Larus: A WSGI HTTP Server [WIP]

Introduction

Larus is a simplified Gunicorn clone

Install

From PyPI:

$ pip install larus

From source:

$ python setup.py install

Usage

Usage: larus [OPTIONS] MODULE_APP

Options:
  -w, --workers INTEGER  number of workers, default: 1
  -b, --bind TEXT        address to bind: host:port, default: 127.0.0.1:8000
  -c, --chdir TEXT       change the working director
  --help                 Show this message and exit.

For example, if myapp.py contains the wsgi app app, we can serve it with larus with:

larus myapp:app

with 4 workers and bind to 0.0.0.0:5000:

larus -w 4 -b 0.0.0.0:5000 myapp:app

Signals

You can use signals to dynamically increment or decrement the number of workers while serving:

• TTIN: Increment the number of workers by one
• TTOU: Decrement the number of workers by one

Performance

In my simple benchmark using ab, Larus performs better than Gunicorn both running 4 sync workers. Perhaps benefits from the C http parser Larus using.

TODO: add detailed comparison.

TODO

• Add more worker types