On 07/10/2014 09:00 PM, Eric Broch wrote:
On 7/10/2014 7:55 PM, Eric Shubert wrote:
I was sad to see Tom Hendrikx announce that he will no longer be
leading the DSPAM project. It looks as though DSPAM will likely just
become stagnate, and probably fall by the wayside.
With this in mind, I'm much less inclined to use DSPAM for delivery
(use-specific) spam filtering. If someone wants to make a case for
DSPAM use, it'll be a long uphill battle.
For those of you using DSPAM, please let me know how you're using it,
and perhaps we can come up with something using dovecot's deliver and
plugins that will provide the desired functionality in this area.
Thanks
That's a shame as I am using DSPAM presently with great results. I was
using it at the domain level with a call in the .qmail-default file
using this format: '| /usr/local/bin/dspamc --user "$EXT@$HOST"
--deliver=stdout | /home/vpopmail/bin/vdelivermail '' bounce-no-mailbox'
, however, I discovered that some attachments, and very few at that,
were being stripped only when the call to DSPAM was present. I am
currently testing on my own server DSPAM's use in each user's
.mailfilter file using this call: 'xfilter "/usr/local/bin/dspamc --user
$EXT@$HOST --deliver=stdout".' Hopefully the later will solve the
'stripped attachment' problem.
Anyway I'd be interested in calling DSPAM from Dovecot LDA using
Dovecot's configuration files but have yet to find a way.
---------------------------------------------------------------------
That might be possible with the dovecot antispam plugin
(http://wiki2.dovecot.org/Plugins/Antispam). Have a look at it.
My thinking for QMT is ultimately to have a global anti-spam scanner run
during the smtp connection (to avoid bounces), and another user-specific
anti-spam scanner run during delivery, which would not create a bounce.
Spamassassin could be used in both places with different configurations,
but that's not been set in stone. I was thinking of using DSPAM in one
place or the other, but now am rethinking that possibility. Ideally,
each user would have their own bayes database which they can train
according to their preferences. That might or might not be practical. My
present vpopmail bayes database is 84M. I'm a little concerned by what
individual bayes databases might do to performance (and storage).
We'll see how things pan out as we get to implementing dovecot's deliver
LDA. That enhancement should be getting close to the top of the list.
I'm eager to get server side filtering implemented.
Thanks.
P.S. Anyone else using DSPAM?
--
-Eric 'shubes'
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
