Ross Patterson <A024012@RUTVM1>
Wed, 22 Apr 87 10:30:23 EDT
|
Hank, I'd be very interested in how DATABASE's code would have
detected the loop. The loop succeeded in evading all of LISTSERV's
loop detection code, which is pretty good in it's own right. LISTSERV
detects several conditions which indicate a potential loop, including:
1) From: is the local mailer;
2) From: contains one of the words MAILER, SYSTEM, and DAEMON;
3) From: is the mailer listed in XMAILER NAMES for the source node;
4) From: is the mailer listed in DOMAIN NAMES for the source domain;
5) Sender: contains one of the strings <listname>@<listnode>,
<listname>%<listnode>, and <listnode>!<listname>;
6) Subject: is UNDELIVERED MAIL or RETURNED NETWORK MAIL.
In addition, any user sending 10 invalid commands in a row to
LISTSERV will have access to LISTSERV suspended, until another user
(any other user) tells LISTSERV to re-instate access.
A sample of the mail in question was (after distribution by the list):
------------------ Start of included file UUCP MAIL -----------------
Received: from BITNIC(MAILER) by RUTVM1 (Mailer X1.23b) id 7768;
Wed, 22 Apr 87 02:32:07 EDT
Received: by BITNIC (Mailer X1.23b) id 6024; Wed, 22 Apr 87 02:33:28 EDT
Date: Tue, 21 Apr 87 09:43:01 PDT
Reply-To: UNIX-to-UNIX Copy <CMCL2!RUTGERS!SRI-SPAM!CSI!UUCP>
Sender: BITNIC IBM-NETS List <IBM-NETS@BITNIC>
From: UNIX-to-UNIX Copy <CMCL2!RUTGERS!SRI-SPAM!CSI!UUCP>
Subject: uuxqt cmd (rmail ghere) status (signal 0, exit 67)
To: Ross Patterson <A024012@RUTVM1>
------------------ End of included file UUCP MAIL -----------------
Please note that the Sender: was added by LISTSERV, and thus
couldn't trip test which path through usenet the file took. The only
tip off was the Subject:, and that required some knowledge of esoteric
bits of Unix to understand. Remember, the first alarm was issued only
after the list had received several of these.
So, how 'bout it?
Ross Patterson
Rutgers University
|
|
|