EXTRA NOTE: This will take very long the first time and Mailscanner can be configured to skip Spamassassin checks during this time or hold all E-mail until finished. If it's a company server, I recommend configuring Mailscanner to hold all E-mails during expiry and starting the task at friday evening. You can run it with "2>&1 > ~/expire.log &" at the end so it redirects all output to ~/expire.log and rund in the background. (you can end your SSH session) (Can someone check this code? I'm not very fluent in output redirection)<br>
<br>You could also delete the whole Bayes DB and download a recent copy of someone that does run expiry every now and than :P<br><br>It is perfectly normal for an E-mail to have a negative score, this just means that the E-mail is definitely not spam. If there are SPAM E-mails that are getting a negative score, you got a problem ;)<br>
A spamfilter aims to give all HAM a score as low as possible and all SPAM a score as high as possible. Alle E-mails that hover around the threshold score (usually 6) are unsure and should be checked manually.<br><br><div class="gmail_quote">
On 17 November 2011 12:01, Markus Nilsson <span dir="ltr"><<a href="mailto:markus@markusoft.se">markus@markusoft.se</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
Please reply to the previous mail to keep the thread!<br>
<div class="im"><br>
----- Original Message -----<br>
> From: "eric le corre" <<a href="mailto:eric_le_corre@msn.com">eric_le_corre@msn.com</a>><br>
> To: <a href="mailto:mailscanner@lists.mailscanner.info">mailscanner@lists.mailscanner.info</a><br>
</div><div class="im">> Sent: torsdag, 17 nov 2011 11:31:33<br>
> Subject: Re: Very bad score spamassassin<br>
><br>
> Spamassassin configuration is in autolearning. So there has to be<br>
> found in<br>
> spam HAM.<br>
> Occasionally, I learn to SpamAssassin manually.<br>
><br>
> But this problem has been the case since I installed mailscanner.<br>
> Since the<br>
> beginning I have negative cores.<br>
><br>
> Do I have to disable the AutoLearn, how?<br>
<br>
</div>No you can keep autolearn, just make sure you manually train those that fail!<br>
<div class="im">><br>
> how can your remove the old tokens?<br>
><br>
<br>
</div>e.g. by running<br>
<br>
sa-learn --force-expire -p /etc/MailScanner/spam.assassin.prefs.conf<br>
<br>
NOTE: This will remove lots of tokens from your Bayes DB, make sure you have reviewed your settings in the config file first, and make a backup of the bayes DB if you are unsure you want to do this!<br>
<br>
Specifically check your<br>
<br>
bayes_expiry_max_db_size<br>
<br>
setting<br>
<font color="#888888"><br>
/Markus<br>
</font><div><div></div><div class="h5">--<br>
MailScanner mailing list<br>
<a href="mailto:mailscanner@lists.mailscanner.info">mailscanner@lists.mailscanner.info</a><br>
<a href="http://lists.mailscanner.info/mailman/listinfo/mailscanner" target="_blank">http://lists.mailscanner.info/mailman/listinfo/mailscanner</a><br>
<br>
Before posting, read <a href="http://wiki.mailscanner.info/posting" target="_blank">http://wiki.mailscanner.info/posting</a><br>
<br>
Support MailScanner development - buy the book off the website!<br>
</div></div></blockquote></div><br>