 'python' command on your system, run the build script as PYTHON=python3.7 ./

Alpine Linux

The Python agent is not regularly tested on Alpine Linux, and support will be on a best effort basis. The Dockerfile shows how to build a minimal image with the agent installed.


Google Cloud Platform

  1. First, make sure that you created the VM with this option enabled:

    Allow API access to all Google Cloud services in the same project.

    This option lets the Python debugger agent authenticate with the machine account of the Virtual Machine.

    It is possible to use the Python debugger agent without it. Please see the next section for details.

  2. Install the Python debugger agent as explained in the Installation section.

  3. Enable the debugger in your application using one of the two options:

    Option A: add this code to the beginning of your main() function:

    # Attach Python Cloud Debugger
      import googleclouddebugger
      googleclouddebugger.enable(module='[MODULE]', version='[VERSION]')
    except ImportError:

    Option B: run the debugger agent as a module:

    python \
        -m googleclouddebugger --module=[MODULE] --version=[VERSION] -- \

    Note: This option does not work well with tools such as multiprocessing or gunicorn. These tools spawn workers in separate processes, but the debugger does not get enabled on these worker processes. Please use Option A instead.

    Where, in both cases:

    • [MODULE] is the name of your app. This, along with the version, is used to identify the debug target in the UI.
      Example values: MyApp, Backend, or Frontend.

    • [VERSION] is the app version (for example, the build ID). The UI displays the running version as [MODULE] - [VERSION].
      Example values: v1.0, build_147, or v20170714.

Outside Google Cloud Platform

To use the Python debugger agent on machines not hosted by Google Cloud Platform, you must set up credentials to authenticate with Google Cloud APIs. By default, the debugger agent tries to find the Application Default Credentials on the system. This can either be from your personal account or a dedicated service account.

Personal Account

  1. Set up Application Default Credentials through gcloud.

    gcloud auth application-default login
  2. Follow the rest of the steps in the GCP section.

Service Account

  1. Use the Google Cloud Console Service Accounts page to create a credentials file for an existing or new service account. The service account must have at least the Stackdriver Debugger Agent role.

  2. Once you have the service account credentials JSON file, deploy it alongside the Python debugger agent.

  3. Set the GOOGLE_APPLICATION_CREDENTIALS environment variable.

    export GOOGLE_APPLICATION_CREDENTIALS=/path/to/credentials.json

    Alternatively, you can provide the path to the credentials file directly to the debugger agent.

    Option A:

    # Attach Python Cloud Debugger
      import googleclouddebugger
    except ImportError:

    Option B:

    python \
        -m googleclouddebugger \
        --module=[MODULE] \
        --version=[VERSION] \
        --service_account_json_file=/path/to/credentials.json \
        -- \
  4. Follow the rest of the steps in the GCP section.

Django Web Framework

You can use the Cloud Debugger to debug Django web framework applications.

The best way to enable the Cloud Debugger with Django is to add the following code fragment to your file:

# Attach the Python Cloud debugger (only the main server process).
if os.environ.get('RUN_MAIN') or '--noreload' in sys.argv:
    import googleclouddebugger
    googleclouddebugger.enable(module='[MODULE]', version='[VERSION]')
  except ImportError:

Alternatively, you can pass the --noreload flag when running the Django and use any one of the option A and B listed earlier. Note that using the --noreload flag disables the autoreload feature in Django, which means local changes to files will not be automatically picked up by Django.

Flag Reference

The agent offers various flags to configure its behavior. Flags can be specified as keyword arguments:


or as command line arguments when running the agent as a module:

python -m googleclouddebugger --flag_name=flag_value --

The following flags are available:

module: A name for your app. This, along with the version, is used to identify the debug target in the UI.
Example values: MyApp, Backend, or Frontend.

version: A version for your app. The UI displays the running version as [MODULE] - [VERSION].
If not provided, the UI will display the generated debuggee ID instead.
Example values: v1.0, build_147, or v20170714.

service_account_json_file: Path to JSON credentials of a service account to use for authentication. If not provided, the agent will fall back to Application Default Credentials which are automatically available on machines hosted on GCP, or can be set via gcloud auth application-default login or the GOOGLE_APPLICATION_CREDENTIALS environment variable.

You can’t perform that action at this time.