How do I restore database to default like a clean install

[Stuart Haas <st...@gmail.com>]

Hi,

In error I ran sa-learn --clear, and now spamassassin misses most
spam.  How would I return the database to the state of a clean
install?

Stuart

Re: How do I restore database to default like a clean install

[jdow <jd...@earthlink.net>]

The user_prefs file may also want to be deleted to really restore to
default configuration.
{^_^}
----- Original Message ----- 
From: "Loren Wilton" <lw...@earthlink.net>


> I suspect you basically are in a clean install state at the moment, with a
> zapped Bayes db.  YOu can do a dump magic to see the number of spam and
ham
> tokens in the db.  Probably you just have to re-learn the mandatory couple
> hundred spam and ham messages to get Bayes running again.  Alternately you
> could completely remove the 3-4 Bayes files and let it start over, but I
> suspect that would be a waste of time.
>
> I'm surprised though that Bayes counts so heavily for you.  Granted it is
> helpful.  But even without net tests enabled, the standard rules should be
> catching will on the order of 50-80% of the incoming spam, if not more.
> Probably quite a lot more, especially if you have any of the standard
add-on
> rulesets.  If you are running net tests and SURBL you should be catching
> probably 95+% of the spam just with the net tests.
>
>         Loren

Re: How do I restore database to default like a clean install

[Loren Wilton <lw...@earthlink.net>]

I suspect you basically are in a clean install state at the moment, with a
zapped Bayes db.  YOu can do a dump magic to see the number of spam and ham
tokens in the db.  Probably you just have to re-learn the mandatory couple
hundred spam and ham messages to get Bayes running again.  Alternately you
could completely remove the 3-4 Bayes files and let it start over, but I
suspect that would be a waste of time.

I'm surprised though that Bayes counts so heavily for you.  Granted it is
helpful.  But even without net tests enabled, the standard rules should be
catching will on the order of 50-80% of the incoming spam, if not more.
Probably quite a lot more, especially if you have any of the standard add-on
rulesets.  If you are running net tests and SURBL you should be catching
probably 95+% of the spam just with the net tests.

        Loren