I got one problem using this setup.. the whitelist/blacklist nor bayesian works, because spamd doesnt look the homedir of the user in the ldap.. just preferences..

With this problem I rewrote one module used to auth in courier to work directly in ldap
http://www.ebrain.com.br/~cassiano/AuthLdap.pm
I save it on spamssassin modules dir (under perl modules) in my debian box the dir is
    /usr/share/perl5/Mail/SpamAssassin/
and after the #!/usr/bin/perl in the begin of spamd i put this line
use Mail::SpamAssassin::AuthLdap;

this module rewrites the getpwuid and getpwnam, with this spamd thinks the ldap user is one real user.. let me show you one spamassassin log
Feb 27 07:41:08 edge spamd[1504]: connection from localhost [127.0.0.1] at port 61376
Feb 27 07:41:09 edge spamd[1504]: info: setuid to [EMAIL PROTECTED] succeeded
Feb 27 07:41:09 edge spamd[1504]: processing message <20050227104106.27865.qmail at mail-1.ebrain.com.br> for caquino at ebrain.com.br:1033.
Feb 27 07:41:09 edge spamd[1515]: clean message (-4.5/4.0) for caquino at ebrain.com.br:1033 in 4.0 seconds, 975 bytes.
Feb 27 07:41:09 edge spamd[1515]: result: . -4 - ALL_TRUSTED,AWL,BAYES_00,NO_REAL_NAME scantime=4.0,size=975,mid=<20050227104056.27746.qmail at mail-1.ebrain.co
m.br>,bayes=1.66533453693773e-16,autolearn=ham

I think the most interesting line is this
Feb 27 07:41:09 edge spamd[1504]: info: setuid to [EMAIL PROTECTED] succeeded
where the spamassassin thinks [EMAIL PROTECTED] is one real user.
now I can create one .spamassassin in the virtual home dir with al the configuration and the spam dbs.

I think the AuthLdap.pm needs some adjustments to fit your needs.. but this is the one I use here.
in the maildrop user config I put

if ( $SIZE < 262144 )
{
        exception {
                xfilter "spamc -f -u [EMAIL PROTECTED]"
        }
}

if (/^X-Spam-Flag: *YES/)
{
        exception {
                to "Maildir/.SPAM.E Spam"
        }
}

I don´t have tested with simscan but I think simscan can pass the -u to spamc.. and the user can haver your spam settings valid before the mail gets the queue.


Ace Suares wrote:
On Saturday 26 February 2005 6:21, you wrote:
  
Matt wrote:
    
This isn't exactly a qmail-ldap question... but I'm sure it's come
up.....   how do you handle user accounts settings in spamassassin
(like if you want to let them have individual settings) when you have
virtual users as in qmail-ldap?  Does it even work?
      
I use simscan to filter virus with freshclam and spam with
spamassassin. All you have to do is define "spamassassin" attribute in
ldap and you can store  there spamassassin user preferences.
    

Wouldn't mind if you gave a link. Google returns some weeeeeeird 
results ;-)

Ace

  
In local.cf you put something like this:

user_scores_dsn
ldap://192.168.X.Y:389/ou=users,dc=example,dc=org?spamassassin?sub?uid=
__USERNAME__ user_scores_ldap_username cn=Manager,dc=example,dc=org
user_scores_ldap_password  secret


And thats all.

Omar
    

  

Reply via email to