Heroku上带有Gunicorn的Bottle应用-工头在本地工作，但不在Heroku上工作

Question

I've been stuck trying to get my simple bottle app starting when deployed on heroku.

After quite some searching and tinkering I've got a setup that works locally, but not on heroku.

In /app.py :

import bottle
import beaker.middleware
from bottle import route, redirect, post, run, request, hook, template, static_file, default_app

bottle.debug(True)
app = beaker.middleware.SessionMiddleware(bottle.app(), session_opts)
...
# app routes etc, no run()

Then in /Procfile :

web: gunicorn app:app --bind="0.0.0.0:$PORT" --debug

Correct me if I misunderstand how gunicorn works, I understand the "app:app" portion as look in the module (=file) called app.py and use whatever is in variable "app" as your WSIG, yes?

• I've checked via $ heroku run bash if $PORT is set, seems ok
• The "0.0.0.0" IP I've got from other heroku examples, that should anyway accept any IPs server end, no?
• Python dependencies seem to get installed fine
• I've got this locally running by setting the $POST variable via an .env file for foreman, everything seems working ok on my setup
• Based on this SO question I checked $ heroku ps

=== web (1X): `gunicorn app:app --bind="0.0.0.0:$PORT" --debug` web.1: crashed 2014/12/24 22:43:00 (~ 1m ago)*)

• And $ heroku logs shows:

2014-12-24T20:42:59.235657+00:00 heroku[web.1]: Starting process with command `gunicorn app:app --bind="0.0.0.0:23177" --debug` 2014-12-24T20:43:00.434570+00:00 heroku[web.1]: State changed from starting to up 2014-12-24T20:43:01.813679+00:00 heroku[web.1]: State changed from up to crashed 2014-12-24T20:43:01.803122+00:00 heroku[web.1]: Process exited with status 3

Not sure really how I could get better debugging results either. Somehow the Procfile web process just doesn't seem to work / start, but how can I get info on what's breaking?

Anybody got ideas what's going on here?

PS: I'm rather new to heroku, python, bottle & gunicorn :O

Answer 1

您可以尝试在web: gunicorn app:app中仅使用web: gunicorn app:app吗？

Answer 2

What version of gunicorn do you use? gunicorn 19.1 doesn't write errorlog by default. Try gunicorn --log-file=- .

-R is also useful option to investigate error.