Hi,

Use Postfix? Use spamd? Have a small mail log? Ever wonder which hosts
are sending the most spam into your system? Wonder no longer -
spamsources.sh is here to answer all your questions about who is
spamming you. Maybe.

http://www.cynistar.net/~apthorpe/code/sa-contrib/spamsources.sh

This 'simple' shell script makes four passes through your mail log:

- gathering the PID of every spamd process that identified a message as
spam

- finding the message-id of each identified spam

- finding the Postfix ID from each message-id

- finding the connecting host and IP address for each Postfix ID

then sorts the list of hosts and arranges them in order of spamminess.
Sample output:
    273 shared-austin.bos.hosting.com[64.55.166.99]
     32 lists.sourceforge.net[66.35.250.206]
      9 daedalus.apache.org[208.185.179.12]
      6 cherry.ease.lsoft.com[209.119.0.109]
      5 sloth.good-day.net[220.218.54.194]
      3 lina.cynistar.net[66.143.181.11]
      3 dsl.80.119.networkiowa.com[209.234.80.119]
      2 xuxa.iecc.com[208.31.42.42]
      1 rrcs-central-24-92-133-152.biz.rr.com[24.92.133.152]
      1 pop1-357.catv.wtnet.de[213.209.65.102]
      1 pc-200-74-21-128.pvaldivia2.pc.metropolis-inter.com[200.74.21.128]
      1 mta08bw.bigpond.com[144.135.24.137]
      1 mta07bw.bigpond.com[144.135.24.134]
      1 maiden.genestate.com[212.21.116.19]
      1 gnu-designs.com[65.172.152.98]
      1 fl-atlnfl-u3-c4b-228.atlsfl.adelphia.net[67.22.99.228]
      1 dyn-81-167-219-125.ppp.tiscali.fr[81.167.219.125]
      1 bellevue.puremagic.com[209.189.198.108]
      1 adsl-dc-20d8b.adsl.wanadoo.nl[81.70.43.139]
      1 adsl-69-104-60-150.dsl.pltn13.pacbell.net[69.104.60.150]
      1 D40A2138.rev.stofanet.dk[212.10.33.56]
      1 73.254.39-62.rev.gaoland.net[62.39.254.73]

(note: there are FPs in here[1])

Consider this to be proof-of-concept code. It's specific to Postfix but
can be easily modified to work with other MTAs. It makes four passes
through the log file; this could be reduced by writing it in perl,
naively storing all the message-ids and MTA IDs and stitching them all
together at the end. This eats more memory both by using perl and by
naive parsing but avoids race conditions if the logs rotate while the
script is running.

A more robust variant could conceivably watch for spam/ham ratios, watch
for anomalies (sudden statistically-significant increases in spam/ham
ratios), and temporarily add firewall blocks or MTA rules (like
dynamically updating an MTA access.db)

Have fun!

-- Bob

[1] shared-austin.bos.hosting.com[64.55.166.99] forwards most of the mail
directed to my ancient soon-to-be-retired address <[EMAIL PROTECTED]>,
lina.cynistar.net is my backup MX, good-day.net runs the Sylpheed list,
lsoft.com hosts SPAM-L, iecc.com hosts the spamtools list, puremagic.com
hosts the greylisting list, gnu-designs.com hosts the Jpilot lists and
loses DNS periodically - aside from Apache and Sourceforge, the rest is
mostly consumer dialup and broadband space hence my irritating rants
about ISPs needing to redirect their dynamic customers' outbound port 25
traffic to internal servers by default.

This doesn't show the number of connections blocked for having no/broken
rDNS or broken HELO, or being listed on xbl.spamhaus.org,
sbl.spamhaus.org, dnsbl.njabl.org, dnsbl.sorbs.net, rhsbl.sorbs.net,
opm.blitzed.org, relays.ordb.org, bogusmx.rfc-ignorant.org, or
list.dsbl.org. The biggest source of crap entering (or trying to enter)
my mail system is dynamically-allocated consumer space. YMMV but I doubt
it.


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk

Reply via email to