> From: Ronald Wiplinger <[EMAIL PROTECTED]> > Date: Fri, 22 Nov 2002 22:57:37 +0800 > > I would like to give it a try to install tmda, but: > > 1. I use sendmail on Linux, with rbl list filtering > > 2. I use system wide SpamAssassin, which flags already recognized Spam, > and reduce so my willingness to read the e-mail. > > 3. After SpamAssassin I would like to use tmda, to get the writer into a > white list, BUT, I would get this whitelist useful into the LDAP > database. That way I have two advantages: > a. The LDAP is filled up automatically (system wide) > b. The LDAP can be used than as the addressbook for all users at the > system with different ways of sending mails, like from the desktop, or > from a webmail program, .... > > Problems so far: > 1. How to setup the above wish at all, SpamAssassin followed by tmda > (procmail) > 2. I hear problems with mailing lists and Amazon and E-bay auction mails > 3. I would like to let tmda only flag messages as "unconfirmed sender", > sending back the requestes for confirming, but still let the messages > through, and filter according to this flag the message into an extra > folder with my desktop program > 4. tmda must be so clever that it does not loop, is it so? E.g., if I > ask for a confirmation, than it should not try to make ping pong with > this confirmation request. > 5. confirmed sender should be added into the LDAP database. > > Can it be done, and if how? (So far I have already compiled it without > error - hahahaha).
Here's what I do: I have Spam Assassin running on my firewall under qmail-scanner so all mail has the X-Spam-Status header before it gets to my mail server. In my account on my mail server, I run qmail with TMDA. My MUA is nmh/exmh. I have multiple inboxes. Some of my inboxes run messages through TMDA, some of them filter on the X-Spam-Status header, some do both, and some do neither. Here's what happens for inboxes which use both: 1) I have a generic incoming filter which looks for .whitelist files (as well as a few other useful lists of users who I want to flag differently) which are stored in the nmh directories. 2) If the message gets past the whitelist, I then look at the X-Spam-Status header and if it contains "Yes", I redirect it to an inbox named "ProbableSpam". Other inboxes may skip either step 1 or step 2. Most inboxes only use TMDA. The only times I actually look at X-Spam-Status is for lists which get SPAM or for inboxes which are intended to be public. As of yesterday, I also look at the X-Spam-Status header in my tmda-pending cron job and only show messages which do not contain "Yes". Clearly this is not quite what you have in mind and I'm not using procmail and am using tools that you aren't using, but some portion of what I'm doing may be of use to you. I use LDAP for other things, but not for this. I'm not quite sure what you're hoping for with the global whitelist. For myself, I prefer each inbox to have it's own whitelist. This is because of spam that uses legitimate addresses from the domain that it's sending to. I will want mail from [EMAIL PROTECTED] to get to my cwg-nina mailbox, because that's where she posts to me, but I won't want mail from her to get to my cwg-tmda mailbox because that's not really her. Chris -- Chris Garrigues http://www.DeepEddy.Com/~cwg/ virCIO http://www.virCIO.Com 716 Congress, Suite 200 Austin, TX 78701 +1 512 374 0500 World War III: The Wrong-Doers Vs. the Evil-Doers.
msg00910/pgp00000.pgp
Description: PGP signature
