SUMMARY (partial) : Spamassassin

From: Daniel Clar (Daniel.Clar@supelec.fr)
Date: Fri Mar 28 2003 - 13:19:22 EST


The original posting was :

>
> Hello Managers,
>
> Is someone currently using spamassassin on Tru64 Unix ?
>
> How is it interfaced with sendmail on the system ?
>
> Thank you and regards,
>
> Daniel
>
> Daniel Clar
> Computer Operations Manager
> Supelec
> France
>

---------------------------
Thank you to :

Bryan Dunlap [bcd@pacific.mps.ohio-state.edu]

I'm running spamassassin on 4.0E, but it's not hooked into sendmail.
Users run it voluntarily via procmail. In my opinion, it works well.

David Warren [warren@atmos.washington.edu]

We are now using it on linux, but we did have it running on suns and decs. We
have writen our own milter.

Didier Godefroy [ldg@ulysium.net]

I'm about to install this myself, I was just reading about it a while ago,
it's a perl script that's called from within a .procmailrc file mostly, and
there's a daemonized version that can be used also from .procmailrc through
a pipe, but it seems not to matter if the mta is sendmail or something else,
it seems to do all its work at the local delivery and not at the mta level.
But perhaps I may be wrong..

Tom Webster [webster@ssdpdc.lgb.cal.boeing.com]

Not currently, but based on the volume of SPAM I've been getting,
I've been thinking about it.

James Sainsbury [J.Sainsbury@chem.usyd.edu.au]

I use spamassassin but probably not in the way you would want to.
(I have a lot of site policy in a modified smap/smapd (TIS fwtk)
that existed before spam filtering.)
ie

incomming mail--> smtp process (smap)
                   \
                    spool --> scanning process
                              / \
                         jail (spam) output spool --> smapd --> sendmail

The scanning process contains spammassassin but determines the spam value
based on the exit code
eg

$PREPROCESS smapfile | $SPAMASS -S -L -P -e > /dev/null
if [ $? -ge 5 ] ; then
        is SPAM
fi
---------

Generally spamassassin is invoke by a delivery agent such as procmail.
Eg with sendmail/procmail under openbsd

http://davespicks.com/writing/programming/spamassassinopenbsd.html

Rustam T. Usmanov [rustam@ruslan.ru]

We're successfully using spamassassin on DigitalUNIX 4.0D as part of
mailscanner configuration
<URL:http://www.sng.ecs.soton.ac.uk/mailscanner/>.

It is interfaced with sendmail via the second message queue -
<URL:http://www.sng.ecs.soton.ac.uk/mailscanner/install/sendmail.shtml>

Nikola Milutinovic [Nikola.Milutinovic@ev.co.yu]

Will be when I emerge under this Berkeley DB 4.1.25 pile, that's screwing up my
Sendmail run...

> How is it interfaced with sendmail on the system ?

As of Sendmail 8.12.x the prefered way of interfacing ANY filtering
approve/disapprove software (anti-virus, SPAM,...) is via Milter, which has to
be enabled during build time. Anything else is making a cludge of the config.

I'm currently using AMaViS+Sophos as AV scanner.

Horst Dieter Lenk [lenk@mpi-muelheim.mpg.de]

we use spamassassin on Tru64 together with an Amavis script
(+ Sophos) for virus scan.
The sendmail has a mailer which starts a command procedure
to scan and deliver the mails.

/etc/mail/aliases: user@savspam
/etc/mail/sendmail.cf: # SAV +SPAM mailer is a pseudo-domain
                          CPSAVSPAM
                          R$+ < @ SAVSPAM. > $#savspam $: $1 user@SAVSPAM
                          Msavspam,P=/usr/sbin/scanmails+spam, F=lsDFMqPenu, S=EnvFromL,
                          R=EnvToL/HdrToL, T=DNS/RFC822/X-Unix,
                          A=savspam $u
/usr/sbin/scanmails+spam: cat ${tmpdir}/receivedmail | ${formail} -f -A \
                          "${SA_Header_String}" | ${spamc} -u $1 | ${deliver} -d $1

For spamassassin I use the the spamc program with the spamd daemon,
because its much faster than statring the perl scipt each time.
If you need more details, I can send you all the configuration files.

stephane BRANCHOUX [stephane.branchoux@univ-perp.fr]

In French.....

Marc Thoelen [marc.thoelen@luc.ac.be] and Andres Henckens [andres.henckens@luc.ac.be]

How to
incorperate procmail into your sendmail.cf is explained in the procmail manual
(it differs from sendmail to sendmail). That should be the first thing to do
and to check if all is OK.

You can download and compile spamassassin from www.spamasssissin.org
On most unix systems it won't compile, you need to change the LIBS= line after
doing a configure to LIBS= -ldb -lm

.procmailrc is used on an account basis and is great for testing, for a
systemwide procmailrc you can create one as /etc/procmailrc and it is
systemwide. What to put into it is also explained in the spamassassin manual.

You are right, with local users it works, when e-mail gets forwarded to
another account it doesn't. If you use a .procmailrc in the users homedir it
works. The structure is the same as in /etc/procmailrc but you have to add a
line to forward the e-mail to the other e-mail box after the spamscan.

I have something like this added.

:0
! user@otherserver.domain

that's it. A .procmailrc replaces a .forward in the homedir. If you
use .forwards, .procmailrc will not be used, so you have remove the .forward.

Xavier Mertens [xavier@rootshell.be]

SpamAssassin as default setup does not interact with sendmail but with
procmail! Anyway there exists not a Milter filter to use spamd directly fro
sendmail.

Tom Smith [smith@cag.lkg.dec.com]

Who helped me a lot to put in place a working configuration with the sendmail.cf modifications.

---------------------------------------------------

So it works if the users are local users. The only problem is that in my configuration the Tru64 box is just the
incoming smtp server who delivers mail, through the aliase mechanism the the smtp mailer, to users mailboxes on an
OpenVMS system.

So even with procmail replacing /bin/mail, it's not used in this configuration.

I've not yet explored the mailscanner and the milter configuration and don't know if they resolve this issue.

If yes, there will be a final summary !

Regards

Daniel



This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:49:14 EDT