Gunicorn RecursionError with gevent and requests in python 3.6.2 #1559
Comments
|
Thank you for the report. Unfortunately, we need to see the full traceback and a minimal reproducer in order to track down the problem. |
|
|
Is this enough? This happens when gunicorn is running, and I use another python script to make a request to the server. My command of running gunicorn is |
|
That looks like gevent/gevent#941, ie, incorrect monkey-patching order. |
|
Is there a way to change the monkey patch order in the gunicorn setup? Right now in my code I've never type any "monkey-patch-all", I just use |
|
Speaking only for myself, I have found the order in which gunicorn preload and monkey-patches to be incompatible. Instead, we use a wrapper script; in principle it's something like: import gevent.monkey; gevent.monkey.patch_all()
import gunicorn
gunicorn.main() |
|
It seems like I develop in a very different (and probably worse) way than how you did it. In my code, my code looks like a standard flask app, and in my code there's no |
|
The way you develop has nothing to do with this. This is simply about deployment. I deploy with (roughly) the script I shared because |
|
I see, would you mind suggest me a deployment script that I can use to run my code that can fix this monkey patch order issue? Originally I thought the deployment script would be a bash script that's similar to the line I share above but it seems like you are also using python for this, then I don't know how can you feed in those parameters (or even simulate the New relic line...?). I just have a gqa.py in the gqa folder that I want to use gunicorn to run it. |
|
It's both trivial and depends on your app. Our app uses Pyramid, so we simply have a setuptools # Note that we must not import *anything* before the patch
from nti.monkey import patch_nti_pserve_on_import
patch_nti_pserve_on_import.patch() # gevent.monkey.patch_all()
import sys
from pkg_resources import load_entry_point, get_distribution
def main():
sys.exit(
load_entry_point('pyramid', 'console_scripts', 'pserve')()
)
if __name__ == '__main__':
sys.exit(main()) |
|
I should add, we absolutely depend on |
|
Thanks, your suggestions are extremely helpful (Although I am still not sure how to use my own deployment script as I have no experience on that). But will probably looking into |
|
Thanks for the details. I think this is more or less a duplicate of #1056. |
|
We can close this one now. |
I have a simple flask script that uses requests to make http requests to third party web service. The way I run the script in gunicorn is
gunicorn abc:APP -b 0.0.0.0:8080 -w 4 -k gevent --timeout 30 --preloadHowever, after I upgrade the code to python 3.6.2, I can still run the server, but whenever the webserver received a request, it shows
RecursionError: maximum recursion depth exceeded while calling a Python objecton every worker, and the server seems are still running. When I change the running command to
gunicorn abc:APP -b 0.0.0.0:8080 -w 4 --timeout 30 --preloadIt all works again. So is there any issue with gunicorn's async worker and requests in python 3.6.2? Is there a way to fix this?
The text was updated successfully, but these errors were encountered: