If you need to run multiple decoders on the same log to extract additional pieces of information (and at the same time do not affect the original decoder), we have a simple way to do so.

Just create multiple child decoders with the same name and no “prematch” and all of them will be evaluated. A good example is for the Microsoft event logs. This is our original decoder for Windows:

<decoder name=”windows-sub1″>
<type>windows</type>
<prematch>^WinEvtLog: </prematch>
<regex offset=”after_prematch”>^\.+: (\w+)\((\d+)\): (\.+): </regex>
<regex>(\.+): \.+: (\S+): </regex>
<order>status, id, extra_data, user, system_name</order>
<fts>name, location, user, system_name</fts>
</decoder>

You will see that we do not extract a soure IP address from there (only user, system name, location, etc). If we wanted to extract the source IP address whenever it is available, we would need to check all the variations (Source Network Address:, Source IP Address:, etc). By using sub-decoders we can check for them without affecting the original:

<decoder name=”windows”>
<type>windows</type>
<prematch>^WinEvtLog: </prematch>
</decoder>

<decoder name=”windows-sub1″>
<type>windows</type>
<parent>windows</parent>
<regex offset=”after_parent”>^\.+: (\w+)\((\d+)\): (\.+): </regex>
<regex>(\.+): \.+: (\S+): </regex>
<order>status, id, extra_data, user, system_name</order>
<fts>name, location, user, system_name</fts>
</decoder>

<decoder name=”windows-sub1″>
<type>windows</type>
<parent>windows</parent>
<regex offset=”after_regex”>Source Network Address: (\S+)</regex>
<order>srcip</order>
</decoder>

<decoder name=”windows-sub1″>
<type>windows</type>
<parent>windows</parent>
<regex offset=”after_regex”>Source IP Address: (\S+)</regex>
<order>srcip</order>
</decoder>

You see that all of them are named “windows-sub1″ and none of them have a prematch. Another option is to use the “after_regex” in the offset so OSSEC won’t check the whole string again…

By default OSSEC has a static timeout on our active response scripts. You specify the action and how long to block the IP Address:

<active-response>
<command>host-deny</command>
<location>local</location>
<level>6</level>
<timeout>600</timeout>
</active-response>

Which works well for most of the time. However, if you need to increase the timeout for repeated offenders (say the same IP trying to brute force your server all the time), you would have to do this manually.

To solve this problem, we added a new option to the active response config (repeated_offenders):

<active-response>
<repeated_offenders>30,60,120</repeated_offenders>
</active-response>

It allows you to specify a comma separated list of timeouts per re-incidence (in minutes). So the first time an IP is blocked it would use the default timeframe (600 seconds). If it gets blocked again, it would use the first entry in the <repeated_offenders> list (30 minutes = 1800 seconds), then 60 minutes and then 120 minutes… Anything above that would use the latest entry.

To use this option, you have to download the latest snapshot from: http://bitbucket.org/dcid/ossec-hids (just click on get source).