The difference between "FILTER_" and "DENIED_" is a little more 
complicated than you've described.  The "FILTER_" message appears 
whenever a filter is triggered, whether that filter will eventually 
block the message or not.  Those messages were originally added to show 
when whitelists were being matched, so administrators could understand 
why a "bad" connection wasn't being blocked (that's why you have to 
increase the log-level to see them).  The "DENIED_" message appears 
whenever a recipient is actually rejected.

So you are correct that one "FILTER_RBL_MATCH" message could match 
multiple "DENIED_RBL_MATCH" messages.  In that scenario, the remote 
server was found on an RBL, which is only checked once, but gave several 
recipients, which caused a number of rejections.  However, if the remote 
server is found on an RBL but the sender authenticates, you could see a 
"FILTER_RBL_MATCH" message with an "ALLOWED" message.  A similar 
situation could happen with a whitelisted sender -- you could see a 
"FILTER_RBL_MATCH" message, followed by a "FILTER_SENDER_WHITELIST" 
message, followed by an "ALLOWED" message.  Worse yet, if the "FILTER_" 
messages are specific to the recipient (e.g. recipient blacklists), you 
could see multiple "FILTER_" messages /and/ multiple "DENIED_" messages.

For the purposes of generating statistics, I think the "DENIED_" 
messages are much more useful because they show what actually happened 
instead of what spamdyke was thinking.  On my server, I just graph the 
"DENIED_" messages and I don't worry about multiple messages coming from 
a single connection.  In my mind, a single connection that generates 
multiple messages is the same as multiple connections that each generate 
one message -- the same number of spam emails were blocked either way.  
If you really want to track how many connections your mail server has 
gotten, I suppose you could parse the log entries to find spamdyke's PID 
and count the unique lines.  Beware, however, that PIDs can roll over 
rapidly on a busy server.

I hope that helps.

-- Sam Clippinger

Sebastian Grewe wrote:
> Hey list,
>
> I just looked at those stats and compared the output to what I am having
> on our boxes and I started wondering:
>
> When I check the log files, Spamdyke logs the following
>
> FILTER_RBL_MATCH : When listed in the RDNS
> DENIED_RBL_MATCH : For each recipient address in the mail
>
> So basically it will result in 1 FILTER match but 1 DENIED match for
> each mail address.
>
> Doesn't that mean that using the DENIED match will not result in the
> actual denied mails but rather in a much higher number? I am currently
> looking for both FILTER_ and DENIED_ flags and sum those up to find out
> how many mails I rejected - but I am guessing here that looking for
> FILTER_ alone would make more sense.
>
> Here my output, wrote the script today - Mirkos' output inspired me :)
> It's tailored to work for our environment though.
>
> Total      : 1571    (100.0000%)
> Legitimate : 123     (7.8200%)
>    |
>    |-    FILTER_WHITELIST                     : 61 (49.5900%)
>                |
>                |-    _RECIPIENT_WHITELIST     : 61 (100.0000%)
>
> Rejected   : 1448     (92.1700%)
>    |
>    |-    FILTER : 539 (37.2200%)
>    |        |
>    |        |-  _RDNS_MISSING                 : 192 (35.6200%)
>    |        |-  _OTHER                        : 12 (2.2200%)
>    |        |-  _RBL_MATCH                    : 297 (55.1000%)
>    |                |
>    |                |- _RBL_MATCH_SPAMHAUS    : 171 (57.5700%)
>    |                |- _RBL_MATCH_SPAMCOP     : 126 (42.4200%)
>    |
>    |-    DENIED : 905 (62.5000%)
>    |        |
>    |        |-  _RDNS_MISSING                 : 415 (45.8500%)
>    |        |-  _RBL_MATCH                    : 446 (49.2800%)
>    |        |-  _EARLYTALKER                  : 0 (0%)
>    |        |-  _SENDER_NO_MX                 : 14 (1.5400%)
>    |        |-  _TOO_MANY_RECIPIENTS          : 0 (0%)
>    |        |-  _UNQUALIFIED_RECIPIENT        : 0 (0%)
>    |
>    |-    Clamav : 4 (.2700%)
>             |
>             |-  Phishing                      : 4 (100.0000%)
>             |-  Trojan                        : 0 (0%)
>
>
> On Tue, 2009-09-01 at 15:52 -0500, Sam Clippinger wrote:
>   
>>>>> -----Original Message-----
>>>>> From: [email protected]
>>>>> [mailto:[email protected]] On Behalf Of Mirko
>>>>>           
>> Buffoni
>>     
>>>>> Sent: 01 September 2009 14:27
>>>>> To: spamdyke users
>>>>> Subject: Re: [spamdyke-users] Spam Stats
>>>>>
>>>>> Goods average between 500 and 2000 daily.  Figures are however
>>>>> pretty standard.  Spamdyke filters out about 60k attempts daily.
>>>>> Here are yesterday stats:
>>>>>
>>>>>      Good :   1025 =   0.68 %
>>>>>     Unsure :    183 =   0.12 %
>>>>>      Virus :     62 =   0.04 %
>>>>> BAD Sender:   5114 =   3.40 %
>>>>> BAD  Rcpt :    212 =   0.14 %
>>>>> Pure SPAM :  45997 =  30.56 %
>>>>>    SPAMMER :  97940 =  65.06 %
>>>>>       |
>>>>>       \.............BLACKLISTED_KEYWORD :  29608 =  30.23 %
>>>>>       \..............DENIED_EARLYTALKER :      3 =   0.00 %
>>>>>       \...............DENIED_IP_IN_RDNS :  30447 =  31.09 %
>>>>>       \................DENIED_RBL_MATCH :  23268 =  23.76 %
>>>>>       \.............DENIED_SENDER_NO_MX :  13070 =  13.34 %
>>>>>       \......DENIED_TOO_MANY_RECIPIENTS :      1 =   0.00 %
>>>>>       \....DENIED_UNQUALIFIED_RECIPIENT :      1 =   0.00 %
>>>>>       \.........................TIMEOUT :   1542 =   1.57 %
>>>>>
>>>>> ------------------------------
>>>>>      Total : 150533 = 100.00 %
>>>>>           
_______________________________________________
spamdyke-users mailing list
[email protected]
http://www.spamdyke.org/mailman/listinfo/spamdyke-users

Reply via email to