On Mon, 2009-08-03 at 22:46 -1000, alexandre wrote:
> I want to setup exim for virtual domains hosting using ldap as backend.
> In addition I want antivirus and antispam checks.
> >From what i can understand:
> As antivirus policy is quite simple (remove viruses for every users),
> antivirus at SMTP (using exim ACL) suits well.
> But for antispam things are differents.
> I want to use spamassassin (not very original...) and its user
> preferences stored in the ldap directory (bayesian db and autowhitelist
> in sql as well). If I am not mistaking... using spamc in ACL at smtp
> time prevents system from using variable expansion such as
> ${local_part} or $domain... and so I cannot use user preferences.
> Then I tried using spamc as a transport in exim:
> It seems to work almost properly but I don't know how to deal with
> spam-tagged message (I would like them to be delivered in a Junk Folder
> but would like to avoid creating .forward files for my each of my users)
> Does anybody knows how to achieve this goals?
If you want to do per user spam scanning there are a few tricks to get
it to work properly. First step is configuring SA to use per user config
(or per profile using a fake user) but that's entirely an SA problem.
I was going to point you out to the FAQ or config examples for doing
this but even when only looking at site:exim.org all I get from google
are the "old" way of doing it via transports. Ironically, the good
example of how to do it is from exim 4.50 with exiscan-acl, but who on
earth is going to google for that if they're new to exim.
Section 6. Having multiple content scanning profiles for several
users or domains.
http://duncanthrax.net/exiscan-acl/exiscan-acl-examples.txt
exiscan-acl is now part of exim so don't go try to patch to add it in.
You may have to tweak things a little .. you don't need to use acl_mX (X
= 0-9) any more since acl variables have been expanded to use any text
eg. acl_mPerRecipientProfile but that's entirely up to you.
--
The Exim manual -
http://docs.exim.org
