All,
I recently upgraded a RedHat 8 mail server to RedHat Enterprise
Linux 3. Rather than do an in-place upgrade, I chose to do a complete
backup of the RH8 server, format and install a fresh RHEL3. After doing
so, I replaced all the necessary config including my centralized bayes
database from my previously working install. Now, spamassassin is not
recognizing all the spam messages that I had learned. If I type
"spamassassin -D --lint" it shows the following:
debug: bayes: 18871 tie-ing to DB file
R/O /etc/mail/spamassassin/bayes/bayes_toks
debug: bayes: 18871 tie-ing to DB file
R/O /etc/mail/spamassassin/bayes/bayes_seen
debug: debug: Only 0 spam(s) in Bayes DB < 200
So it's claiming that there are no spams in the DB. It doesn't complain
that it can't open the DB as you might expect with different db
versions. In fact, I've tried to do dump and load between db4 and db3
in numerous different ways to no avail. I've even tried downloading the
script that is mentioned on the Wiki to do a text dump and recreate the
DB - all to no avail.
The thing that's confusing is that if I do a sa-learn --rebuild on the
DB that I restore from my old install, I see this:
[EMAIL PROTECTED] bayes]# sa-learn --rebuild
..
synced Bayes databases from journal in 1 seconds: 2206 unique entries
(3871 total entries)
Which CLEARLY shows that I have 3871 entries!! So why is SA ignoring
this!?
By the way - I know that it's not working b/c I've had a huge increase
in spam since the upgrade and none of my headers are showing a
substantial use of the BAYES test....as they did before.
I've searched the FAQs, searched the web, and spent a good deal of time
on this and just can't figure it out. I appreciate any help.
Thanks,
Brian
