bob@vpk3.UUCP (Bob Kyryliuk) (04/25/89)
In article <1989Apr22.142911.1685@ziebmef.uucp> cks@ziebmef.UUCP (Chris Siebenmann) writes: >In article <429@becker.UUCP> bdb@becker.UUCP (Bruce Becker) writes: >| To add insult to injury, there seems to be a recirculation >| of week-old to 10-day-old news in parts of the local feeds. >| Since some news implementations expire history entries at >| the same time as articles, these won't be seen as duplicates. > > I suspect that we're getting close to having loops just over the >critical time, perhaps due to all the sites that have converted to >Cnews. I've seen a lot of duplicates in tor.news.stats and various >other groups; the late-arriving duplicates all have rather long paths. >Relatively central sites still running Bnews may want to up their >history retention times for tor.* groups to try and choke this off. I've noticed a small number of duplicate articles on attcan in tor.news.* that actually were processed as separate articles ... It can't be explained by expire times, since we don't expire tor.* for 7 days, and yet some of the articles arrived on the same day, or a day apart. The usual check of the Message-ID showed that they were identical. I've never heard of this problem before ... and nobody has reported the problem to me internally on any but the tor.news.* groups. We're still running B-News 2.11.14 (although we'll move up to C-News when it becomes available). Here's a sample duplicate article from mark@sickkids: article headers (note identical Message-ID's): 148:Path: attcan!utgpu!lsuc!sickkids!mark 148:Message-ID: <259@sickkids.UUCP> 151:Path: attcan!utzoo!yunexus!maccs!lsuc!sickkids!mark 151:Message-ID: <259@sickkids.UUCP> history.d entries (note that they arrived the same day): <259@sickkids.UUCP> 04/22/89 02:23 tor.news/148 <259@sickkids.UUCP> 04/22/89 13:02 tor.news/151 As we often get numerous duplicate articles because of our multiple feeds (ie 1000-10000 per day :-) which are handled correctly, it seems odd that a few of the tor.news.* articles are slipping in past the B-News software. Anyone have any clues? Regards, Bob -- ------------------------------------------------------------------------------- ____ _______ _____ _______ ------- Bob Kyryliuk / __ \ |__ __| / _ \ |__ __| -====------ Manager, Special Projects | (__) | | | \ \ \_\ | | -======------ National Technical Support | __ | | | / \ __ | | --====------- AT&T Canada Inc. | | | | | | | (\ / / | | ----------- |_| |_| |_| \_____/ |_| ------- (416) 756-5183 AT&T MAIL: ..!attmail!bkyryliuk UUCP:{att|cuuxb|uunet|utzoo}!attcan!bob
bdb@becker.UUCP (Bruce Becker) (04/25/89)
In article <4777@vpk3.UUCP> bob@attcan.UUCP (Bob Kyryliuk) writes: | [...] |Here's a sample duplicate article from mark@sickkids: | | article headers (note identical Message-ID's): | | 148:Path: attcan!utgpu!lsuc!sickkids!mark | 148:Message-ID: <259@sickkids.UUCP> | | 151:Path: attcan!utzoo!yunexus!maccs!lsuc!sickkids!mark | 151:Message-ID: <259@sickkids.UUCP> | | history.d entries (note that they arrived the same day): | | <259@sickkids.UUCP> 04/22/89 02:23 tor.news/148 | <259@sickkids.UUCP> 04/22/89 13:02 tor.news/151 | |As we often get numerous duplicate articles because of our multiple feeds |(ie 1000-10000 per day :-) which are handled correctly, it seems odd that |a few of the tor.news.* articles are slipping in past the B-News software. | |Anyone have any clues? I have duplicates consistently in my news, and not just in "tor" either. The reason why I ended up complaining was that a lot of the recirculated articles *ought* to have been rejected but weren't for some reason. I had put the problem down to a possible flaw in the port of dbm to this machine (which theory I have yet to test), but "history.d" sounds like the non-dbm type of history mechanism, so there may be a flaw in the logic of the B News history system (version 4.3.1.4). I have put off dealing with this problem until I upgrade my News software, which I'm told will be available RSN... Cheers, -- __ Bruce Becker Toronto, Ont. w \cc/ Internet: bdb@becker.UUCP, bruce@gpu.utcs.toronto.edu `/v/-e BitNet: BECKER@HUMBER.BITNET _< >_ "A divine sparc turned me from nextrophilia" - NoNuke of the North