gazimbeisaac / kent

Fake Sentry server for local development, debugging, and integration testing

Kent

Kent is a service for debugging and integration testing Sentry.

Code:	https://github.com/willkg/kent/
Issues:	https://github.com/willkg/kent/issues
License:	MPL v2

Goals

Goals of Kent:

1. make it possible to debug before_send and before_breadcrumb sanitization code when using sentry-sdk
2. make it possible to debug other sentry error submission payload issues
3. make it possible to write integration tests against a fake sentry instance

Quick start

Installing and running on your local machine

1. Install Kent.

   You can install Kent from PyPI with pipx or pip or whatever:

   pipx install kent
   

   You can install a REVISH ("main", branch name, commit, whatever) from GitHub:

   pipx install https://github.com/willkg/kent/archive/refs/heads/<REVISH>.zip
   

   You can install from a checked out version of this repository:

   pipx install .
   

2. Run Kent:

   kent-server run [-h HOST] [-p PORT]
   

Running in a Docker container

I'm using something like this:

FROM python:3.10.1-alpine3.15

WORKDIR /app/

ENV PYTHONUNBUFFERED=1 \
    PYTHONDONTWRITEBYTECODE=1

RUN pip install -U 'pip>=8' && \
    pip install --no-cache-dir 'kent==VERSION'

USER guest

ENTRYPOINT ["/usr/local/bin/kent-server"]
CMD ["run"]

Replace VERSION with the version of Kent you want to use. See https://pypi.org/project/kent for releases.

Then:

$ docker build -t faksentry:latest .
$ docker run --rm --publish 8000:8000 fakesentry:latest run --host 0.0.0.0 --port 8000

Things to know about Kent

Kent is the fakest of fake Sentry servers. You can set up a Sentry DSN to point to Kent and have your application send errors to it.

Kent is a refined fake Sentry service and doesn't like fast food.

Kent will keep track of the last 100 payloads it received in memory. Nothing is persisted to disk.

You can access the list of errors and error data with your web browser by going to Kent's index page.

You can also access it with the API. This is most useful for integration tests that want to assert things about errors.

GET /api/errorlist/

List of all errors in memory with a unique error id.

GET /api/error/ERRORID

Retrieve the payload for a specific error by id.

POST /api/flush/

Flushes the error manager of all errors.

You can use multiple project ids. Kent will keep the errors separate.

If you run kent-server run with the defaults, your DSN is:

http://public@localhost:5000/1    for project id 1
http://public@localhost:5000/2    for project id 2
...

Kent definitely works with:

• Python sentry-sdk client
• Python raven client (deprecated)

I don't know about anything else. If you use Kent with another Sentry client, add an issue with details or a pull request to update the README.

Development

Install

1. mkvirtualenv kent
2. pip install -e '.[dev]'

Release process

1. Check out main tip.

2. Update __version__ in src/kent/__init__.py.

3. Run:

   $ make checkrot
   $ make test
   $ make testdocker
   $ check-manifest
   

   Fix any issues from that.

4. Update HISTORY.rst.

5. Push any changes to GitHub.

6. Release:

   $ git tag --sign 1.0.0
   # Use contents from HISTORY.rst for tag comment
   
   $ make clean
   $ python setup.py sdist bdist_wheel
   $ twine upload dist/*