BUG - Nothing in the corpus folder but qu
Printed From: LogSat Software
Category: Spam Filter ISP
Forum Name: Spam Filter ISP Support
Forum Description: General support for Spam Filter ISP
URL: https://www.logsat.com/spamfilter/forums/forum_posts.asp?TID=5292
Printed Date: 31 December 2025 at 11:44pm
Topic: BUG - Nothing in the corpus folder but qu
Posted By: fischer
Subject: BUG - Nothing in the corpus folder but qu
Date Posted: 05 August 2005 at 1:31am
|
I've seen this reported before, but haven't seen a fix. I just installed 2.6.3.473 as a new install on Win2k3 Web edition, and the corpus folder is empty except for an empty queue folder. Other thread responses from admins seem to indicate that there should be more there. I even did the whole delete the folder and restart SF thing, and that didn't change anything. Consequently bayesian filtering does not work right now. Anyone have any ideas?
|
Replies:
Posted By: LogSat
Date Posted: 05 August 2005 at 7:51am
Please check to see if on the "Settings - Bayesian Filter" tab, the
checkbox for "Learn new incoming emails" is checked. Underneath it
there should be a lable that describes the "Bayesian Learning Status".
Can you please let us know what that value is? Also, note that it may
take 10 minutes for the database files to appear in the corpus
directory as incoming email is processed. Also note that there needs to
be a steady arrival of incoming emails to be processed for the Bayesian
database to be built.
-------------
Roberto Franceschetti
http://www.logsat.com" rel="nofollow - LogSat Software
http://www.logsat.com/sfi-spam-filter.asp" rel="nofollow - Spam Filter ISP
|
Posted By: fischer
Date Posted: 05 August 2005 at 9:20am
|
Learn new incoming emails is checked, but Bayesian Learning Status shows as inactive. Unchecking the box and saving then rechecking and saving again does not change the status. My SF installation has been running for several hours now and still nothing in the corpus folder except the empty queue folder.
|
Posted By: LogSat
Date Posted: 07 August 2005 at 6:51pm
Can you please zip and email a copy of your SpamFilter.ini file, along
with SpamFilter's activity logfile for the day you checked/unchecked
that box?
-------------
Roberto Franceschetti
http://www.logsat.com" rel="nofollow - LogSat Software
http://www.logsat.com/sfi-spam-filter.asp" rel="nofollow - Spam Filter ISP
|
Posted By: fischer
Date Posted: 07 August 2005 at 7:47pm
|
What email address should I sent it to?
|
Posted By: LogSat
Date Posted: 07 August 2005 at 8:55pm
support @ logsat.com
-------------
Roberto Franceschetti
http://www.logsat.com" rel="nofollow - LogSat Software
http://www.logsat.com/sfi-spam-filter.asp" rel="nofollow - Spam Filter ISP
|
Posted By: LogSat
Date Posted: 08 August 2005 at 10:33pm
Form the 1st logfile you sent for the 4th, we can see that the
timeframe it contains, from 22:03 to 23:57 (almost two hours) you only
received one email. As mentioned in our first reply, that there needs
to
be a steady arrival of incoming emails to be processed for the Bayesian
database to be built.
From the 2nd logfile from the following day instead, while the email
traffic is slightly higher, it may not be enough to trigger the saving
of the statistical database. The statistical data is cached in memory
for a few minutes, and only on occasion is then saved to disk (to
optimize performance). If there is not enough data in memory (in the
order of an email every 4-5 minutes) that data is lost and not saved.
The Bayesian filter really needs a steady flow of data for the
statistical information to be built, and will not activate itself until
at least 5000 spam and 5000 good emails are received. At the rate of an
email every 5 minutes you can see that it will not be practical to use
that filter.
Furthermore, from the SpamFilter.ini file you sent us, we see that your
settings for the Bayesian filter are set to 100%, which, as stated in
the GUI itself, indicates that the Bayesian filter is disabled. When
disabled, the filter will not create/update the bayesian database.
-------------
Roberto Franceschetti
http://www.logsat.com" rel="nofollow - LogSat Software
http://www.logsat.com/sfi-spam-filter.asp" rel="nofollow - Spam Filter ISP
|
|