From: Digestifier To: Linux-Admin@senator-bedfellow.mit.edu Reply-To: Linux-Admin@senator-bedfellow.mit.edu Date: Mon, 30 Aug 93 17:24:08 EDT Subject: Linux-Admin Digest #25 Linux-Admin Digest #25, Volume #1 Mon, 30 Aug 93 17:24:08 EDT Contents: Re: Linux trusted by SPARC (Drew Eckhardt) Re: booting .99P12 (new kernel) (Drew Eckhardt) Re: booting .99P12 (new kernel) (Drew Eckhardt) Rnews problem in UUCP (Steve W. Wynen) Re: restricting newsgroups (Bill Heiser) Re: booting .99P12 (new kernel) (Bill Heiser) Re: Linux bbs software? (Mark Buckaway) Re: booting .99P12 (new kernel) (Drew Eckhardt) Re: Rnews problem in UUCP (Andreas Klemm) Re: big UUCP spool files (Jay Pfaffman) Re: booting .99P12 (new kernel) (James A Robinson) Re: big UUCP spool files (John Henders) Re: booting .99P12 (new kernel) (John Henders) ---------------------------------------------------------------------------- From: drew@juliet.cs.colorado.edu (Drew Eckhardt) Subject: Re: Linux trusted by SPARC Date: Mon, 30 Aug 1993 00:19:18 GMT In article entropy@world.std.com (Lawrence Foard) writes: >In article , >C Wayne Huling wrote: >>I have been working on setting up a small network of linux machines for my >>company. Part of the reasoning is, many people here have to share a single >>SPARC station, and I suggested using the Linux machines X capability to >>expand the amount of people capable of working on the SPARC. Well any how, >>I have all the Linux hosts trusting each other and I am capable of using xon >>(when specifing a full path?) but I cannot rsh to the SPARC and hence cannot >>xon to the SPARC. Has anyone succefully accomplished this? > >Atleast with the SPARC machines I've used all thats required is a >.rhosts file in the directory of the user you want to rsh into. >However they have all used SUN OS I don't know is solaris is any >different. ~/.rhosts is a generic UNIX thing, and will let users login / rsh without a password. Host names should be fully qualified, and this file should be read-write owner only. Unix system also support /etc/hosts.equiv, although I don't recommend using /etc/hosts.equiv for security reasons. -- Boycott USL/Novell for their absurd anti-BSDI lawsuit. | Condemn Colorado for Amendment Two. | Drew Eckhardt Use Linux, the fast, flexible, and free 386 unix | drew@cs.Colorado.EDU Will administer Unix for food | ------------------------------ From: drew@juliet.cs.colorado.edu (Drew Eckhardt) Subject: Re: booting .99P12 (new kernel) Date: Mon, 30 Aug 1993 01:10:07 GMT In article <25mn0s$o38@news.bu.edu> heiser@bumetb.bu.edu (Bill Heiser) writes: >I installed Linux .99P12 last week using SLS 1.03. Today I decided >to try building a new kernel, to make sure the "clean distribution" >would build, before trying any new kernels for patches that may come >out, etc. > >I ran "make config", and selected options appropriate for my system >(scsi disks, etc). A half hour later (!!!!!yikes!!!) (on a 486/33!!), The most important factor in determining build speed is the amount of memory you have. GCC2 is a pig, and will thrash in 4M of memory. In 8M or more, it's quite tolerable. >the build completed. I then tried to boot the new kernel. It got >to the point where it got through LILO and started "loading kernel" >(or whatever wording it uses right after leaving LILO). Then the >machine cold-booted. > >So unable to get to a point where I could manually specify a kernel to >load, I booted from my boot floppy. I copied the old kernel that I'd >saved earlier, back to /zImage. But even now, the same symptoms occur >when I try to boot from the system disk! The only way it will boot is >if I use the floppy. > >Is this a common problem? Yes. >How do I restore the bootability of the Linux system? Traiditional unices only support one local filesystem, and the bootstrap loader knows enough about the layout to read any file off of it. This was the case with Linux, the minix filesystem, and shoelace, but is no more now that Linux supports a myriad of local filesystems. Supporting every filesystem in the boot code would be a nightmare, so instead we leave the boot program a map of what blocks the kernel(s) occupy on the raw device(s). The LILO bootstrap finds the map, and reads the appropriate blocks off the raw device. If you merely copy a kernel onto the disk, and don't update this mapping, Bad Things(tm) happen when the wrong blocks get loaded. Rerun /etc/lilo/lilo with the appropriate switches and config file, and things will straighten themselves out. -- Boycott USL/Novell for their absurd anti-BSDI lawsuit. | Condemn Colorado for Amendment Two. | Drew Eckhardt Use Linux, the fast, flexible, and free 386 unix | drew@cs.Colorado.EDU Will administer Unix for food | ------------------------------ From: drew@juliet.cs.colorado.edu (Drew Eckhardt) Subject: Re: booting .99P12 (new kernel) Date: Mon, 30 Aug 1993 01:13:54 GMT In article <93240.115501BDA104@psuvm.psu.edu> Kyr Velusian writes: >In article <25mn0s$o38@news.bu.edu>, heiser@bumetb.bu.edu (Bill Heiser) says: >> >>I ran "make config", and selected options appropriate for my system >>(scsi disks, etc). A half hour later (!!!!!yikes!!!) (on a 486/33!!), > >used to take me about 10 hours on my 386DX20 back in the .98 days... >;-) used to take me about 10-15 minutes on my 386-33 back in .11 days. With 4M of memory and no swap :-) -- Boycott USL/Novell for their absurd anti-BSDI lawsuit. | Condemn Colorado for Amendment Two. | Drew Eckhardt Use Linux, the fast, flexible, and free 386 unix | drew@cs.Colorado.EDU Will administer Unix for food | ------------------------------ From: swwynen@undergrad.math.uwaterloo.ca (Steve W. Wynen) Crossposted-To: comp.os.linux.help Subject: Rnews problem in UUCP Date: 30 Aug 93 01:26:52 GMT Hello Everybody; Hope you can help. I am trying to set up a UUCP site to receive email and news. i have got the UUCP faq, the recommended reading, and this one still escapes me. I have the UUCP feed mostly working, can send and receive email, but i cannot get my news to come in properly, it gets sent over , but then seems to dissappear into the bit bucket. In the file /usr/spool/uucp/.Log/joesys I get many many messages like: Executing X.joesysX0ctr (rnews) ERROR: Execution: Exit Status 1 Execution failed (X.joesysX0ctr) I have don e everything that the Cnews README said to do but, it doesnt work. Any suggestions would be appreciated. ------------------------------ From: heiser@bumetb.bu.edu (Bill Heiser) Subject: Re: restricting newsgroups Date: 30 Aug 1993 03:11:36 GMT Reply-To: heiser@bumetb.bu.edu (Bill Heiser) In article sperry@well.sf.ca.us (Michael P. Sperry) writes: > I would like to offer certain infamous newsgroups to my users (a.b.p.e). >However, I would like to be able to have people under age eighteen on my system. >Does anyone know a method whereby I could prevent children from receving these >newsgroups (besides not letting them on my system)? I was trying to work >out a way using Unix groups, but this is more of a C-news type problem. This isn't a LINUX question, but I'll answer it :-) You could try changing the directory "group ownerships" of the spool directories you want to restrict. Then put your users in appropriate groups and set the read permissions of those spool directories accordingly. This might be a pain to keep up though, if you allow your Cnews to automatically process "newgroups". ------------------------------ From: heiser@bumetb.bu.edu (Bill Heiser) Subject: Re: booting .99P12 (new kernel) Date: 30 Aug 1993 03:15:29 GMT Reply-To: heiser@bumetb.bu.edu (Bill Heiser) In article <1993Aug30.011007.9334@colorado.edu> drew@juliet.cs.colorado.edu (Drew Eckhardt) writes: > >The most important factor in determining build speed is the amount of memory >you have. GCC2 is a pig, and will thrash in 4M of memory. In 8M or >more, it's quite tolerable. Ahhh ... I have 16mb of memory ... but was running X when I did the kernel build, so there wasn't much memory free. Next time I'll quit X before rebuilding the kernel :-) Speaking of which ..h ... there must be a memory leak, possibly in X386. When I first start X, I have about 4mb of memory free, but it seems to disappear even when I don't start new apps :-) ------------------------------ From: mark@datasoft.north.net (Mark Buckaway) Subject: Re: Linux bbs software? Date: Sun, 29 Aug 1993 15:26:19 GMT Bob Myers (bmyers@dseg.ti.com) wrote: : One of the main reasons is that you'd want some type of protection mechanism : for accessing either files for downloading and/or newsgroup reading. I'm sure : that you wouldn't want kids getting onto your bbs and downloading files/messages : from alt.bin.pic.erotica.* or getting into discussions on the alt.sex.* tree : either. I have solved this problem by limiting the age for an account to 18. If I did want kiddies online, access restrictions would be easy. Incidently, I am not new to being a sysop. I have run about various BBS programs on different OS's over 5-6 years. None of them every suited my needs exactly. I have chosen Linux now because it appears something I can grow with. : I'm running Uniboard right now, and have had some good responces so far. I do : have the ability to limit newsgroup access (yep, i get over 2100 newsgroups on : my Linux system at home -- thank god for WorldBlazers and TurboPep.) One : of the drawbacks so far is file transfers; I need to get some of the protocol : sources and recompile them. Once of the biggest farses with most UNIX bbs program is they expect you to have your user login under a BBS account. The user is then prompted to a username and passwd from the BBS program running as a shell. (I am not sure how UNIBOARD does this, but I know you'll defend BS program to the death). This makes accounting almost impossible and BBS setup a headache. I am running a pay for access BBS and want to charge by the hour. The ablility to go to 32 lines with ease is the reason I am now with UNIX (Linux). With my setup, the user will have a real UNIX login, and the .profile will controls what the user see. As for not permitting users access to certain groups.....most newsreader have an option to point to WHAT active file to read from. If this is not an option, it would be easy to add in. : -bob : p.s. If you need an ftp address for uniboard, let me know... Oh...that's quite alright. I have no intention of running it. Mark (Follows mark to alt.bbs.allsysop, this has nothing to Linux) Mark -- Mark Buckaway | "UNIX and OS/2 are operating systems, DataSoft Communications | Windows is a pitiful shell, System Administrator | DOS is an installible virus." root@datasoft.north.net | uunorth!datasoft!root | ====================================== ------------------------------ From: drew@kinglear.cs.colorado.edu (Drew Eckhardt) Subject: Re: booting .99P12 (new kernel) Date: Mon, 30 Aug 1993 13:48:22 GMT In article <25rrch$ed4@news.bu.edu> heiser@bumetb.bu.edu (Bill Heiser) writes: >In article <1993Aug30.011007.9334@colorado.edu> drew@juliet.cs.colorado.edu (Drew Eckhardt) writes: >> >>The most important factor in determining build speed is the amount of memory >>you have. GCC2 is a pig, and will thrash in 4M of memory. In 8M or >>more, it's quite tolerable. > >Ahhh ... I have 16mb of memory ... but was running X when I >did the kernel build, so there wasn't much memory free. Next time >I'll quit X before rebuilding the kernel :-) With a build time that high (My perceptions of what's right are probably skewed by my experience with ancient Linux revisions, GCC 1, and partial builds where I only tweak one source file), I assumed it was a memory shortage, but with 16M you shouldn't have any problems compiling the Linux kernel, even under X, and the bulk of the Linux code is device drivers, networking code, etc. that don't usually find there way into the systems I build. Other programs are a different story - I ran into a case under BSD, m68k, where I had to raise the hard RLIMIT, since GCC2.3.3 needed 32M to compile a file including an X bitmap and the default hard limit was only 16M! GCC 2.4.5 did nicely with a 16M limit in place, but memory usage was still frightening, not to mention the size of the .s files it spewed without using the -pipe option! Try -pipe, you should see some improvement since disk access is reduced because cpp and assembler intermediate files are never committed to disk. (BTW, why isn't this an option in the default CFLAGS? If you aren't thrashing, it reduces disk access) Is this on an otherwise idle system, or with other things going on? With how generic / pruned down of a kernel? Most of Linux's weight is in the diverse driver support, this should have a real impact on compile time. >Speaking of which ..h ... there must be a memory leak, possibly in X386. >When I first start X, I have about 4mb of memory free, but it seems >to disappear even when I don't start new apps :-) If you're running the networking code, there are reliable reports of memory leaks in the net-2 code in .99.12. Some X server implementations have memory leaks too, I'm not sure about Xfree86 specifically. On the systems I administer, I use DisplayManager*terminateServer: true in my xdm configuration files so I get a new Xserver for each login session, minimizing the impact of an Xserver memory leak. You always have the problem where a user level free() can't return freed pages to the heap, so this is probably a good idea even if there isn't a real memory leak. There's also the Linux unfied buffer cache, which basically grows to fill all available memory that's not in use for something else. People used to traditional unices often confuse this with a real memory leak, but you can consider "real free memory" as that marked free and that marked buffer cache. -- Boycott USL/Novell for their absurd anti-BSDI lawsuit. | Drew Eckhardt, Condemn Colorado for Amendment Two. | Profesional Linux Use Linux, the fast, flexible, and free 386 unix | Consultant Will administer Unix for food | drew@cs.Colorado.EDU ------------------------------ Crossposted-To: comp.os.linux.help From: andreas@knobel.knirsch.de (Andreas Klemm) Subject: Re: Rnews problem in UUCP Date: Mon, 30 Aug 1993 13:07:36 GMT swwynen@undergrad.math.uwaterloo.ca (Steve W. Wynen) writes: >Hello Everybody; >Hope you can help. I am trying to set up a UUCP site to receive email >and news. i have got the UUCP faq, the recommended reading, and this >one still escapes me. Why don't you say what kind of Linux version you are running ? Would be fine, too, if you would mention, which uucp version you run ... >I have the UUCP feed mostly working, can send and receive email, but >i cannot get my news to come in properly, it gets sent over , >but then seems to dissappear into the bit bucket. >In the file /usr/spool/uucp/.Log/joesys I get many many messages like: Strange, which kind of uucp logginf report do you use ? SLS comes with HDB and taylor compiled in, but is setup to use HDB kind (Honey Dan Beer) of config/spooling/logging .... >Executing X.joesysX0ctr (rnews) >ERROR: Execution: Exit Status 1 >Execution failed (X.joesysX0ctr) Could it be the case, that your rnews isn't found by uux ? Check your local config files in /usr/local/lib/uucp (Permissions is a good start) if your news feed is permitted to run the rnews program .... >I have done everything that the Cnews README said to do but, it doesnt work. >Any suggestions would be appreciated. Why can't your newsfeed help ?! He has obviously a running system ?! I think you aren't experienced in configuring uucp/news/mail ?! Then your newsfeed should help you in this - general - points of questions .... -- /-\ Andreas Klemm +-----------------+ |@|########################################################-@ "pay for it !" | \-/ 41469 Neuss Germany phone +49/ 2137 12609 +-----------------+ ------------------------------ From: pfaffman@relax.des.edu (Jay Pfaffman) Subject: Re: big UUCP spool files Date: 30 Aug 93 11:37:13 GMT gomez@enuxsa.eas.asu.edu (JL Gomez) writes: >Is there anything I can do short of adding cron entries to reduce >the size of those UUCP spool files in .Log, .Admin, etc? Short of adding cron entries, you can delete them by hand. -- Jay Pfaffman pfaffman@relax.des.edu POBox 128, Ripton, vt 05766 pfaffman@pilot.njin.net 802-388-6503 (H) 802-388-1754 (O) 802-388-7305 (FAX) ------------------------------ From: jcg@world.std.com (James A Robinson) Subject: Re: booting .99P12 (new kernel) Date: Mon, 30 Aug 1993 19:40:37 GMT drew@kinglear.cs.colorado.edu (Drew Eckhardt) writes: >>Ahhh ... I have 16mb of memory ... but was running X when I >>did the kernel build, so there wasn't much memory free. Next time >>I'll quit X before rebuilding the kernel :-) >With a build time that high (My perceptions of what's right are >probably skewed by my experience with ancient Linux revisions, GCC 1, >and partial builds where I only tweak one source file), I assumed it >was a memory shortage, but with 16M you shouldn't have any problems >compiling the Linux kernel, even under X, and the bulk of the Linux >code is device drivers, networking code, etc. that don't usually >find there way into the systems I build. I agree, I was able to compile the kernel in about 30minutes under X, and I only have 8 megs of 80ns RAM. This is kernel v0.99p6-26 (or so the SLS says) without SCSI or NFS, but everything else. Jim Please send all email to: jimr@world.std.com ------------------------------ From: jhenders@jonh.wimsey.bc.ca (John Henders) Subject: Re: big UUCP spool files Date: Mon, 30 Aug 1993 20:50:08 GMT pfaffman@relax.des.edu (Jay Pfaffman) writes: >gomez@enuxsa.eas.asu.edu (JL Gomez) writes: >>Is there anything I can do short of adding cron entries to reduce >>the size of those UUCP spool files in .Log, .Admin, etc? >Short of adding cron entries, you can delete them by hand. In the contrib/ directory in the Taylor 1.4 source there's a program called savelog that will save the Taylor style log files. I don't know if it could be adapted to HDB. Otherwise, a simple script run daily could accomplish the same thing. -- John Henders GO/MU/E d* -p+ c+++ l++ t- m--- s/++ g+ w+++ -x+ ------------------------------ From: jhenders@jonh.wimsey.bc.ca (John Henders) Subject: Re: booting .99P12 (new kernel) Date: Mon, 30 Aug 1993 20:55:40 GMT drew@kinglear.cs.colorado.edu (Drew Eckhardt) writes: >... GCC 2.4.5 did nicely with a 16M limit in place, but memory >usage was still frightening, not to mention the size of the .s files >it spewed without using the -pipe option! >Try -pipe, you should see some improvement since disk access is >reduced because cpp and assembler intermediate files are never >committed to disk. I just tried using pipe on an 8 meg machine, while using nn, with just the normal sendmail and selection daemons running. Under this setup, using gcc 2.4.5 and compiling 99pl12, it was a tossup on whether -pipe did any good at all. Real time was 35 minutes for -pipe and 34 without. Free showed 1.5 meg swapped out compared to 800-1meg without -pipe. As a comparison, compiling 99pl9 with gcc2.3.3 would take 20-25 minutes on the same amchine and didn't go into swap at all. -- John Henders GO/MU/E d* -p+ c+++ l++ t- m--- s/++ g+ w+++ -x+ ------------------------------ ** FOR YOUR REFERENCE ** The service address, to which questions about the list itself and requests to be added to or deleted from it should be directed, is: Internet: Linux-Admin-Request@NEWS-DIGESTS.MIT.EDU You can send mail to the entire list (and comp.os.linux.admin) via: Internet: Linux-Admin@NEWS-DIGESTS.MIT.EDU Linux may be obtained via one of these FTP sites: nic.funet.fi pub/OS/Linux tsx-11.mit.edu pub/linux sunsite.unc.edu pub/Linux End of Linux-Admin Digest ******************************