tablets and chemists

[Mike Schrauder <MI...@specialtyblades.com>]

Hi all,	
	I really wish I had time to contribute to this list, but you
folks are on a completely higher level.  I have sa 2.63 on debian woody
using the filter account to protect an exchange server.
I set it up w/ the very helpful info at
http://advosys.ca/papers/postfix-filtering.html.
This was about 2 years ago and it has been working VERY well.  Now some
more spam is getting through and I have two questions.  
1.  how scary for a novice would it be to apt-get upgrade spammassassin
to 3.0.2-1 and expect everything to keep on working?

2.  if that is to stressful, who has already formed rulesets that is
compatible w/ 2.63 that
I could drop in to catch all this drug spam that is getting through? 

I am already using razor, dcc, bayes and 

SPAMCOP_URI_RBL
WS_URI_RBL
OB_URI_RBL
AB_URI_RBL   not that I can really tell if they are working optimally
:^{


Mike Schrauder
Specialty Blades, Inc.
www.specialtyblades.com
www.olfablades.com

Re: tablets and chemists

[Bob Proulx <bo...@proulx.com>]

Matt Kettler wrote:
> Mike Schrauder wrote:
> >I have sa 2.63 on debian woody using the filter account to protect
> >...
> >This was about 2 years ago and it has been working VERY well.  Now some
> >more spam is getting through and I have two questions.  
> >1.  how scary for a novice would it be to apt-get upgrade spammassassin
> >to 3.0.2-1 and expect everything to keep on working?

On Debian if you upgrade the Berkeley DB libraries on the system you
will need to upgrade the SA bayes databases using the db4.2_upgrade
upgrade command in the db4.2-util package.  Change the 4.2 in that
name to the exact version you are moving to after a Berkeley DB
upgrade.  The BDB file format upgrade has always worked easily for me.

This is really outside spamassasin and is controlled by the underlying
system perl in use.  If you don't change the perl BDB libraries then
SA won't change BDB formats either.  But better knowing this than
searching for it later.  But don't fear it.  It is not a serious
problem.  I would definitely upgrade to SpamAssassin 3.x.

> 3.x has some configuration option changes, which are documented in the
> upgrade file:
> http://svn.apache.org/repos/asf/spamassassin/branches/3.0/UPGRADE

Good stuff in the UPGRADE file.  The only real thing I remember
needing to change here, there may have been others, it was quite a
while ago, was the -P option going away.  But really everything worked
well and any problems were very minor.

One thing I did was to remove all of the custom rule processing I had
added with the 2.6x series and started out fresh with only the stock
3.x rules.  Then tweaked up the scores such as increasing the bayes
score points.  Almost all of the additional rules I had installed into
2.x are now standard in 3.x.  So things got quite a bit simpler.

Bob

Re: tablets and chemists

[Matt Kettler <mk...@evi-inc.com>]

Mike Schrauder wrote:

>Hi all,	
>	I really wish I had time to contribute to this list, but you
>folks are on a completely higher level.  I have sa 2.63 on debian woody
>using the filter account to protect an exchange server.
>I set it up w/ the very helpful info at
>http://advosys.ca/papers/postfix-filtering.html.
>This was about 2 years ago and it has been working VERY well.  Now some
>more spam is getting through and I have two questions.  
>1.  how scary for a novice would it be to apt-get upgrade spammassassin
>to 3.0.2-1 and expect everything to keep on working?
>  
>
3.x has some configuration option changes, which are documented in the
upgrade file:
http://svn.apache.org/repos/asf/spamassassin/branches/3.0/UPGRADE

I'd strongly recommend the upgrade to 3.0.x, or at MINIMUM an upgrade to
2.64 to correct the DoS vulnerability that 2.63 has.

Upgrading to 2.64 should be completely painless if you're afraid of 3.x.

3.x also makes broken trust path issues painfully obvious, whereas in
2.6x the problem would be more subtle in nature (mostly DUL  FP's). If
you see spam matching ALL_TRUSTED with 3.x read:

http://wiki.apache.org/spamassassin/TrustPath

>2.  if that is to stressful, who has already formed rulesets that is
>compatible w/ 2.63 that
>I could drop in to catch all this drug spam that is getting through?
>
If you don't go to 3.x, you might want to add antidrug.cf to
/etc/mail/spamassassin:

http://mywebpages.comcast.net/mkettler/sa/antidrug.cf

3.0 and higher include this ruleset by default as a part of
{$PREFIX}/share/spamassassin/20_drugs.cf

Re: tablets and chemists

[Jeff Chan <je...@surbl.org>]

On Friday, April 29, 2005, 5:52:39 AM, Mike Schrauder wrote:
> I am already using razor, dcc, bayes and

> SPAMCOP_URI_RBL
> WS_URI_RBL
> OB_URI_RBL
> AB_URI_RBL   not that I can really tell if they are working optimally
> :^{

Be sure to add the JP list.  It's very good:

  http://www.surbl.org/quickstart.html

Jeff C.
-- 
Jeff Chan
mailto:jeffc@surbl.org
http://www.surbl.org/