Giorgio Salluzzo

If I am not mistaken, the problem seems quite clear to me: apart from access logs, everything else is logged through `self.error_log` (`gunicorn.error`) which outputs to `stderr`. @rokcarl approach makes...

My PR uses the same approach used by @rokcarl. Not sure why they did not consider it. https://github.com/benoitc/gunicorn/pull/2752

I was trying to tackle this issue, since it's happening to us every time Gunicorn terminates an `uvicorn` worker which reached the `max_requests` limit. At some point of my research,...

A second approach would be logging it as `DEBUG` instead of `INFO`. Let me know if you like any of the above, I could even submit a PR for it.

> It sounds kind of like some people think this is a benign error that shouldn't be logged, but in our case it appears to be killing the worker -...

The problem is exactly when gunicorn logs to stderr INFO logs, this patch fixes this behaviour.

Hi there, I am the author of `mocket`, which supports `gevent` and has a compatibility layer for using it the same way you use `HTTPretty`. You could give it a...

Mocket supports both `gevent` and `asyncio`/`aiohttp`. https://pypi.python.org/pypi/mocket/2.0.0

Not using `PEM` format worked for me. So, basically: ``` ssh-keygen -t rsa -b 4096 -C [email protected]:foo/bar.git ```

Also, in the metadata the package is called `anaconda-client`. ``` WARNING: Generating metadata for package anaconda-cli produced metadata for project name anaconda-client. Fix your #egg=anaconda-cli fragments. ``` So, maybe the...