Ok, this is how it works.  A message arrives to the server.

1. The message is sent immediately to a procmail script that checks to see
if the to: or cc: address contain [EMAIL PROTECTED]  This blocks any
messages that may have been bcc:'ed to the list.  By doing this, we prevent
about 10 spam messages a day from being sending on the list.

2. Next the message is run through a script that checks the senders address.
What we are checking here is if this particular sender sent over 15
messages/day to the list.  The assumption is over 15 messages/day, there
might be a mail-loop and we should kill the messages.  The script then
checks the total number of messages on the list.  I believe I have this list
set to 250 messages a day.  Over 250 messages/day will cause the "circuit
breakers" to fire preventing any additional messages on the list.  Again
this is to  prevent mail-loops or mail-bombs from filling our e-mail
accounts.  It has fired a few times.  Last time, some server in Texas was
sending messages back to the list that were over 6 months old.  So once the
list hit 250, the list will shutdown.

3. Next the message is sent through a program that converts any HTML e-mail
to plain-text.  You should always be sending plain-text through the list,
but sometimes people forget.  An HTML message usually takes twice the
bandwidth as a plain-text message.  Since the server currently uses 6-8
GB/day we want to limit messages to plain-text only.  If you send a link
using HTML, the link may become corrupt as it passes through the filter.  By
sending plain-text, you can be assured you message will go through
unchanged.

4. Next the message is sent to a program that posts the message on the
web-board.  I would encourage people to read the web-board instead of
subscribing to the mailing list, as it uses less bandwidth since you are not
receiving every message.  Before it gets posted on the web-board, a series
of checks are made.  First, any message larger then 40k characters is
rejected.  This is usually not a problem as the message has already been
converted to plain-text.  Then a series of keywords are checked (see George
Carlin).  Besides profanity, known advertisers are also checked.  So for
example, if you have a great new product, it may seem spamming the list with
advertisements is a great form of free advertising.  But this will place
your product on the profanity list, preventing even legitimate discussion of
the product.  So in the long run, it hurts sales and discourages spamming of
the list.

5. If the message makes it through the web posting, it is re-written as an
e-mail and sent to the mailing list (customized version of majordomo).   It
use to take 7 hours for the message to be fully delivered.  Now the
mailing-list is sorted by domain and broken into groupings of 50 domains
each.  Each grouping  runs in parallel delivering the message in about an
hour.  So depending upon where you fall in the list, delivery might be
faster or slower, but you should receive the message in about an hour.

6. At the same time a copy of the messages is converted to news format for
posting on the newsfeed.  The message usually is posted on the newsfeed at
about the time the first e-mail is being sent.  So reading the newsfeed is a
more timely option then reading the list.

7.  While the message is being sent through the mailing list, and newsfeed,
a copy is saved in the archives queue.  Once a week, a program runs that
posts the messages to the archives.

Most of the software is customized/home grown for this website, written in
Perl, PHP, a little C with the database backend being MySQL.  Remember I am
not a programmer and this was all self-taught.  My guess a "real" programmer
would cringe at some of my code, but hey it works.

The entire thing is being run on a customized version of RedHat Linux.

What was the question again? Oh' yea.  The message in step 5 is re-written
as a to:.  That is why every message from the list begins with to:
[EMAIL PROTECTED]  Messages that are cc:'ed to the list end up being
converted.  I can not imagine this being a problem.  Is it?

Paul Borghese




Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=8155&t=8155
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]

Reply via email to