auto-whitelist, bayes_journal, bayes_seen, and bayes_toks.
The bayes_toks file was practically empty so I just went ahead and
deleted all 4 of those, and re-ran sa-learn. It seems to have fixed the
issue for whatever reason. Now when sa-learn is re-ran it doesn't
"learn" from the same messages over and over.
I guess one of those files became corrupted for whatever reason.
Thanks!
Theo Van Dinter wrote:
When you say "the database", do you mean "bayes_toks" or "bayes_toks
and bayes_seen"? If the former, you need to grant write privs to
bayes_seen as well.
Also, when in doubt, run w/ -D to see what's going on.
On Sun, May 31, 2009 at 1:41 PM, Russell Jones <rjo...@eggycrew.com> wrote:
I am running a global bayes database. The file permissions for the database
is 0666. For some reason I just realized that sa-learn is not remembering
the messages it's already learned from. I've checked the bayes file
permissions and everything else I could think of, but if you run sa-learn,
wait for it to finish, and then run it again it learns off of the same
messages. (IE, learned 5 tokens from 5 messages. This will show everytime
you run sa-learn). Nothing has changed from my configuration that should
have caused this.
I am running SpamAssassin 3.2.5, Perl 5.8.8. There are no errors shown when
sa-learn finishes. Any ideas what I could look at as the cause for this?
