smk@linus.UUCP (Steven M. Kramer) (06/21/83)
The following are two articles we received. The first appeared via seismo!harpo!decvax!genrad and the second thru seismo!philabs. Can anyone limit the problem now? Which sites look suspect? From solomon@uwvax.UUCP Sat Jun 18 15:17:22 1983 Relay-Version: version B 2.10 5/3/83; site linus.UUCP Path: linus!genrad!decvax!harpo!seismo!uwvax!solomon From: solomon@uwvax.UUCP Newsgroups: net.news Subject: Re: A gripe about 2.10 readnews - (nf) Message-ID: <222@crystal.UUCP> Date: Sat, 18-Jun-83 15:17:22 EDT Article-I.D.: crystal.222 Posted: Sat Jun 18 15:17:22 1983 Date-Received: Sun, 19-Jun-83 07:14:19 EDT Lines: 72 Yep, if PAGER is /usr/ucb/more, having it read its input from the news article itself rather than a pipe is a real win. It's even better if ... From solomon@uwvax.UUCP Sat Jun 18 15:17:22 1983 Relay-Version: version B 2.10 5/3/83; site linus.UUCP Posting-Version: version B 2.10 5/3/83; site crystal.ARPA Path: linus!philabs!seismo!uwvax!solomon From: solomon@uwvax.UUCP Newsgroups: net.news Subject: Re: A gripe about 2.10 readnews - (nf) Message-ID: <222@crystal.ARPA> Date: Sat, 18-Jun-83 15:17:22 EDT Article-I.D.: crystal.222 Posted: Sat Jun 18 15:17:22 1983 Date-Received: Sun, 19-Jun-83 15:30:42 EDT References: <2248@uiucdcs.UUCP> Organization: U of Wisconsin CS Dept Lines: 72 Yep, if PAGER is /usr/ucb/more, having it read its input from the news article itself rather than a pipe is a real win. It's even better if ... -- --steve kramer {allegra,genrad,ihnp4,utzoo,philabs,uw-beaver}!linus!smk (UUCP) linus!smk@mitre-bedford (ARPA)
jim@uw-beaver.UUCP (06/25/83)
This dup was caused by someone dropping the message-id. Crystal is using ".ARPA" as their domain, so any time the message-id is dropped, you get a dup. I know the article was ok when it left seismo, so I would guess the problem is at philabs. I also got a duplicate here, and the message-id in this case was definitely dropped by microsoft. I'm afraid we'll have to put up with dups until sites like these get around to updating their software, or until we decide to drop message-ids and go back to article-ids.
chris@grkermit.UUCP (Chris Hibbert) (06/27/83)
... And uw-beaver!jim's article got here with duplicate Message-ID's. Both copies had "Article-I.D.: uw-beave.679", but there were different Message-ID's: One was "Message-ID: <679@uw-beaver>", and the other was "<679@uw-beave.UUCP>." The second clearly reconstructed from the article-ID, which had gotten truncated. The paths they used were: "grkermit!genrad!linus!wivax!decvax!harpo!floyd!vax135!cornell!uw-beaver!jim" for <679@uw-beaver> and "grkermit!genrad!decvax!harpo!floyd!vax135!cornell!uw-beaver!jim" for <679@uw-beave.UUCP> Maybe decvax<==>genrad is at fault? Genrad did recently change over to news 2.10, and I don't think decvax has.
jim@mcvax.UUCP (06/28/83)
Our only news-feed from "Over There" is through philabs, and we are getting a lot of duplicate articles. I have had a prod around, and it looks like genrad<=>decvax to me too: diff 454 457 2,3c2 < Posting-Version: version B 2.10 5/3/83; site uw-beaver < Path: mcvax!philabs!linus!wivax!decvax!harpo!floyd!vax135!cornell!uw-beaver!jim --- > Path: mcvax!philabs!linus!genrad!decvax!harpo!floyd!vax135!cornell!uw-beaver!jim 7c6 < Message-ID: <679@uw-beaver> --- > Message-ID: <679@uw-beave.UUCP> 11,13c10 < Date-Received: Mon, 27-Jun-83 23:37:40 (DT) < References: <26911@linus.UUCP> < Organization: U of Washington Computer Science --- > Date-Received: Mon, 27-Jun-83 23:50:16 (DT) Other duplicate messages that we receive always have genrad in them, but others have decvax (without genrad) and are not duplicated. Note that the Posting-Version, References and Organisation fields are also gone. We had a similar problem when we changed to 2.10, as a couple of sites we talk to "Over Here" were running VERY old 2.9 versions, which didn't understand all the fields (like they threw the Path field away and used the From field, so we got ALL the news back!). They also stripped the above fields out. Does that help? Jim McKie ....decvax!mcvax!jim ...philabs!mcvax!jim
smk@linus.UUCP (Steven M. Kramer) (06/30/83)
Has anyone considered that there may still be a bug in the news code? It looks like each of us has checked out the code (except decvax, but I'm not accusing them) that has been identified along the routes. The information to go on is so fragile, it's hard to debug. -- --steve kramer {allegra,genrad,ihnp4,utzoo,philabs,uw-beaver}!linus!smk (UUCP) linus!smk@mitre-bedford (ARPA)
mark@cbosgd.UUCP (07/01/83)
One possible cause of a dropped message id is if a site sends news to another site in A format. I just discovered that decvax, while running 2.10 and having mostly 2.10 neighbors, is sending news to several sites in A format. A format will lose much information, including the message ID. Could I ask news admins to double check their sys files to make sure that A format isn't being used, unless their neighbor is really running A news? (A+ understands B format.) I don't know of many sites on the net still running A (microsoft is the only one I can think of, and I'm not sure about them), so A's are unlikely to be needed anymore. Mark Horton
rob@genrad.UUCP (Rob Wood) (07/02/83)
There have been a number of people suggesting that genrad is at the root of the multiple copies. This in not correct. I appreciate everyone trying to find out what is happening, but while my site does pass both articles, it is because someone has already truncated the r from uw-beaver in one of the articles, so our news thinks that they are different articles. We are running 2.10. We have exactly the same software as most of the New England sites. Our link to decvax is a 2.10 link. I hope that someone can find the problem, as about 50% of our articles received are duplicates and thus our spool directory is overutilized. Rob Wood (decvax!genrad!rob)
jim@uw-beaver.UUCP (07/06/83)
I got a call this morning from Dan at philabs who assures me that he isn't dropping the message-id. He is running news 2.10 with the header.c fixes. I'm out of ideas now, because I was pretty sure that all the other sites in that path were OK. My apologies for falsely accusing philabs. They apparently take at least as much pride as I do in running bug-free software, and were justifiably annoyed at being publicly maligned.