<html><head>
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
</head><body bgcolor="#FFFFFF" text="#000000">Nikolaos Milas wrote:
<blockquote cite="mid:5048F267.2050104@noa.gr" type="cite">Hello,
<br>
<br>
I am storing spam (from Amavis) to:
<br>
<br>
$QUARANTINEDIR = '/var/blockedmail'
<br>
<br>
i.e. to '/var/blockedmail/spam/' as:
<br>
<br>
$spam_quarantine_method = "local:spam/%n-%m" ;
<br>
<br>
Question: Can I use stored spam files (e.g. "02792-01-bzM11CH7LOH8") as
they are (i.e. without any modification/conversion) to train
spamassassin for ham (of course when spam is not real spam)? That is,
can I simply move/copy such files in a "ham" directory and then train
spamassassin? Will SpamAssassin recognize these files as spam email
messages?
<br>
</blockquote>
This is just a quarantine directory. Amavis won't do anything to it
except when you want to release a mail. If you want Amavis to re-learn a
message you'll have to send it through sa-learn or spamc, moving it to
another directory won't do anything.<br>
<br>
If you want to release Spam you'll have to use amavisd-release: "The
utility <i>amavisd-release</i> tells the amavisd daemon to fetch
a mail from a local quarantine, and send it to MTA through its regular
channels ($notify_method), bypassing re-checking."<span><meta
http-equiv="content-type" content="text/html; charset=UTF-8"> (</span><span><a class="moz-txt-link-freetext" href="http://www.ijs.si/software/amavisd/amavisd-new-docs.html">http://www.ijs.si/software/amavisd/amavisd-new-docs.html</a>).<br>
<br>
Amavisd-new should be your last means of defense (see e.g.
<a class="moz-txt-link-freetext" href="http://www.postfix.org/POSTSCREEN_README.html">http://www.postfix.org/POSTSCREEN_README.html</a>) and 95-99% of all Spam
should already have been blocked. Because of that I just have spam
delivered to a mailbox called Spam. And I use the Dovecot antispam
plugin: when mail is moved out of Spam to another directory (except
Trash) it is retrained as Ham; when mail is from a map to Spam it is
trained as Spam, e.g. my users have to train the filter and decide what
is ham or spam.<br>
<br>
Rgds,<br>
N.<br>
</span></body></html>