LSTOWN-L Archives

LISTSERV List Owners' Forum

LSTOWN-L

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Topic: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
"Duane D. Weaver" <[log in to unmask]>
Tue, 25 Feb 1992 10:27:12 EST
text/plain (105 lines)
Some of the list keywords are documented in LISTKEYW MEMO available
from Listserv.   The following is a description of loopcheck, obtained
and edited from archives of LSTSRV-L.
 
 
(...)    LISTSERV  looks into the body  of the mail, in  an attempt to
recognize  mail that  may be  looping between  a LISTSERV  list and  a
foreign mailer.  Quite a  few checks were put in as  part of 1.5l, but
they don't apply to your 1.5i  system.  The only pre-1.5l checks I can
recall are that the body must not contain either:
 
   1) "From: <list>";
or 2) "Sender: <list>"
 
where <list> contains:     ({x} means "x is optional")
 
   1) {"<"}<listname>"@"<listnode>{">"};
   2) {"<"}<listnode>"!"<listname>{">"} (for UUCP bigots);
or 3) {"<"}<listname>"%"<listnode>{">"}.
 
    Perhaps you might find an incentive to upgrade to 1.5l in the fact
that  it offers  a new  list keyword,  "LoopCheck =  {Full /  NoBody /
NoToCount}, MAXTO(nnn)".  With LoopCheck=NoBody,  the above checks are
suppressed.
 
Ross Patterson
Rutgers University
                default
LOOPCHECK:       'FULL'
 
 
The "Loopcheck=" keyword was implemented  because a number of people were
requesting it so aggressively that I  decided it would be better to spend
10 minutes implementing it than hours answering their endless downpour of
arguments, plus  they were making local  changes to turn it  off on their
server anyway -  with the side-effect that lists where  they did not want
it turned off  were also affected. Anyway to re-enable  loop checking all
you have to do is remove the keyword.
 
  Eric
 
>>> Item number 3023, dated 90/09/10 00:56:58 -- ALL
Date:         Mon, 10 Sep 90 00:56:58 O
Reply-To:     "Forum on LISTSERV release 1.6" <[log in to unmask]>
Sender:       "Forum on LISTSERV release 1.6" <[log in to unmask]>
From:         Eric Thomas <[log in to unmask]>
Subject:      Re: loopcheck
In-Reply-To:  Your message of Sun, 9 Sep 90 18:26:45 EDT
 
The best value for "Loopcheck=" is no  value, ie edit the list header and
remove the string "Loopcheck= Nobody". Any other value apart from "Full",
which  is the  default,  means that  some  or all  of  the loop  checking
routines are  disabled. "Nobody"  means that LISTSERV  is not  allowed to
look  into the  message body  to  decide whether  or not  something is  a
delivery error; it is to rely solely  on the 'Subject:' field. As you can
imagine, this  is a very dangerous  option: it means that  any mailer not
using one of the "standard" subjects known to LISTSERV will cause a loop.
"None" is  even worse,  it tells LISTSERV  that, by  definition, anything
that reaches its reader is NOT a delivery error. These two options should
never be used unless the list is moderated. Even if the list is moderated
they are  not recommended, especially  if the  moderator is not  the list
owner  (ie administrator).  I  actually regret  having implemented  these
options, it really scares me to  think that people are using them without
knowing what they do.
 
"Notocount" means that LISTSERV should not  count the 'To:' fields in the
message body and decide  it is seeing a delivery notice  if there are too
many; "MaxTo(nnn)" orders it to allow up to the specified amount of 'To:'
fields, the default is 10. This  is the one sub-option of the "Loopcheck"
keyword that can be useful - for  digests. The code that counts the 'To:'
field  is here  as a  kind  of "last  resort" action  anyway, to  prevent
infinite loops, but  it won't stop the redistribution of  the first 'nnn'
nastygrams.  If  you disable  this  routine  it  means loops  can  become
infinite, but it  doesn't have any effect on the  actual detection of new
loops.
 
Finally, the latest  addition, "NoCRC", means you don't  want to suppress
duplicate messages on the list, ie if RSCS at FOO becomes crazy and sends
you 200 copies of the same posting, as it happened for LSTSRV-L, INFO-VAX
and many other lists about 6 weeks ago, you will happily redistribute the
200 copies. I added  this option as a kind of  "emergency bypass" in case
the  new  code  that  detects  such duplicates  should  break  and  start
rejecting  all postings.  Fortunately this  did  not happen,  so you  can
forget about it.
 
  Eric
 
>>> Item number 4057, dated 91/10/10 10:53:43 -- ALL
From:         Greg Kroll <[log in to unmask]>
>   LOOPCHECK: (default value FULL)
 
This controls which of the loop-detecting heuristics are used for this
list.  Sorry, I don't have time to compile a complete list of options.
The important ones are FULL and NONE, with meanings that should be
obvious.
 
>>> Item number 4180, dated 91/11/25 06:20:19 -- ALL
From:         Eric Thomas <[log in to unmask]>
Subject:      Changes from release 1.7a to 1.7b
  New       New loop-detector option,  "Loopcheck= NoOrigin", permits the
            list owner  to disable the  check for "known  mailer origins"
            such as  MAILER, POSTMASTER,  ROOT, UUCP,  et al.  Mail whose
            'From:' field  is the  address of the  local mailer  is still
            trapped, but wildcard checks on the mail origin are disabled.

ATOM RSS1 RSS2