Log SMTP disconnect as warning instead of error

[Paul Everitt]

This is tripping Nagios alarms several times a week. Can we lower from warning to error?

Traceback (most recent call last):
  File "/srv/osfkarl/production/81/eggs/repoze.sendmail-4.2+agendaless.4-py2.7.egg/repoze/sendmail/queue.py", line 232, in _send_message
    self.mailer.send(fromaddr, toaddrs, message)
  File "/srv/osfkarl/production/81/eggs/repoze.sendmail-4.2+agendaless.4-py2.7.egg/repoze/sendmail/mailer.py", line 100, in send
    connection.sendmail(fromaddr, toaddrs, message)
  File "/usr/lib/python2.7/smtplib.py", line 748, in sendmail
    (code, resp) = self.data(msg)
  File "/usr/lib/python2.7/smtplib.py", line 511, in data
    (code, msg) = self.getreply()
  File "/usr/lib/python2.7/smtplib.py", line 365, in getreply
    + str(e))
SMTPServerDisconnected: Connection unexpectedly closed: ('The read operation timed out',)

[Carlos de la Guardia (cguardia)]

Where is this called? Is this the complete traceback? I can catch the exception inside repoze.sendmail, which should stop the errors, but there's no logging or anything set up in there, so not sure how I would log the warning.

[Paul Everitt (paul-agendaless)]

Yep, that's the entire traceback. I believe we've done this in the past, where we suppressed an unhandled repoze.sendmail traceback, but I don't remember how. Maybe we overrode the method in our implementation's subclass.

[Carlos de la Guardia (cguardia)]

Fixed, but turns out I don't have access to the repository. Sent a pull request:

https://github.com/karlproject/repoze.sendmail/pull/2

[Paul Everitt (paul-agendaless)]

Hmm, today I got four exceptions that looked like this:

Error while sending mail from to <email address hidden>.

Traceback (most recent call last):
  File "/srv/osfkarl/production/87/eggs/repoze.sendmail-4.2+agendaless.5-py2.7.egg/repoze/sendmail/queue.py", line 232, in _send_message
    self.mailer.send(fromaddr, toaddrs, message)
  File "/srv/osfkarl/production/87/eggs/repoze.sendmail-4.2+agendaless.5-py2.7.egg/repoze/sendmail/mailer.py", line 108, in send
    connection.quit()
  File "/usr/lib/python2.7/smtplib.py", line 772, in quit
    res = self.docmd("quit")
  File "/usr/lib/python2.7/smtplib.py", line 393, in docmd
    self.putcmd(cmd, args)
  File "/usr/lib/python2.7/smtplib.py", line 341, in putcmd
    self.send(str)
  File "/usr/lib/python2.7/smtplib.py", line 333, in send
    raise SMTPServerDisconnected('please run connect() first')
SMTPServerDisconnected: please run connect() first

> On Nov 30, 2016, at 10:01 PM, Carlos de la Guardia <email address hidden> wrote:
>
> Fixed, but turns out I don't have access to the repository. Sent a pull
> request:
>
> https://github.com/karlproject/repoze.sendmail/pull/2
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1613259
>
> Title:
> Log SMTP disconnect as warning instead of error
>
> Status in KARL4:
> In Progress
>
> Bug description:
> This is tripping Nagios alarms several times a week. Can we lower from
> warning to error?
>
> Traceback (most recent call last):
> File "/srv/osfkarl/production/81/eggs/repoze.sendmail-4.2+agendaless.4-py2.7.egg/repoze/sendmail/queue.py", line 232, in _send_message
> self.mailer.send(fromaddr, toaddrs, message)
> File "/srv/osfkarl/production/81/eggs/repoze.sendmail-4.2+agendaless.4-py2.7.egg/repoze/sendmail/mailer.py", line 100, in send
> connection.sendmail(fromaddr, toaddrs, message)
> File "/usr/lib/python2.7/smtplib.py", line 748, in sendmail
> (code, resp) = self.data(msg)
> File "/usr/lib/python2.7/smtplib.py", line 511, in data
> (code, msg) = self.getreply()
> File "/usr/lib/python2.7/smtplib.py", line 365, in getreply
> + str(e))
> SMTPServerDisconnected: Connection unexpectedly closed: ('The read operation timed out',)
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/karl4/+bug/1613259/+subscriptions

[Carlos de la Guardia (cguardia)]

I'll take a look Saturday morning.

[Carlos de la Guardia (cguardia)]

I made another PR to fix this.