aiosmtpd: RuntimeError after upgrading from 1.2.2

After upgrading from version 1.2.2 to 1.2.4 we receive the following error when trying to start the controller:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python3.6/site-packages/aiosmtpd/controller.py", line 165, in start
    raise RuntimeError("Unknown Error, failed to init SMTP server")
RuntimeError: Unknown Error, failed to init SMTP server

This error can even be triggered by running the sample code from the documentation:

import asyncio
class ExampleHandler:
  async def handle_RCPT(self, server, session, envelope, address, rcpt_options):
    return '250 OK'

from aiosmtpd.controller import Controller
controller = Controller(ExampleHandler())
controller.start()

After executing “controller.start()” the RuntimeError from above appears.

Python version is 3.6.8 OS is Red Hat 7.9 aiosmtpd was installed with pip3. After downgrading via “sudo pip3 install aiosmtpd=1.2.2” everything works again and aiosmtpd receives and processes emails without problems.

About this issue

Original URL
State: closed
Created 3 years ago
Comments: 32

Commits related to this issue

Test newly-implemented TimeoutError Inspired by #244 — committed to pepoluan/aiosmtpd by pepoluan 3 years ago
Wait until factory _is_ invoked Also inspired by #244 We add another threading.Event to wait until _factory_invoker() completes. Plus a test case to test this timeout. — committed to pepoluan/aiosmtpd by pepoluan 3 years ago
Pin aiosmtpd to latest working version Version higher than 1.2.2 cause the server start to fail. See https://github.com/aio-libs/aiosmtpd/issues/244 for more info. — committed to LeoVerto/email2signal by LeoVerto 3 years ago
Fix Controller Bug on hostname="" or IPv6 not available (PR #248, Issue #244) * Implement _factory_invoked Event * Suppress only socket.timeout exception Other exceptions should bubble up, no lon... — committed to aio-libs/aiosmtpd by pepoluan 3 years ago

Most upvoted comments

Hi, sure. I was thinking it can be similar one 😃 New one opened.

msztolcman on Mar 5, 2021

This has fixed the issue for me, thank you so much for the quick debugging and fix, @pepoluan!

LeoVerto on Feb 18, 2021

Hello @pepoluan thank you very much! Looks very good, except that on my system the error 97 is thrown:

>>> from aiosmtpd.controller import Controller
>>> cont = Controller(ExampleHandler())
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python3.6/site-packages/aiosmtpd/controller.py", line 107, in __init__
    self.hostname = get_localhost() if hostname is None else hostname
  File "/usr/local/lib/python3.6/site-packages/aiosmtpd/controller.py", line 41, in get_localhost
    with makesock(AF_INET6, SOCK_STREAM) as sock:
  File "/usr/lib64/python3.6/socket.py", line 144, in __init__
    _socket.socket.__init__(self, family, type, proto, fileno)
OSError: [Errno 97] Address family not supported by protocol

Errno 97 translates to errno.EAFNOSUPPORT, however get_localhost() is catching errno.EADDRNOTAVAIL. If I replace errno.EADDRNOTAVAIL with errno.EAFNOSUPPORT, version 1.3.1 works!

gnugnug on Feb 18, 2021

v1.3.1 is marinating, to be released tomorrow.

Definitely will fix @LeoVerto 's issue.

Not sure if it will fix @gnugnug 's issue, but at least 1.3.1 has more control points that can give more visibility into what’s going on.

pepoluan on Feb 17, 2021

Hey, someone else here!

I’ve been getting this issue with Python 3.9 in both alpine-based docker images and on my dev machine running Arch Linux, in both cases using a fresh virtualenv created by pipenv.

To reproduce this, clone my repository at this commit and replace line 90 in app.py with config = {} to avoid having to set all the config environment variables. Next run pipenv install and finally pipenv run python app.py to start the application.

LeoVerto on Feb 16, 2021