On Thu, Apr 25, 2002 at 10:46:38PM -0400, Brian Kendig wrote:
> Does anyone have an easier (but reliable) way to have SpamAssassin
> automatically report spam to the upstream providers?
The method I use is that all incoming mails that are flagged as spam
(and from my spam traps) are saved into a folder called spam-work.
I go through that occasionally (at least once an hour typically) and
take the actual spam (99%+) and save it to another folder called hs.
I then have a cron job that takes that folder, use another script I
wrote called handlespam, and go through each message:
- Report to Razor if RAZOR_CHECK didn't match the message
- Check for open relays and report to various sites
- Forward the message to SpamCop (which I use for my reporting needs)
- Display new entries for my sendmail access database
- If SpamAssassin didn't flag the message as spam (score < 5.0),
send (as an attachment) to spamassassin-sightings.
- Archive the spam to a specific folder (called spammers for me)
It works very well for me -- so far I have almost 8500 entries in my
access database which cuts out a lot of spam. I block sites in various
DNSBL's at the SMTP level (mostly open relays since the false positive
rate is much smaller for that category.) SA catches almost all of the
stuff that gets through after that, and the cycle begins anew. ;)
If you're interested, you can find a lot of my information at these URLs:
http://www.kluge.net/mailfiltering/
- Mostly what I do for filtering and info for my users
http://www.kluge.net/~felicity/random/
- Various scripts including the handlespam and testrelay
scripts.
http://www.kluge.net/~felicity/spam-reported-{day,week,month}.png
http://www.kluge.net/~felicity/spam-syslog-{week,month}.png
- Just informational. I stole (er, borrowed?) some
GNUplot configs from someone who posted it to the SA
list (thanks!), and then rigged up a spam archive/syslog
parser -> MySQL script -> GNUplot script. It now
runs through cron and plots the overall SA scores per
message over time. The syslog ones update once a week
and are the SA scores from all messages that have been
scanned via spamd. The reported one is a parsing of
my spam archive updated every 6 hours.
I've been thinking about also plotting which ones were
in razor to get a better feel for how effective it is
as a test/system. At last check, razor is around 55%
for me. Not great, but not bad. SA, OTOH, is at 94%
effectiveness so far this month.
--
Randomly Generated Tagline:
"I did it the old fashioned way... I begged." - Wolf
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
