Top posting for a reason. Does anyone else think that this entire reply
should be up on a wiki??? Bob did a great explanation to an often asked
question. Mind if I clean it up and post to the wiki, Bob?

--Chris

> -----Original Message-----
> From: Bob Apthorpe [mailto:[EMAIL PROTECTED]
> Sent: Tuesday, February 10, 2004 9:05 AM
> To: [EMAIL PROTECTED]
> Subject: Re: Very Slow SA respons v.211
> 
> 
> Hi,
> 
> On Tue, 10 Feb 2004 06:19:22 -0600 George Kasica 
> <[EMAIL PROTECTED]> wrote:
> 
> > >On Tue, 10 Feb 2004 16:54:00 +1300, you wrote:
> > 
> > >At 16:48 10/02/2004, George Kasica wrote:
> > >>In the last 3-4 days I'd suddenly seen a massive drop in 
> speed in SA
> > >>2.11 here I occasionally see the an error series like this:
> > >
> > >Sounds extremely ancient to me, considering 2.63 is the 
> latest. Any version 
> > >earlier than 2.6 is going to have "issues" out of the box, 
> such as trying 
> > >to query RBL lists that no longer exist, and if you really 
> are running 2.11 
> > >this is likely why...
> >
> > OK, I'll admit its old, let me ask the next question, how 
> hard is the
> > upgrade to the current 2.63 version. I'm not a perl wizard, but am a
> > decent Unix admin. Is there a way to convert the existing rule sets?
> 
> Your existing custom rules or the stock rules that come with 
> SA? If you
> put your custom rules under /etc/mail/spamassassin, they will be
> preserved between upgrades. The stock rules that ship with SA will be
> upgraded since many rules have been added and dropped as the character
> of spam has changed, plus some rules depend on SA code which also may
> have changed. If you've added custom rules to the config files under
> /usr/share/spamassassin, you'll need to diff those against 
> the originals
> and extract your changes. Custom rules in ~/.spamassassin/user_prefs
> will be preserved but like those in /etc/mail/spamassassin, 
> they'll need
> to be reviewed to see if they conflict with the current version of SA.
> 
> From INSTALL:
> 
> ----
> "Note For Users Upgrading From SpamAssassin 2.3x or 2.4x
> -------------------------------------------------------
>  
> SpamAssassin no longer includes code to handle local mail 
> delivery, as it
> was not reliable enough, compared to procmail.  So now, if 
> you relied on
> spamassassin to write the mail into your mail folder, you'll have to
> change your setup to use procmail as detailed below.  If you used
> spamassassin to filter your mail and then something else 
> wrote it into a
> folder for you, then you should be fine.
>  
> Support for versions of the optional Mail::Audit module is no longer
> included.
>  
> The default mode of tagging (which used to be ***SPAM*** in 
> the subject
> line) no longer takes place.  Instead the message is rewritten."
> ----
> 
> One important question: which version of perl do you have? 
> IIRC, 2.6x is
> the last version that will support perl 5.005. Aside from that, you'll
> want to freshen your perl installation (perl -MCPAN -e shell; then use
> the 'r' command to see what's out of date and upgrade those modules as
> prudent.) Then install the prerequisites for 2.63:
> 
>   - ExtUtils::MakeMaker >= 5.45 (included in Perl 5.6.1 and later)
>   - HTML::Parser >= 3.24 (from CPAN)
>   - Sys::Syslog          (from CPAN)
> 
> and probably
> 
>   - DB_File         (from CPAN, included in many distributions)
>   - Digest::SHA1    (from CPAN)
>   - Net::DNS        (from CPAN)
>   - Time::HiRes    (from CPAN)
> 
> Maybe it's best to install SA to a local user account, test it, then
> upgrade the system once you're comfortable with the new version.
> 
> > I've heard 2.63 needs to get "trained" where as 2.11 did not, how
> > hard is this process as we don't save spam here but just mark it and
> > thats all.
> 
> You can run SA with or without network tests, and with or without the
> Bayesian (statistical) classifier. Training SA is fairly 
> simple; if you
> can save at least 200 pieces each of spam and ham (not-spam), you can
> use sa-learn to initially train the Bayesian analyzer. It can also
> autolearn, learning from messages above or below some learning
> thresholds but this will take much longer. If you don't care, just
> disable Bayes. SA should work just fine without it.
> 
> It's pretty easy for people here to tell you to upgrade, it's another
> matter completely to upgrade without your users ever knowing. I think
> you'll be happier with a more recent version; if you have any 
> questions,
> please ask.
> 
> Good luck!
> 
> -- Bob
> 

Reply via email to