Re: speaking of the new FH_HOST_EQ_D_D_D_D tests...

2007-05-31 Thread Per Jessen
Jason Haar wrote:

 Since upgrading from SA-3.1* to 3.2.0, we have discovered that it
 appears most small New Zealand businesses run mail servers on DSL
 links with PTR records of the format NN-NN-NN-NN.isp.carrier.nz.
 Hence they end up with 2.2 points
 (FH_HOST_EQ_D_D_D_D,FH_HOST_EQ_D_D_D_DB,RDNS_DYNAMIC) added without
 any real effort. That plus their sputty HTML mails pushes them into
 the 5-7 range.
 
 I know SA isn't really doing anything wrong, but are people in other
 countries seeing this too? If so, it may imply the default scores are
 too high?

I'm not using those particular rules, but the mail-server setup you
describe is fairly typical - and still wrong.  They need to ask the
provider for a proper reverse DNS entry, or use the providers
SMTP-server as a relay.


/Per Jessen, Zürich



Re: speaking of the new FH_HOST_EQ_D_D_D_D tests...

2007-05-31 Thread Justin Mason

Jason Haar writes:
 We are getting heaps of false positives off these rules - ironically
 mainly from our IT services dealers/sellers/etc.
 
 Since upgrading from SA-3.1* to 3.2.0, we have discovered that it
 appears most small New Zealand businesses run mail servers on DSL links
 with PTR records of the format NN-NN-NN-NN.isp.carrier.nz. Hence they
 end up with 2.2 points
 (FH_HOST_EQ_D_D_D_D,FH_HOST_EQ_D_D_D_DB,RDNS_DYNAMIC) added without any
 real effort. That plus their sputty HTML mails pushes them into the 5-7
 range.
 
 I know SA isn't really doing anything wrong, but are people in other
 countries seeing this too? If so, it may imply the default scores are
 too high?

Well, FH_HOST_EQ_D_D_D_D is the main issue I'd say, since it has the
higher FPs and a much more significant score:

  score FH_HOST_EQ_D_D_D_D 2.599 1.992 1.692 1.212 # n=2
  score FH_HOST_EQ_D_D_D_DB 0.102 0.095 0.055 0.223 # n=2

  STATISTICS-set3.txt:  7.010  10.9120   0.13160.988   0.640.22  
FH_HOST_EQ_D_D_D_DB
  STATISTICS-set3.txt: 20.447  31.7873   0.45650.986   0.581.21  
FH_HOST_EQ_D_D_D_D

Since they overlap heavily with RDNS_DYNAMIC, perhaps we should consider
reducing/zeroing one or both scores for 3.2.1.  could you open a bug?

--j.

 Obviously I'm going to have to lower those scores to compensate - I bet
 more spam will come through too :-(  
 
 -- 
 Cheers
 
 Jason Haar
 Information Security Manager, Trimble Navigation Ltd.
 Phone: +64 3 9635 377 Fax: +64 3 9635 417
 PGP Fingerprint: 7A2E 0407 C9A6 CAF6 2B9F 8422 C063 5EBB FE1D 66D1


speaking of the new FH_HOST_EQ_D_D_D_D tests...

2007-05-30 Thread Jason Haar
We are getting heaps of false positives off these rules - ironically
mainly from our IT services dealers/sellers/etc.

Since upgrading from SA-3.1* to 3.2.0, we have discovered that it
appears most small New Zealand businesses run mail servers on DSL links
with PTR records of the format NN-NN-NN-NN.isp.carrier.nz. Hence they
end up with 2.2 points
(FH_HOST_EQ_D_D_D_D,FH_HOST_EQ_D_D_D_DB,RDNS_DYNAMIC) added without any
real effort. That plus their sputty HTML mails pushes them into the 5-7
range.

I know SA isn't really doing anything wrong, but are people in other
countries seeing this too? If so, it may imply the default scores are
too high?

Obviously I'm going to have to lower those scores to compensate - I bet
more spam will come through too :-(  

-- 
Cheers

Jason Haar
Information Security Manager, Trimble Navigation Ltd.
Phone: +64 3 9635 377 Fax: +64 3 9635 417
PGP Fingerprint: 7A2E 0407 C9A6 CAF6 2B9F 8422 C063 5EBB FE1D 66D1