Gunicorn with Flask using wrong Python

Question

I'm trying to bootstrap a Flask app on a Gunicorn server. By putting the two tools' docs together, plus searching around on SO, this is what I have so far... but it's not quite working.

app.py:

from flask import Flask, render_template
from flask.ext.sqlalchemy import SQLAlchemy
from werkzeug.contrib.fixers import ProxyFix

app = Flask(__name__)
app.wsgi_app = ProxyFix(app.wsgi_app)
db = SQLAlchemy(app)

@app.route('/')
def index():
    return render_template('index.html')

what I ran:

From the same directory as app.py,

gunicorn app:app

Even starting this small, I've missed something. The error message is not very helpful:

2013-09-12 20:13:07 [11461] [INFO] Starting gunicorn 0.14.5
2013-09-12 20:13:07 [11461] [INFO] Listening at: http://127.0.0.1:8000 (11461)
2013-09-12 20:13:07 [11461] [INFO] Using worker: sync
2013-09-12 20:13:07 [11528] [INFO] Booting worker with pid: 11528
2013-09-12 20:13:07 [11528] [INFO] Worker exiting (pid: 11528)
2013-09-12 20:13:08 [11461] [INFO] Shutting down: Master
2013-09-12 20:13:08 [11461] [INFO] Reason: Worker failed to boot.

By the way, I'm running this on a Debian Linux system. Many thanks in advance for your help!

Update

After turning on debugging, I got some more instructive error messages. This has become a very specific problem very fast: ImportError: No module named flask. Usually I get this sort of error when I'm not using my virtualenv--but I am. And upon closer inspection, Gunicorn seems to be using a different version of Python than my virtualenv uses, that is Python3. So... my particular python seems not to be getting used. How do I fix this, and tell Gunicorn to use the right Python?

Answer 1

The gunicorn utility may be running out of the system path rather than your virtualenv.

Make sure to pip install gunicorn into the virtualenv.

Here's the pip freeze of a virtualenv I setup to run your app:

(so_2)20:38:25 ~/code/tmp/flask_so$ pip freeze
Flask==0.10.1
Flask-SQLAlchemy==1.0
Jinja2==2.7.1
MarkupSafe==0.18
SQLAlchemy==0.8.2
Werkzeug==0.9.4
gunicorn==18.0
itsdangerous==0.23
wsgiref==0.1.2

In reality, I only ran these pip installs:

$ pip install flask
$ pip install gunicorn
$ pip install Flask-SQLAlchemy

Answer 2

I have the same problem as You. The problem is that gunicorn for some reason load the enviroment outside your virtual env. I solved by uninstalling the package gunicorn outside virtual enviroment;

(env) $ deactivate
$ sudo pip uninstall gunicorn

So you come back to your env and try to run. In my case env folder I typed:

$ source env/bin/activate
(env) $ pip install gunicorn
(env) $ gunicorn server:app
2013-10-19 20:40:56 [11923] [INFO] Starting gunicorn 18.0
2013-10-19 20:40:56 [11923] [INFO] Listening at: http://127.0.0.1:8000 (11923)
2013-10-19 20:40:56 [11923] [INFO] Using worker: sync
2013-10-19 20:40:56 [11926] [INFO] Booting worker with pid: 11926

Answer 3

Gunicorn may be installed at multiple location in your system. It may be present in

1. OS default Python Path
2. Anaconda Python Path

By default when you specify

gunicorn -w 4 -b 127.0.0.1:5000 flaskApp:app

You are referrng to operating system's default Python where in the same path flask package is not installed results in error. Better specify which gunicorn you are reffering to by providing proper path to gunicorn

/home/sunil/anaconda2/bin/gunicorn -w 4 -b 127.0.0.1:5000 flaskApp:app

Answer 4

Assuming your virtual environment is called env and your app is called app and gunicorn was install properly.You may try:

sudo env/bin/gunicorn --bind 0.0.0.0:5432 wsgi:app

This force the app to use the gunicron in your virtual environment.