I thought I had solved my sa-learn issues by always doing su qscand
whenever I want to run sa in debug mode or run sa-learn. Today I got
some false positives and some false negatives and decided to train the
bayes db using them. After I ran sa-learn for both ham and spam I
looked at the bayes db's file sizes and they were the same. It looked
like the timestamp on the file got updated, but the size didn't
increase. Is that normal?
First I just ran sa-learn my usual way, but from the user qscand:
/var/spool/qscan/.spamassassin
[EMAIL PROTECTED] ~/.spamassassin $ ls -la
total 173752
drwxrwxrwx 2 qscand qscand 4096 Apr 1 10:08 .
drwxr-xr-x 7 qscand root 4096 Apr 1 03:16 ..
-rw------- 1 qscand qscand 41400 Apr 1 10:13 bayes_journal
-rwxrw-rw- 1 qscand qscand 2609152 Apr 1 10:08 bayes_seen
-rw------- 1 qscand qscand 167100416 Feb 22 14:35 bayes_seen.old
-rw------- 1 qscand qscand 5218304 Apr 1 10:08 bayes_toks
-rw------- 1 qscand qscand 5308416 Feb 22 14:35 bayes_toks.old
-rw-r--r-- 1 qscand qscand 1487 Jan 31 18:23 user_prefs
[EMAIL PROTECTED] ~/.spamassassin $ sa-learn --showdots --mbox --ham
/home/domainmail/mail/Ham
....
Learned tokens from 4 message(s) (4 message(s) examined)
[EMAIL PROTECTED] ~/.spamassassin $ sa-learn --showdots --mbox --spam
/home/domainmail/mail/Spam
....
Learned tokens from 4 message(s) (4 message(s) examined)
[EMAIL PROTECTED] ~/.spamassassin $ ls -la
total 173708
drwxrwxrwx 2 qscand qscand 4096 Apr 1 10:15 .
drwxr-xr-x 7 qscand root 4096 Apr 1 03:16 ..
-rwxrw-rw- 1 qscand qscand 2609152 Apr 1 10:15 bayes_seen
-rw------- 1 qscand qscand 167100416 Feb 22 14:35 bayes_seen.old
-rw------- 1 qscand qscand 5218304 Apr 1 10:15 bayes_toks
-rw------- 1 qscand qscand 5308416 Feb 22 14:35 bayes_toks.old
-rw-r--r-- 1 qscand qscand 1487 Jan 31 18:23 user_prefs
You can see from the above that the file sizes didn't change.
So, then I tried it this way:
[EMAIL PROTECTED] ~ $ sa-learn --progress --dbpath
/var/spool/qscan/.spamassassin/ --mbox --spam
/home/domainmail/mail/Spam
100%
[============================================================================]
74.35 msgs/sec 00m00s DONE
Learned tokens from 0 message(s) (4 message(s) examined)
[EMAIL PROTECTED] ~ $ sa-learn --progress --dbpath
/var/spool/qscan/.spamassassin/ --mbox --ham /home/domainmail/mail/Ham
80% [============================================================
] 12.72 msgs/sec 00m00s DONE
Learned tokens from 0 message(s) (4 message(s) examined)
and it looks like it's already seen those messages.
[EMAIL PROTECTED] ~ $ ls -la .spamassassin/total 173720
drwxrwxrwx 2 qscand qscand 4096 Apr 1 10:27 .
drwxr-xr-x 7 qscand root 4096 Apr 1 03:16 ..
-rw------- 1 qscand qscand 11616 Apr 1 10:27 bayes_journal
-rwxrw-rw- 1 qscand qscand 2609152 Apr 1 10:27 bayes_seen
-rw------- 1 qscand qscand 167100416 Feb 22 14:35 bayes_seen.old
-rw------- 1 qscand qscand 5218304 Apr 1 10:27 bayes_toks
-rw------- 1 qscand qscand 5308416 Feb 22 14:35 bayes_toks.old
-rw-r--r-- 1 qscand qscand 1487 Jan 31 18:23 user_prefs
Is sa-learn hitting the files in this directory and if so, why isn't
the file size increasing? Anything else I should be doing when I'm
training?
(by the way, the user qscand's home directory is /var/spool/qscan)
____________________________________________________________________________________
Don't get soaked. Take a quick peek at the forecast
with the Yahoo! Search weather shortcut.
http://tools.search.yahoo.com/shortcuts/#loc_weather
