[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Re[6]: [vps-mail] Large Spamassassin bayes_toks file



I've got plenty of ham and spam in some of my user's token files with my
/home/jsmith/.spamassassin having well over the 5 meg limit. The point
Abigail notes that the debug reports shows no ham is because it is looking
at /root/.spamassassin/bayes_toks and I can't figure out how to convince it
to look at my user's database's instead. 

And after a month or so of "training" it and building up a database of spam
& ham, I don't want to tell my client's that they need to start over again
from zero every time it reaches 5 megs because we can't figure out how to
use the expire feature on a shared VPS2 server.

Matt's suggestion of doing a symlink from root to each user's files may be
the manual hack I'll have to try until we can figure out how to make this
process less cumbersome. Any better solutions that anyone can come up with
to try?

Thanks,

Jim Smith

> -----Original Message-----
> From: owner-vps-mail@xxxxxxxxxxxx 
> [mailto:owner-vps-mail@xxxxxxxxxxxx] On Behalf Of Abigail Marshall
> Sent: Sunday, March 07, 2004 1:28 AM
> To: Jim Smith
> Subject: Re[6]: [vps-mail] Large Spamassassin bayes_toks file
> 
> JS> Ok, I upgraded to Spamassassin's version 2.63 from 2.61. 
> I'm still 
> JS> not able to reduce the size of my token or journal.
> 
> Jim, your lint shows the following result:
> 
> JS> debug: bayes: found bayes db version 2
> JS> debug: bayes: Not available for scanning, only 0 ham(s) 
> in Bayes DB 
> JS> < 200
> 
> Bayes need at least 200 ham emails to function.
> 
> Maybe the problem is that the program won't allow reduction 
> in the token database unless there is the minimum required 
> number of spam + ham.
> 
> This makes sense from a programming standpoint, because if a 
> database function eliminating older tokens in order to reduce 
> the size of the database were to allow the corpus to go below 
> the minimum requisite to function, then the Bayes functioning 
> would be compromised.
> 
> I really think you need to feed your system a bunch of ham.
> Or else delete your entire database and start over, because 
> what you have is worthless. Bayes isn't running in any case 
> without sufficient ham.
> 
> -Abigail
> 
> ======================================================================
> This is <vps-mail@xxxxxxxxxxxx>       <http://www.perlcode.org/lists/>
> Before posting a question, please search the archives (see above URL).
> 

======================================================================
This is <vps-mail@xxxxxxxxxxxx>       <http://www.perlcode.org/lists/>
Before posting a question, please search the archives (see above URL).


Main Index | Thread Index
Match: Format: Sort by:
Search: