I'm not sure about your specific environment, but in my experience the things 
that helped me the most in blocking spam are:

- enable network tests
- enable the URIDNSBL plugin (init.pre)
- using sa-update with the spamassassin and SARE rule sets.

The SARE rules helped the most with stock quote spams.

I have most of the other plugins running, as well as having installed Pyzor and 
FuzzyOCR. I'm pretty happy so far! The one thing I would want is better spam 
processing performance. If a huge wash of spam hits the server, the server load 
goes up to 2 or 3 (shouldn't happen on a dual-core 3.2GHz receiving as little 
mail as we do).

Here is the output from the sa-stats program 
(<http://www.rulesemporium.com/programs/sa-stats-1.0.txt>) from my server for 
the past 24 hours:


Email:     5034  Autolearn:   417  AvgScore:  12.18  AvgScanTime:  6.02 sec
Spam:      3518  Autolearn:   341  AvgScore:  18.24  AvgScanTime:  6.11 sec
Ham:       1516  Autolearn:    76  AvgScore:  -1.89  AvgScanTime:  5.79 sec

Time Spent Running SA:         8.41 hours
Time Spent Processing Spam:    5.97 hours
Time Spent Processing Ham:     2.44 hours

TOP SPAM RULES FIRED
----------------------------------------------------------------------
RANK    RULE NAME                       COUNT  %OFMAIL %OFSPAM  %OFHAM        
----------------------------------------------------------------------
   1    HTML_MESSAGE                     2796    69.47   79.48   46.24
   2    URIBL_BLACK                      1577    32.20   44.83    2.90
   3    RCVD_IN_SORBS_DUL                1530    30.91   43.49    1.72
   4    URIBL_JP_SURBL                   1389    27.59   39.48    0.00
   5    RCVD_IN_NJABL_DUL                1387    28.01   39.43    1.52
   6    MY_CID_AND_STYLE                 1362    27.10   38.72    0.13
   7    SARE_GIF_ATTACH                  1335    27.55   37.95    3.43
   8    EXTRA_MPART_TYPE                 1229    24.89   34.93    1.58
   9    TVD_FW_GRAPHIC_ID1               1201    23.86   34.14    0.00
  10    PART_CID_STOCK                   1197    23.78   34.03    0.00
  11    MY_CID_ARIAL_STYLE               1141    22.67   32.43    0.00
  12    MY_CID_AND_ARIAL2                1141    22.69   32.43    0.07
  13    URIBL_OB_SURBL                   1027    20.44   29.19    0.13
  14    MIME_HTML_ONLY                   1004    22.77   28.54    9.37
  15    SARE_GIF_STOX                     963    19.19   27.37    0.20
  16    URIBL_SC_SURBL                    937    18.63   26.63    0.07
  17    HTML_IMAGE_ONLY_28                827    16.71   23.51    0.92
  18    URIBL_WS_SURBL                    806    16.29   22.91    0.92
  19    URIBL_SBL                         793    15.85   22.54    0.33
  20    PART_CID_STOCK_LESS               681    13.53   19.36    0.00
----------------------------------------------------------------------

TOP HAM RULES FIRED
----------------------------------------------------------------------
RANK    RULE NAME                       COUNT  %OFMAIL %OFSPAM  %OFHAM        
----------------------------------------------------------------------
   1    HTML_MESSAGE                      701    69.47   79.48   46.24
   2    NO_REAL_NAME                      665    14.88    2.39   43.87
   3    MIME_HTML_ONLY                    142    22.77   28.54    9.37
   4    BAYES_00                          130     2.74    0.23    8.58
   5    AWL                               128     3.06    0.74    8.44
   6    HTML_FONT_BIG                     128     9.24    9.58    8.44
   7    SPF_HELO_PASS                      59     4.59    4.89    3.89
   8    HTML_IMAGE_RATIO_02                53     2.82    2.53    3.50
   9    SARE_UNI                           52     1.23    0.28    3.43
  10    SARE_GIF_ATTACH                    52    27.55   37.95    3.43
  11    INFO_TLD                           50     3.08    2.98    3.30
  12    HTML_TAG_EXIST_TBODY               45     1.15    0.37    2.97
  13    URIBL_BLACK                        44    32.20   44.83    2.90
  14    MISSING_HB_SEP                     41     1.61    1.14    2.70
  15    USER_IN_WHITELIST                  35     0.70    0.00    2.31
  16    UNPARSEABLE_RELAY                  35     3.32    3.75    2.31
  17    EMPTY_MESSAGE                      32     1.13    0.71    2.11
  18    FORGED_RCVD_HELO                   27     3.95    4.89    1.78
  19    RCVD_IN_SORBS_DUL                  26    30.91   43.49    1.72
  20    BAYES_50                           26     0.87    0.51    1.72
----------------------------------------------------------------------

Quinn




On Mon, 13 Nov 2006 15:24:55 -0600, Ryan Gibbons wrote:
> My server (not just my domain) is getting hit hard with spam related to
> stock quotes.  It is plan text, no links no html, and of course the
> envlope changes each time.  I have go through with sa-learn and try to
> mark them individually but they are still getting through, some are even
> being learned as ham b/c they are generated a score of over -3, (*note
> to self, I might want to bump that up) and very few are being marked
> anything lower then 3.  On overage, it is coming across as zero.
> 
> Thunderbird sees it has spam, so it is possible to catch these, I just
> don't know enough about spamassassin to create a rule set to catch it. 
> I use rules de jour and moderate RBL block list.
> 
> Anybody have any hits, If you want to see the message, let me know and I
> can put it up here.

---------------------------------------------------------------------
     QmailToaster hosted by: VR Hosted <http://www.vr.org>
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to