Cancels and NoCems (was: Re: ninpaths updated)

Rich Skrenta skrenta at textport.net
Fri Jan 12 06:12:22 UTC 2001


> ftp://ftp.cistron.nl/pub/people/miquels/news/perl-nocem

Thank you for posting this link.  The link to the perl nocem-on-spool
on the nocem registry is broken, and c-nocem doesn't build cleanly
with inn-2.3.0.


I have a question about what most INN admins do about honoring
regular cancels vs. NoCems.

One of my users recently complained that soc.history.living on my site
was missing all of its articles.  I checked some other nntp sites and
sure enough, mine had all apparently been deleted by cancels (the numbers
in the active file indicated that posts had once been in that group...)
I set an archive feed up to log any cancels coming by for that group
so I could see what was going on (I had some fantasy about being able
to alias out the bad guy) but haven't spotted any new rogue ones yet.

Then today I noticed that some joker had toasted all of the posts in
news.announce.newgroups.  I assume this is a regular news.groupie thing
to do when you don't like the way the discussion is going.

So my conclusion is that today we live with waves of rogue cancels
routinely washing across Usenet, improperly deleting posts.  The only
alternative for the newsadmin seems to be to disable honoring regular
cancels in INN.  However, I didn't want to miss the beneficial
cancels which target spam, so I figured I needed to install and run
nocem-on-spool software to process pgp-signed cancels from alt.nocem.misc
and news.lists.filters.  (I already run cleanfeed, but it doesn't catch
all spam of course.)

Question #1:  Is this what all of you do, or is there another solution?
I notice that Newsguy's groups look reasonable, and even Best.com still
has its news.announce.newgroups messages, and most of soc.living.history.
Is it standard practice now to turn off regular cancels and install
some form of nocem-on-spool?

Question #2:  I turned off honoring of cancels by setting verifycancels
to true in inn.conf and changing ARTcancelverify() to always return NULL.
Is there some way to do this through configuration files rather than
hacking the source?

  *  *  *

Optional question #3:  The nocem mailing list doesn't seem to exist
anymore at nocem-l-request at mail.mpcs.com.  Does anyone know of a new one?

Optional question #4:  The Cancel-Key: header proposal at
http://www.cm.org/about-cancels.html seems to be a good one:

	For every posted message there is a "Cancel-Key" which is the
	message-id of the message hashed with a secret password. The MD5
	of the cancel-key is the "Cancel-Challenge" which is posted
	as a header in every post you make. To cancel that post,
	the cancel message must have a copy of the Cancel-Key in the
	headers.  [...]

If we come to a point where most news software ignores cancels, are there
plans to go forward with something like this in the transport software,
and encourage news client maintainers to add support?




More information about the inn-workers mailing list