campbell@maynard.UUCP (Larry Campbell) (06/28/86)
Eighteen out of the last 33 articles posted to mod.comp-soc have arrived
here with material eaten out of the front of the article. The paths involved
are:
Path: maynard!wjh12!husc7a!harvard!caip!ut-sally!pyramid!hplabs!hplabsc!taylor
Path: maynard!wjh12!husc7a!harvard!ut-sally!pyramid!hplabs!hplabsc!taylor
The Message-ID's are:
<337@hplabsc.UUCP> <338@hplabsc.UUCP> <339@hplabsc.UUCP>
<340@hplabsc.UUCP> <341@hplabsc.UUCP> <342@hplabsc.UUCP>
<343@hplabsc.UUCP> <344@hplabsc.UUCP> <345@hplabsc.UUCP>
<346@hplabsc.UUCP> <347@hplabsc.UUCP> <348@hplabsc.UUCP>
<350@hplabsc.UUCP> <351@hplabsc.UUCP> <352@hplabsc.UUCP>
<353@hplabsc.UUCP> <354@hplabsc.UUCP> <355@hplabsc.UUCP>
Would news administrators at the sites along these paths please check their
copies of these articles and post the results?
--
Larry Campbell The Boston Software Works, Inc.
ARPA: campbell%maynard.uucp@harvard.ARPA 120 Fulton Street, Boston MA 02109
UUCP: {alliant,wjh12}!maynard!campbell (617) 367-6846weemba@brahms.BERKELEY.EDU (Matthew P. Wiener) (06/29/86)
>Eighteen out of the last 33 articles posted to mod.comp-soc have arrived >here with material eaten out of the front of the article. The paths involved >are: What do you mean line eater again? I thought it was Whaxre! ucbvax!brahms!weemba Matthew P Wiener/UCB Math Dept/Berkeley CA 94720
taylor@hplabsc.UUCP (Dave Taylor) (06/29/86)
The Problem is Fixed!!! ------------------------- Well, some time has passed and we've managed to figure out what the problem was with the mod.comp-soc postings! As it turns out, I've written a set of scripts that allow me to post news to our news machine (hplabsc) while actually running on my workstation... Unfortunately, it appears that inews/RFA has a bit of a problem when stdin is redirected to a remote machine. The symptoms, strangely enough, are exactly what we were seeing in the group (namely that a piece of the artile was gulped up). My suspicion is that we're sending data faster than the inews (version 2.10.3) program can absorb it, so while it's off fussing with headers, it drops the beginning of the article. NOTE that this doesn't mean that news 2.10.3 is bad or messed up or anything - we're still planning on running it at our site - just that it might exhibit strange behaviour when used in a heavily networked environment. ----- more details for those really interested; on my workstation, I invoke a script that does.. echo "Subject:\\nFrom:\\nReply-To:\\nApproved:\\n" > /tmp/pnews.$$ cat $file >> /tmp/pnews.$$ /usr/hosts/hplabsc /usr/local/lib/news/inews -h < /tmp/pnews.$$ echo article posted [This isn't the real script by any means, but it's indicative of the functionality. Note that the 'inews' program is being run on the remote machine while the file being fed to it is on the local machine...] This is what caused the problems with 2.10.3 inews. To test that it was indeed inews and not the networking software itself, we tried the following test; /usr/hosts/hplabsc cat - < /tmp/pnews.$$ which faithfully listed every character in the file to the screen. Then we tried the following sequence; cp /tmp/pnews.$$ hplabsc:/tmp/rpnews.$$ /usr/hosts/hplabsc "/usr/local/lib/news/inews -h < /tmp/rpnews.$$" This met with success, so that's what my script does now! In fact, I've already reposted all the articles that were trashed... I'd like to than Mike Rodriquez and Greg Fowler for their help in tracking down the problem - without them we might just have folded as a news-site entirely! (Just joking! Don't get so enthused!! *chuckle*) ------ -- Dave Taylor facilitator of mod.comp-soc and mod.conferences and G-d knows what else!
phil@amdcad.UUCP (Phil Ngai) (06/29/86)
It is NOT the line eater. I got mangled stuff through decwrl and
decwrl is pretty clean. As final proof, I got two copies of #21, one
mangled and one unmangled through the same path. (The paths for these
articles include pyramid which is also a clean site. I have other
mangled articles which went amdcad!decwrl!hplabs!hplabsc!taylor.) They
were posted at two different times with two different message ids. I
suspect more notes stupidity at hplabs. I enclose the head of these
two articles.
I suggest we sent core dumps to hplabs!root and the inventor of notes.
==> 133 <==
Article 133 of mod.comp-soc:
Relay-Version: version B 2.10.3 4.3bsd-beta 6/6/85; site amdcad.UUCP
Path: amdcad!decwrl!pyramid!hplabs!hplabsc!taylor
From: taylor@hplabsc.UUCP (Dave Taylor)
Newsgroups: mod.comp-soc
Subject: Computers and Society Digest, #21
Message-ID: <329@hplabsc.UUCP>
Date: 25 Jun 86 18:16:58 GMT
Date-Received: 26 Jun 86 10:57:17 GMT
Reply-To: hplabs!taylor (The Moderator)
Organization: Hewlett Packard Labs, Palo Alto CA
Lines: 159
Approved: taylor@hplabs
--------
Computers and Society Digest, Number 21
Monday, May 12th 1986
Topics of discussion in this issue...
==> 134 <==
Article 134 of mod.comp-soc:
Relay-Version: version B 2.10.3 4.3bsd-beta 6/6/85; site amdcad.UUCP
Path: amdcad!decwrl!pyramid!hplabs!hplabsc!taylor
From: taylor@hplabsc.UUCP (Dave Taylor)
Newsgroups: mod.comp-soc
Subject: Computers and Society Digest, #21
Message-ID: <336@hplabsc.UUCP>
Date: 27 Jun 86 18:53:49 GMT
Date-Received: 28 Jun 86 03:59:13 GMT
Reply-To: hplabs!taylor (The Moderator)
Organization: Hewlett-Packard Laboratories
Lines: 137
Approved: taylor@hplabs
esented
a finalizing and codifying process which encouraged well thought-out
communication. Care needed to be taken, since a single error could
necessitate re-typing the entire letter or document.
There is a hidden risk in the new media, in that they have enabled us to
--
Bring back the phone company!
Phil Ngai +1 408 749 5720
UUCP: {ucbvax,decwrl,ihnp4,allegra}!amdcad!phil
ARPA: amdcad!phil@decwrl.dec.comwunder@hpcea.HP (Walter R. Underwood) (06/30/86)
> They were posted at two different times with two different message ids. > I suspect more notes stupidity at hplabs. I enclose the head of these > two articles. > > I suggest we sent core dumps to hplabs!root and the inventor of notes. > > Phil Ngai +1 408 749 5720 Sorry, Phil. Can't pin this one on Notes. The paths go through News-only machines. I think that the machines at labs have been installing new News software. Anyway, this is not the kind of problem that you get from a News/Notes gateway. The Message-ID's are all wrong for that. Sending core dumps? Was this cross-posted to net.politics.terror? Walter Underwood
geoff@suneast.uucp (Geoff Arnold) (06/30/86)
They were smashed here, too. A representative sample: >Article 36 of 41: <350@hplabsc.UUCP> >Subject: License Plate Risks >From: taylor@hplabsc.UUCP (Dave Taylor) >Path: sunne!security!linus!philabs!prls!pyramid!hplabs!hplabsc!taylor >Organization: Hewlett-Packard Laboratories >Newsgroups: mod.comp-soc >Date: Fri, 27-Jun-86 18:37:55 EDT >Reply-To: hplabs!price@src.DEC.COM >more tickets, from all over the state [2500 in >all!]. It seems that when a police officer writes a parking ticket for a >car with no license plates, he writes ``NO PLATE'' on the ticket. > >Our friend took his problem to the DMV, which informed him that he should >change his plates. [etc.] -- "To disclaim, or not to disclaim... " <<<<<< Geoff Arnold, Sun Microsystems Inc. (East Coast Division) >>>>>> SnailMail: One Cranberry Hill, Lexington, MA 02173; 617-863-8870 x136 UUCP: {hplabs,ihnp4,nsc,pyramid,decwrl}!sun!suneast!geoff
geoff@suneast.uucp (Geoff Arnold) (06/30/86)
As I meant to continue, the problem is obviously localizable to:
pyramid!hplabs!hplabsc!taylor
--
"To disclaim, or not to disclaim... "
<<<<<< Geoff Arnold, Sun Microsystems Inc. (East Coast Division) >>>>>>
SnailMail: One Cranberry Hill, Lexington, MA 02173; 617-863-8870 x136
UUCP: {hplabs,ihnp4,nsc,pyramid,decwrl}!sun!suneast!geoffphil@amdcad.UUCP (Phil Ngai) (06/30/86)
In article <12173@amdcad.UUCP> phil@amdcad.UUCP (Phil Ngai) writes: > >I suggest we sent core dumps to hplabs!root and the inventor of notes. > Just in case it's not clear, the above was not meant to be taken literally. -- Bring back the phone company! Phil Ngai +1 408 749 5720 UUCP: {ucbvax,decwrl,ihnp4,allegra}!amdcad!phil ARPA: amdcad!phil@decwrl.dec.com
loverso@sunybcs.UUCP (John Robert LoVerso) (07/02/86)
Well, it definitely happened somewhere in HP Labs! Some of mod.comp-soc here went thru Path: sunybcs!rocksanne!rocksvax!parcvax!hplabs!hplabsc!taylor and the only common portions with other posted paths is from hplabs on down...
jerry@oliveb.UUCP (Jerry Aguirre) (07/02/86)
In article <21500004@hplabsc.UUCP> taylor@hplabsc.UUCP (Dave Taylor) writes: > This met with success, so that's what my script does now! In fact, I've > already reposted all the articles that were trashed... And did you bother to cancel the trashed articles you sent out into the net?
rees@apollo.uucp (Jim Rees) (07/03/86)
While I can claim to be the most rabid anti-notes admin on the net, let me point out that notes does very well at what it was intended to do, and that the author, Ray Essick, is a reasonable guy. He's even a smart guy (one of my highest compliments). The problem lies in the notes to news interface. Just as we were bitten by the line-eater having been included on the 4.2bsd tape, the notes folks were bitten by the obsolete A-news interface being included on that same tape. There is a more modern B-news interface that works well, but as long as there are 4.2 tapes out there, we will be plagued by the line-eater and by orphaned responses.
rs@mirror.UUCP (07/03/86)
>I suspect more notes stupidity.
"Orphaned responses" and dumping old stuff I can accept.
This is the first time I have ever heard anyone blame
notes for mangling articles -- I though that was purely
a news problem.
--
Rich $alz {mit-eddie, ihnp4!inmet, wjh12, cca, datacube}!mirror!rs
Mirror Systems 2067 Massachusetts Avenue Cambridge, MA, 02140
Telephone: 6,176,610,777
"Hi, mom!"rees@apollo.uucp (Jim Rees) (07/03/86)
This is what caused the problems with 2.10.3 inews. To test that it was
indeed inews and not the networking software itself, we tried the following
test;
/usr/hosts/hplabsc cat - < /tmp/pnews.$$
That doesn't mean diddly. Cat doesn't try to seek on its input; inews
does. It looks to me like your network software is to blame.
(I will resist putting in a plug for my company's network software here.)spaf@gatech.CSNET (Gene Spafford) (07/04/86)
In article <12173@amdcad.UUCP> phil@amdcad.UUCP (Phil Ngai) writes: >I suggest we sent core dumps to hplabs!root and the inventor of notes. > hplabs DOES NOT run notes!! Don't dump on Greg and Mike! They run real news! hplabs is a major backbone site and they run the "good stuff" there, and Greg does a stellar job keeping things running so please don't abuse hplabs. However, many of the HP sites hplabs talks to use "notes" and it could be that one of them is to blame. And if you mail core dumps to the inventor of notes, please don't mail 'em through gatech! -- Gene Spafford Software Engineering Research Center (SERC), Georgia Tech, Atlanta GA 30332 CSNet: Spaf @ GATech ARPA: Spaf@Gatech.GATECH.EDU uucp: ...!{akgua,decvax,hplabs,ihnp4,linus,seismo,ulysses}!gatech!spaf
kre@munnari.OZ (Robert Elz) (07/06/86)
> This is the first time I have ever heard anyone blame > notes for mangling articles -- I though that was purely > a news problem. What about the mangling that happens when the entire rear end of a news article is trashed... (those news articles > 64Kb). This might be "fixed in the latest release" but it's certainly been (and still is) out there.
jsdy@hadron.UUCP (Joseph S. D. Yao) (07/17/86)
In article <1164@munnari.OZ> kre@munnari.OZ (Robert Elz) writes: >> This is the first time I have ever heard anyone blame >> notes for mangling articles ... >What about the mangling that happens when the entire rear >end of a news article is trashed... >(those news articles > 64Kb). Isn't notes doing us a service thereby? I don't know that we really need articles > 64Kb. The exceptions, I think, would be mod.sources and MAYBE net.sources; but since the custom there is to break up source archives anyway, that shouldn't be as much of a problem. (Just leave 1K for header ;-).) -- Joe Yao hadron!jsdy@seismo.{CSS.GOV,ARPA,UUCP} jsdy@hadron.COM (not yet domainised)
rs@mirror.UUCP (07/31/86)
Older versions (at least pre-1.7) of notes didn't allow articles >64K;
this problem has been fixed for quite some time. There are mail systems
out there that don't allow messages to be more than 64K, too. I agree
with Joe that this is also a good idea vis-a-vis human limits, too.
I make sure that no individual mod.sources (and soon, mod.sources.games)
posting is greater than 55K (sans headers). Of course, this does have
its drawbacks: the backlog is almost two weeks right now...
----
Rich $alz {mit-eddie, ihnp4, wjh12, cca, cbosgd, seismo}!mirror!rs
Mirror Systems 2067 Massachusetts Avenue Cambridge, MA 02140
Telephone: 617-661-0777
"Hi, mom!"