Path: utzoo!utgpu!attcan!lsuc!eci386!clewis
From: clewis@eci386.uucp (Chris Lewis)
Newsgroups: news.admin
Subject: Re: News delivery problems - old news again
Message-ID: <1989Aug8.153447.14095@eci386.uucp>
Date: 8 Aug 89 15:34:47 GMT
References: <43675@bbn.COM> <651@vector.Dallas.TX.US> <1989Aug3.180304.6252@eci386.uucp> <653@vector.Dallas.TX.US>
Reply-To: clewis@eci386.UUCP (Chris Lewis)
Organization: R. H. Lathwell Associates: Elegant Communications, Inc.
Lines: 59

In article <653@vector.Dallas.TX.US> chip@vector.Dallas.TX.US (Chip Rosenthal) writes:
>clewis@eci386.UUCP (Chris Lewis) writes:
>>[re: problems with running do-cancel under C-news]
>>For the moment, I've created a version of inews that doesn't fork,
>>though I expect it to take several hours to run the several hundred
>>cancels.
>
>Yoicks.  Is that really true?  It took me about 5 mins to run the thing
>here under B2.11 on a 386.

Quite true - running do-cancel with C news inews modified to not spawn in the
background took approximately 3 hours, and (very approx) about 6,000 
processes for 600 some-odd articles ....  (the 6,000 is very approximate - 
other activity was going on at the same time and $$ went up by about 12,000
during that time).  (16Mhz 386, 386/ix 1.0.6, pre-comp.sources.unix C news.)

[In contrast, with an unmodified inews, the system hung up almost completely
in 30-40 seconds with lots of simultaneous inews invocations....  Fortunately,
I was able to kill it ...]

As Henry points out though, this is hardly unexpected.  Inews is intended for
human-generated messages.  If you pump these articles into relaynews directly,
I expect it'll only take a few minutes.  David Kuder (david@indetech) sent
me a modified version of do-cancel that does just this, but I didn't need
to run it because I had already run it with a modified inews.  I would have
posted it, but I haven't asked David for permission.  David, if you're reading
this, I suggest that you post it.  Officially via comp.sources.unix or ...misc
would be nice.  Or, even integrated with C news or somthin.

[Incidentally, we're running a version of C news predating the 
comp.sources.unix distribution and it doesn't appear to support the -W
flag that Henry mentions.  One of these days we'll upgrade]

On a 3b1, with B2.11 (patch level 14) it took approximately 10 minutes.

>BTW, in answer to those who are wondering why these messages weren't
>rejected by history, the offending site stepped on the "Date:" header.
>I've been told that TELECOM Digest issue 202 was reposted to comp.dcom.telecom
>in this slew of messages.  Well, I sent out that issue on June 18.

I don't think so.  History don't know nothing 'bout dates.  On our site,
the path that these articles took was sufficiently wierd to suggest that
some machine held on to them so long that when kicked back out into
their outgoing feed, that many sites had expired the message-ids.
Then, these articles wandered hither-and-yon seeking sites that had also
expired them.

However,

B-news, I believe, has a facility for rejecting articles that are over a 
certain age, but the default's rather high (3 weeks if I remember correctly).
As a suggestion, people might consider turning this down to a week or so -
anything that takes that long to get to you is likely to be a system
vomiting its spool rather than true slow delivery.
-- 
Chris Lewis, R.H. Lathwell & Associates: Elegant Communications Inc.
UUCP: {uunet!mnetor, utcsri!utzoo}!lsuc!eci386!clewis
Phone: (416)-595-5425