rees@apollo.uucp (Jim Rees) (07/09/85)
Recently I've been getting duplicate articles again. There doesn't seem to be any pattern this time. They are rejected here, and I don't know why they aren't getting rejected downstream. We tend to get these in batches of a dozen or so, and then go several days before getting any more. Here is an example. Duplicate article <526@down.FUN> rejected From: honey@down.FUN (Peter Honeyman) Path: wanginst!decvax!mcnc!unc!ulysses!allegra!princeton!down!honey Path: wanginst!linus!decvax!mcnc!unc!ulysses!allegra!princeton!down!honey
davest@daemon.UUCP (Dave Stewart) (07/11/85)
In article <279d59d3.1de6@apollo.uucp> rees@apollo.uucp (Jim Rees) writes: >Recently I've been getting duplicate articles again. There doesn't >seem to be any pattern this time. They are rejected here, and I >don't know why they aren't getting rejected downstream. ... >Path: wanginst!decvax!mcnc!unc!ulysses!allegra!princeton!down!honey >Path: wanginst!linus!decvax!mcnc!unc!ulysses!allegra!princeton!down!honey Note that when a duplicate article is rejected by rnews, it will get retransmitted to downstream sites anyway. The reasoning is that this makes for higher network connectivity if there is a higher degree of article redundancy. Articles are prevented from "looping" around a cycle in the graph because rnews will not forward an article to any site that appears in the "Path" header line. I might add that our local news administrator has added a sys file flag that prevents duplicates from being forwarded along. We use this on "tektronix", our gateway, when forwarding to other Tek sites, because they are a captive part of the network being fed by just one site (tektronix). Send mail to tektronix!rdoty or tektronix!usenet for diffs of rnews code, if you are interested. -- David C. Stewart uucp: tektronix!davest Small Systems Support Group csnet: davest@TEKTRONIX Tektronix, Inc. phone: (503) 627-5418
rees@apollo.uucp (Jim Rees) (07/17/85)
Note that when a duplicate article is rejected by rnews, it will
get retransmitted to downstream sites anyway.
I'm not sure I believe that. Duplicates are rejected around line 327 of
inews.c, just after the call to input(). Broadcast() isn't called until
the very end of insert(), which is called after the duplicates are
rejected. Experiments seem to confirm this analysis.
Maybe my software is out of date, or yours has been hacked up?davest@daemon.UUCP (Dave Stewart) (07/25/85)
Well, it looks like I may have goofed on this one. Sorry - I was going from a faulty memory rather than checking the sources. -- David C. Stewart uucp: tektronix!davest Small Systems Support Group csnet: davest@TEKTRONIX Tektronix, Inc. phone: (503) 627-5418