<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=utf-8">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<p><br>
So I recently did a bunch of upgrades when I discovered that my
DSPAM integration needed some realignment.
<br>
<br>
I currently have the following
<br>
<br>
new spam-scanners setting in my amavisd.conf
<br>
<br>
@spam_scanners = (
<br>
['DSPAM', 'Amavis::SpamControl::ExtProg', $dspam,
<br>
[ qw(--stdout --process --deliver=innocent,spam --mode=tum
<br>
--tokenizer=chained,noise --user), $daemon_user ],
<br>
learn_ham => [ qw(--class=innocent --source=error --user),
$daemon_user ],
<br>
learn_spam => [ qw(--class=spam --source=error --user),
$daemon_user ],
<br>
mail_body_size_limit => 64000, score_factor => 0.1,
<br>
],
<br>
['SpamAssassin', 'Amavis::SpamControl::SpamAssassin' ]
<br>
);
<br>
<br>
Which is getting me the following log entries
<br>
<br>
Feb 8 16:46:54 death amavis[12497]: (12497-08) Checking:
PoWVVtEZIQzZ [195.245.230.171] <a class="moz-txt-link-rfc2396E"
href="mailto:Liam.Mannion@rullion.co.uk"><Liam.Mannion@rullion.co.uk></a>
-> <a class="moz-txt-link-rfc2396E"
href="mailto:barry.wimlett@endless.co.uk"><barry.wimlett@endless.co.uk></a>
<br>
Feb 8 16:46:54 death amavis[12497]: (12497-08) p002 1
Content-Type: multipart/mixed
<br>
Feb 8 16:46:54 death amavis[12497]: (12497-08) p001 1/1
Content-Type: text/html, 7bit, size: 4141, SHA1 digest:
ddebd686fcf56a0c63aa4be9e1a86638be3e790f
<br>
Feb 8 16:46:55 death amavis[12497]: (12497-08) DSPAM result:
Innocent\r, score=-0.100, sig=589b4b7f171141540518149\r
<br>
Feb 8 16:46:58 death amavis[12497]: (12497-08) SpamControl:
scanner DSPAM, auto-learn as ham / -0.921 (was: innocent\r /
-0.100)
<br>
Feb 8 16:46:58 death dspam[17118]: Unable to find a valid
signature. Aborting.
<br>
Feb 8 16:46:58 death dspam[17118]: process_message returned error
-5. dropping message.
<br>
<br>
I <b class="moz-txt-star"><span class="moz-txt-tag">*</span>suspect<span
class="moz-txt-tag">*</span></b> that the extraneous '\rs' are
confusing amavis string recognition for 'ham vs spam' 'spam vs
innocent'; and then also causing the signature to be unrecognised.
<br>
<br>
If I manually enter a dspam command-line with source=error
something like this
<br>
dspam --stdout --debug --user vscan --source=error
--signature=5899e56780281410093335 --class=innocent
<br>
<br>
It all seems to work... I also notice that other mailing-list
users also have the extra '/r's in the postings
<br>
<a class="moz-txt-link-freetext"
href="https://groups.google.com/forum/#%21searchin/mailing.unix.amavis-user/dspam%7Csort:relevance/mailing.unix.amavis-user/tFKiT3HzqNQ/FNxpWTCQ7XUJ">https://groups.google.com/forum/#!searchin/mailing.unix.amavis-user/dspam%7Csort:relevance/mailing.unix.amavis-user/tFKiT3HzqNQ/FNxpWTCQ7XUJ</a>
<br>
<br>
Also I notice that the spam_scanners has zero means of instructing
the learn_ham or lean_spam what the signature is. Is there a
syntax I can use to wrapper --signature=X-DSPAM-Signature to make
it meaningful ?
<br>
<br>
thanks
<br>
<br>
Barry
</p>
</body>
</html>