schwager@m.cs.uiuc.edu (Michael Schwager) (04/24/91)
Hi, Today is April 23. I have some files like these: 57837 39 -rw-rw-r-- 1 news notes 39906 Apr 15 09:34 ./1179 57675 30 -rw-rw-r-- 1 news notes 30310 Apr 15 09:35 ./1180 57678 24 -rw-rw-r-- 1 news notes 24299 Apr 15 09:35 ./1181 57681 1 -rw-rw-r-- 1 news notes 465 Apr 15 09:40 ./1182 57691 1 -rw-rw-r-- 1 news notes 468 Apr 15 09:40 ./1183 57762 30 -rw-rw-r-- 1 news notes 30314 Apr 15 09:41 ./1184 57838 30 -rw-rw-r-- 1 news notes 30308 Apr 15 09:41 ./1185 57839 30 -rw-rw-r-- 1 news notes 30315 Apr 15 09:41 ./1186 57840 30 -rw-rw-r-- 1 news notes 30309 Apr 15 09:42 ./1187 (I even have some that are older!) They are in a local newsgroup called uiuc.cs.maint.nfs . Looking at the explist, I show: uiuc.cs.maint.nfs x 4-7 - Now, from what I know, this means that the articles will be removed from our disk in 4 days, but as you can see it's not happening. Can anyone explain why? -Mike P.S. here is a sample header from a file, specifically file 1183 listed above: From: root@cs.uiuc.edu Date: 14 Apr 91 23:43 CDT Newsgroups: uiuc.cs.maint.nfs Subject: Outbound a.cs.uiuc.edu:/usr/dcs/sun Message-ID: <78007042@m.cs.uiuc.edu> Path: m.cs.uiuc.edu!m.cs.uiuc.edu!seefromline Nf-ID: #N:m.cs.uiuc.edu:78007042:000:167 Nf-From: cs.uiuc.edu!root Apr 14 23:43:00 1991 (text here) -Mike Schwager | Machine: Amiga 500, 3 MB RAM/30 HD INTERNET:schwager@cs.uiuc.edu | Bike: '83 Kawasaki KZ750 LTD UUCP:{uunet|convex|pur-ee}!uiucdcs!schwager| Band: Poi Dog Pondering // BITNET:schwager@mike.cs.uiuc.edu | Hero: Robert Bly \\ // University of Illinois, Dept. of Comp. Sci.| DoD: #0301 \X/Amiga
henry@zoo.toronto.edu (Henry Spencer) (04/24/91)
In article <1991Apr23.170825.8303@m.cs.uiuc.edu> schwager@m.cs.uiuc.edu (Michael Schwager) writes: >Now, from what I know, this means that the articles will be removed from >our disk in 4 days, but as you can see it's not happening. Can anyone >explain why? The first thing you need to check is whether expire is, in fact, being run. Is there a history.o file in /usr/lib/news? How old is it? (It should date to the last time expire was run.) Is anyone reading the mail to "usenet" (or whoever you specified as the destination for trouble reports)? -- And the bean-counter replied, | Henry Spencer @ U of Toronto Zoology "beans are more important". | henry@zoo.toronto.edu utzoo!henry
schwager@m.cs.uiuc.edu (Michael Schwager) (04/24/91)
henry@zoo.toronto.edu (Henry Spencer) writes: >In article <1991Apr23.170825.8303@m.cs.uiuc.edu> schwager@m.cs.uiuc.edu (Michael Schwager) writes: >>Now, from what I know, this means that the articles will be removed from >>our disk in 4 days, but as you can see it's not happening. Can anyone >>explain why? >The first thing you need to check is whether expire is, in fact, being >run. Is there a history.o file in /usr/lib/news? Yes >How old is it? (It >should date to the last time expire was run.) It's dated as of around 2 am. I start the expire at 1am. I have it send me mail every night when expire is run; I've been getting it regularly. Of course, this only means that cron has fired it up. >Is anyone reading the >mail to "usenet" (or whoever you specified as the destination for trouble >reports)? Yup, I am. What should I look for? -Mike
henry@zoo.toronto.edu (Henry Spencer) (04/25/91)
In article <1991Apr24.160705.29971@m.cs.uiuc.edu> schwager@m.cs.uiuc.edu (Michael Schwager) writes: >>...Is there a history.o file in /usr/lib/news? > Yes [and the date is right] Okay, sounds like expire is running. Also, you're not getting any bizarre complaints in mail. So the next question is: are those persistent articles in your history file? The newshist command can tell you that. Actually, another question is, are you *sure* that the explist line for that newsgroup is in fact the one that's being applied? An earlier line which also covers that group will cause its line to be ignored. A look at your whole explist, not just that line, is in order. -- And the bean-counter replied, | Henry Spencer @ U of Toronto Zoology "beans are more important". | henry@zoo.toronto.edu utzoo!henry
schwager@m.cs.uiuc.edu (Michael Schwager) (04/25/91)
henry@zoo.toronto.edu (Henry Spencer) writes: >In article <1991Apr24.160705.29971@m.cs.uiuc.edu> schwager@m.cs.uiuc.edu (Michael Schwager) writes: >>>...Is there a history.o file in /usr/lib/news? >> Yes [and the date is right] >Okay, sounds like expire is running. Also, you're not getting any bizarre >complaints in mail. So the next question is: are those persistent articles >in your history file? The newshist command can tell you that. >Actually, another question is, are you *sure* that the explist line for that >newsgroup is in fact the one that's being applied? An earlier line which >also covers that group will cause its line to be ignored. A look at your >whole explist, not just that line, is in order. Well, unfortunately I went and cleaned up a files throughout my directory, so I can't do a newshist. But if the problem persists, I'm sure it will turn up again. I went and cleaned up my explist file considerably. I'm somewhat embarrassed to admit that I didn't catch the part about the line ordering. So this could very well take care of it. -Mike
molenda@s1.msi.umn.edu (Jason Molenda) (04/25/91)
henry@zoo.toronto.edu (Henry Spencer) writes: > An earlier line which >also covers that group will cause its line to be ignored. I've known about this for a long time and I know it is documented and everything, but am I the only person who thinks it's a huge hole inviting mistakes by new and old sysadmins alike? I don't know much (read: nothing) about the internals of expire, so I have no idea how hard it would be to make the ordering of the lines non-important, but it sure would be nice. Am I missing something obvious here, or shouldn't I be able to say local 5 x - local.test 1 x - I mean, I know this won't work right with the current expire, but would it be that hard to make it work the way I expect it to? -- Jason Molenda, Tech Support, Iris & News Admin, Minnesota Supercomputer Inst molenda@msi.umn.edu || "You can tune a piano but you can't tuna fish."
schwager@m.cs.uiuc.edu (Michael Schwager) (04/25/91)
molenda@s1.msi.umn.edu (Jason Molenda) writes: >henry@zoo.toronto.edu (Henry Spencer) writes: >> An earlier line which >>also covers that group will cause its line to be ignored. >I've known about this for a long time and I know it is documented and >everything, but am I the only person who thinks it's a huge hole inviting >mistakes by new and old sysadmins alike? Indeed, if I may make a comment regarding News (we ran notes up until December of last year): Seems like there are all kinds of pitfalls, troubles, and caveats one can enter when bringing News up. For example, one must make more inodes than the default on our News partitions, because News stores every blasted article in a seperate file. I'm guessing that that's the way it is elsewhere, too. News is so big and cantankerous, that it's so dang confusing to figure out problems. It's taken me a few months to get it down, and even now (as my previous problem will attest), I'm still confused. Like, what all does the history file do? Which programs use it? (I know, relaynews and expire use it... any others? Don't answer that. It's rhetorical.) I had a huge problem one time, where I was sending some files over to our machine from another machine and for some reason, they didn't get put on disk, but they did get entered into our history file (mea culpa- I didn't document the whole sad scene; I was too busy being frustrated). Anyway, I have a good idea of how the whole mass fits together now, but it's taken awhile, and I'm always nervous that another gremlin is gonna jump out of the closet... In a nutshell: News is big, bad, and ugly. I would not wish upon my worst enemy that they be forced to bring it up from scratch. -Mike Schwager | Machine: Amiga 500, 3 MB RAM/30 HD INTERNET:schwager@cs.uiuc.edu | Bike: '83 Kawasaki KZ750 LTD UUCP:{uunet|convex|pur-ee}!uiucdcs!schwager| Band: Poi Dog Pondering // BITNET:schwager@mike.cs.uiuc.edu | Hero: Robert Bly \\ // University of Illinois, Dept. of Comp. Sci.| DoD: #0301 \X/Amiga
flee@cs.psu.edu (Felix Lee) (04/26/91)
>I have no idea how hard it would be to make the ordering of the lines >non-important, but it sure would be nice. It's a matter of using a different "best match" rule. C News expire currently uses "first match". Nearly every alternative is incompatible with this behavior. One alternative is "last match". This is possibly more natural than first match. I tend to want "all" near the top, since a general rule should appear before exceptions to the rule. Another rule is "earliest expire": pick the matching line that expires the article soonest. This rule makes it hard to say something like: news.announce x 30 - news x 5 - The opposite rule is "latest expire", which has the opposite problem. A good rule is "strongest match": X is a stronger match than Y if the set of X is a proper subset of the set of Y. This will probably do what everyone expects most of the time. Unfortunately, it isn't specific enough. What if you have: all.test x 3 - misc.all x 5 - How do you expire misc.test? You have to defer to another rule. -- Felix Lee flee@cs.psu.edu
a3@earth.rivm.nl (Adri Verhoef) (04/26/91)
Henry writes: >> An earlier line which >>also covers that group will cause its line to be ignored. Maybe it's a good idea to put this text in the 'explist' file, so that people are warned after they start editing it. In this simple way you can prevent some of the expire problems showing up. Just put in a short notice. In the meantime, do it for the sake of people that come to administer the news system after you will have been doing so.
henry@zoo.toronto.edu (Henry Spencer) (04/26/91)
In article <1991Apr25.001832.19898@s1.msi.umn.edu> molenda@s1.msi.umn.edu (Jason Molenda) writes: >> An earlier line which >>also covers that group will cause its line to be ignored. > >I've known about this for a long time and I know it is documented and >everything, but am I the only person who thinks it's a huge hole inviting >mistakes by new and old sysadmins alike? ... I thought about the issue somewhat. The trouble is that it's useful to be able to state a general rule and then establish exceptions, and any other definition of which rule dominates is more complicated, harder to understand, and more mistake-prone. -- And the bean-counter replied, | Henry Spencer @ U of Toronto Zoology "beans are more important". | henry@zoo.toronto.edu utzoo!henry
alexis@panix.uucp (Alexis Rosen) (04/27/91)
schwager@m.cs.uiuc.edu (Michael Schwager) writes: >Indeed, if I may make a comment regarding News (we ran notes up until >December of last year): > >Seems like there are all kinds of pitfalls, troubles, and caveats one can >enter when bringing News up. [...] >[...] News is so big and cantankerous, that it's so dang >confusing to figure out problems. It's taken me a few months to get it [...] > >In a nutshell: News is big, bad, and ugly. I would not wish upon my worst >enemy that they be forced to bring it up from scratch. I'd have to disagree with this. When I brought up Cnews, I had never done any sort of news administration. I had *no* experience writing in C under Unix (though I had a reading knowledge of C left over from earlier days) and in fact still haven't had much time to learn about it. I had not much experience with administering Unix in general, and was working with a fairly oddball system (A/UX, getting less odd all the time...) which nobody could offer much help with. That was over a year ago. It took me about a day to _thoroughly_ read all the docs and stuff, and to figure out what the appropriate answers were to the various questions that the Cnews config script asks you. I made careful notes that day about what each file in Cnews was and where it went. The next day I brought up Cnews. Took most of the day to get it to work- I had to figure out a few missing headers, and that sort of thing. Then I spent five minutes writing up the explist, sys, batchparams, and so on, and that was it. Done. Since then it's run for over a year without a single problem. I don't attribute this to any particular brilliance on my part. It comes from knowing that, even though it's running on a Mac, it's not Mac software. :-) Seriously, you can't just browse the docs for a few minutes and hope things will work out, because they won't. Especially given the unix-variant-independance of Cnews Henry and Geoff have done a remarkable job of uniformly superb quality. Sure it's big, but it has an awful lot to do. As for the bad and the ugly, I'd have to lay that at the door of sysops who don't read documentation and (at least as frequently) weird variants of Unix that barely live up to their name. The best thing you can do if you want to bring up Cnews from scratch is forget what you think you know and learn it right, the first time. Do that and you'll be very happy with the results. Cnews is at least as good as my professional work (and it's free!)... Those who know what an egotist I am will know what high praise that is. :-) --- Alexis Rosen Owner/Sysadmin, PANIX Public Access Unix, NY {cmcl2,apple}!panix!alexis
henry@zoo.toronto.edu (Henry Spencer) (04/28/91)
In article <1991Apr26.124402.4151@rivm.nl> a3@earth.rivm.nl (Adri Verhoef) writes: >>> An earlier line which >>>also covers that group will cause its line to be ignored. > >Maybe it's a good idea to put this text in the 'explist' file, >so that people are warned after they start editing it. I have limited sympathy for people who start editing without reading the documentation, and I have a strong feeling that such people will not read a warning in explist either. However, I've added a note to remind the careless that the order of lines is significant. -- And the bean-counter replied, | Henry Spencer @ U of Toronto Zoology "beans are more important". | henry@zoo.toronto.edu utzoo!henry
jad@nyama.UUCP (04/29/91)
In <1991Apr25.161229.18843@m.cs.uiuc.edu>, schwager@m.cs.uiuc.edu (Michael Schwager) writes: > >In a nutshell: News is big, bad, and ugly. I would not wish upon my worst >enemy that they be forced to bring it up from scratch. Ah, excuse me? A friend of mine and I got c-news and rn intalled, receiving and reading news in 3 hours. And that included a coffee break! C-news is big, but it's not bad and not ugly. I find the use of shell for most of simple use to be an advantage. What you do need to do is answer each and every question in conf/build *very carefully* or else. Chances are the defaults are ok, but if not then don't bother "fixing" doit.bin and friends: start over, it will make you happier in the end. This is not meant as a flame, on the contraire. You have to be carefull because c-news assumes you know what you're doing:-) >-Mike Schwager | Machine: Amiga 500, 3 MB RAM/30 HD >INTERNET:schwager@cs.uiuc.edu | Bike: '83 Kawasaki KZ750 LTD >UUCP:{uunet|convex|pur-ee}!uiucdcs!schwager| Band: Poi Dog Pondering // >BITNET:schwager@mike.cs.uiuc.edu | Hero: Robert Bly \\ // >University of Illinois, Dept. of Comp. Sci.| DoD: #0301 \X/Amiga -- Jose Dias jad@nyama.UUCP Who me? I didn't say anything! -- Jose Dias jad@nyama.UUCP Who me? I didn't say anything!