On Sat, 19 Jul 2003, Mike Burger wrote: > On Sat, 19 Jul 2003, Aaron Konstam wrote: > > > On Sat, Jul 19, 2003 at 10:23:20AM +0100, Bryan Hepworth wrote: > > > Aaron > > > > > > I met the mailscanner guy at the recent linux expo in Birmingham. I'm also > > > looking at setting it up on a couple of servers running Shrike. If you could > > > give me any pointers or pitfalls I'd be really grateful. > > > > > > Thanks > > > > > > Bryan > > The basic use is very simple. You run mailscanner instead of > > sendmail and mailscanner calls sendmail to identify the spam. > > There are instructions. You train the Basyian part of sapmassassin > > which must be at least version 2.5 (the shrike version won't work) > > by running a program sa-learn and puts the database in a user > > directory called .spamassassin. To be truthful there are things > > about it I have not figured out yet but it does identify spam. > > I'm guessing that that should have read "You run mailscanner instead of > spamassassin and mailscanner calls spamassassin to identify the spam." > > Amavis is supposed to also support SA, but I have never been able to > easily get Amavis to call SA. Use amavis-new. It is trivial to get SA working with postfix. From what I have been told getting it working with sendmail requires milter or something. Since I do not use sendmail I am sure someone will correct me if I am wrong. > The main problem, of course, with this approach is that by running it at > the MTA level (which is what happens when Sendmail/Postfix, itself, has to > pass the mail to the scanner), you negate the ability for users to have > their own whitelists, blacklists, or other preferences (such as not > munging the subject line, not killing HTML in non-spam messages, and > placing the report into the headers). User level options are only > available if SA (either by spamc or the spamassassin executable) is run > under the recipient's UID. I have read on another list that it is possible to do the per user stuff it just needs to be setup differently. As I understand things you need some sort of db to hold the user preferences in. > The mail administrator winds up having to constantly futz with the > local.cf file, adding/removing entries, changing config options, etc. As I understand things the users can adjust their own prefs. > Calling it through the MTA is best used, IMO, if you're running a front > end mail relay (mail comes in, is scanned, and then forwarded to an > internal system for actual delivery). You also have the ability to have SA simply add a SPAM header and let the users filter on their own. Obviously there are some trade offs doing it this way but it is still better than nothing. HTH, -- ......Tom Registered Linux User #14522 http://counter.li.org tdiehl@xxxxxxxxxxxx My current SpamTrap -------> mtd123@xxxxxxxxxxxx -- Shrike-list mailing list Shrike-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/shrike-list
