I am trying to learn enough about writing decoders and local rules to
use ossec-hids in protecting my spam-filter gateways against
distributed zombie attacks but I've run into one that I haven't been
able to figure out.

The maillog entry that I want to decode is:

Nov 30 00:05:41 mxgf2 sm-mta-in[52687]: lAU558MF052687: Milter:
from=<[EMAIL PROTECTED]>, reject=550 5.7.1 Blocked, look at
http://cbl.abuseat.org/lookup.cgi?ip=155.69.124.178

That should trigger the `sendmail-reject' decoder so I added a child:

<decoder name="sendmail-blocked-cbl">
  <parent>sendmail-reject</parent>
  <prematch>ip=</prematch>
  <regex offset="after_prematch">^(\d+.\d+.\d+.\d+)</regex>
  <order>srcip</order>
</decoder>

Then in local_rules.xml:

<rule id="100060" level="0">
    <decoded_as>sendmail-blocked-cbl</decoded_as>
    <description>Zombie pc connection</description>
    <description>Blocked by cbl.abuseat.org.</description>
    <group>spam,</group>
</rule>

<rule id="100062" level="5">
    <if_sid>100060</if_sid>
    <description>Zombie pc detected and blocked.</description>
    <group>spam,</group>
</rule>

<rule id="100064" level="10" frequency="5" timeframe="60">
    <if_matched_sid>100062</if_matched_sid>
    <same_source_ip />
    <description>Multiple Zombie connections. </description>
    <group>multiple_spam,</group>
  </rule>

But - nothing.  Not even a 100062 alert (log_alert_level 1).

I have similar rules operating to detect multiple smf-sav-rcpt
connections that work OK, the only difference being that they do not
use child decoders.

Hopefully some one can point me in the right direction.  I've read
everything I can find but obviously I don't have something right.

Reply via email to