> Date: Thu, 18 Jul 2013 16:30:51 -0400 - timestamp of your mail
> Thu, 18 Jul 2013 13:30:51 -0700 (PDT) - you posted it via HTTP
> Thu, 18 Jul 2013 22:31:39 +0200 (CEST) - it reached python.org
> Fri, 19 Jul 2013 09:56:14 +0200 (CEST) - it began delivery to my address
> Fri, 19 Jul 2013 00:56:15 -0700 (PDT) - it reached gmail and me.
>
> So the big delay is somewhere within python.org. Since this does NOT
> happen to everyone, I would be looking for something like:
> * Moderation requirements (does your message not comply with spec?)
> * Spam greylisting
> * Slow/glitchy DNS when attempting to verify your message
>
> Most likely I think is moderation, given the extreme length of time.
> Something would have to be pretty glitchy to lag your mail out for
> half a day in spam filtering. It's possible, I guess, though. I've no
> idea what anti-spam features the list is using.
Python.org runs Mailman 2.1.something. In addition to all the various
front-end filters running in the Postfix SMTP daemon (which trap most
spam), SpamBayes is used as a last check to score messages intended
for delivery to most mailing lists hosted on mail.python.org. It
leaves two headers behind as proof of its existence. Here are those
headers for Matthew's message that started this thread:
X-Spam-Status: OK 0.017
X-Spam-Evidence: '*H*': 0.97; '*S*': 0.00; 'subject:Python': 0.06;
'problem?': 0.07; 'subject:posting': 0.09; 'python': 0.11;
'posted': 0.15; 'effect.': 0.16; 'missing?': 0.16; 'python-dev,':
0.16; 'server,': 0.19; 'all:': 0.24; "i've": 0.25; 'message-
id:@mail.gmail.com': 0.30; 'obscure': 0.31; 'lists': 0.32;
'trouble': 0.34; 'anybody': 0.35; 'but': 0.35;
'received:google.com': 0.35; 'there': 0.35; 'subject:List': 0.36;
'list.': 0.37; 'gmail': 0.38; 'to:addr:python-list': 0.38;
'issue': 0.38; 'moving': 0.39; 'to:addr:python.org': 0.39;
'mailing': 0.39; 'first': 0.61; 'hours': 0.66
so SpamBayes thought it looked fine. Had it been marked as "UNSURE"
or "SPAM" in the status header, Mailman would have held it for
moderator attention. It is possible it was held by Mailman for some
other reason. Some obvious reasons: sender isn't subscribed, message
had too many recipients, python-list@python.org was BCC'd (implicit
destination). If you message was held by Mailman, moderator overload
is the most likely cause of the delay. It's also possible Mailman got
overwhelmed, but I've seen no postmaster messages suggesting there
were any systemic problems.
BTW, although SpamBayes serves as a last resort for mail arriving via
SMTP, it is the only line of defense for mail gatewayed from Usenet.
I'm sure I can dredge up the code I wrote for that if anyone wants it
for another application.
Skip
--
http://mail.python.org/mailman/listinfo/python-list
