Re: Again: Most programs don't run. Please help!
On Saturday 26 January 2002 05:48 pm, Andreas Goesele wrote: [snip] The problem BTW returned again, now the third time, again only for some minutes. It started during a cron job and started and *ended* during heavy disk activity. as i mentioned earlier, you should determine whether one of the cron scheduled programs is causing the crashes. to check the dependencies for the apps that cron runs, do apt-cache show filename and check the output for dependency conflicts.
Re: Again: Most programs don't run. Please help!
* Andreas Goesele ([EMAIL PROTECTED]) spake thusly: Dimitri Maziuk [EMAIL PROTECTED] writes: I'd ask on the kernel mailing list and/or on comp.os.linux.system. Thanks. But where would I find comp.os.linux.system? My news-server doesn't provide it nor did I find it on googles groups. Sorry, make that comp.os.linux.development.system. It's on google, as well as kernel list (linux.kernel; it is read-only, of course). Dima
Re: Again: Most programs don't run. Please help!
On 27 Jan 2002, Andreas Goesele wrote: Cameron Kerr [EMAIL PROTECTED] writes: Well, if the error is indeterminate, then I would start to suspect your hardware. Errors such as this often indicate bad memory, in my experience. Getting segfaults all the time is a classic example of this. Get hold of memtest86 (http://www.teresaudio.com/memtest86/) and give your memory a good going over. Three completed runs, all 16 tests, no error. OK, so thats your memory tested... Running crashme off a boot floppy might also be, ah, interesting... Would this, run by a simple computer user - except possibly crashing my system - produce information which could help to pinpoint the problem? Its not meant to pinpoint the problem. If the system crashes, that would be a problem, if not, the CPU/kernel is (as far as running operations go) okay. But, yeah. I've never used it myself, I just pointed it out because it may come in useful ;) The problem BTW returned again, now the third time, again only for some minutes. It started during a cron job and started and *ended* during heavy disk activity. (Actually it ended when I started memtest for a third run which seemed to cause some swapping activity.) Maybe, I should test my hard disks. What programs would you recommend for that? Well then, if it only happens during periods of high disk activity, I would still suspect your hardware. We've already determined its not likely to be your memory, but it could well be some other conflict. Perhaps if you remove _all_ unnecessary drivers from the kernel (easiest if you are running a very modular kernel), then you can start to tell what is conflicting. Make sure you are running a recent kernel. Also, please give details as to your (detailed) hardware specs, paying particular attention to chipsets, and the drivers used. I hope this may shed some light on your predicament. If not, start removing hardware, such as sound cards, and seeing if that starts to make any difference. Good luck. Cameron Kerr -- [EMAIL PROTECTED] http://homepages.paradise.net.nz/~cameronk/
Re: Again: Most programs don't run. Please help!
* Andreas Goesele ([EMAIL PROTECTED]) spake thusly: Hi, my problem as described in [EMAIL PROTECTED] and with some more details in [EMAIL PROTECTED] returned, though only for some minutes. Simple description: Suddenly most programs won't start anymore but will give the error message: foo: relocation error: /lib/libnss_compat.so.2: symbol rectory, version GLIBC_2.0 not defined in file libc.so.6 with link time reference As the problem returned there is - without any doubt - something deeply wrong with my system, but I don't have any clue what and how to go about finding out. So please, if you have any suggestion about how I could locate and eventually solve the problem, let me know. I'd ask on the kernel mailing list and/or on comp.os.linux.system. Dima -- I have not been able to think of any way of describing Perl to [person] Hello, blind man? This is color. -- DPM
Re: Again: Most programs don't run. Please help!
* Andreas Goesele ([EMAIL PROTECTED]) spake thusly: Simple description: Suddenly most programs won't start anymore but will give the error message: foo: relocation error: /lib/libnss_compat.so.2: symbol rectory, version GLIBC_2.0 not defined in file libc.so.6 with link time reference OK, GLIBC_2.0 refers to glibc5 (or 4?). Anyway, the likely reason that it is not defined in libc.so.6 is because it only supports GLIBC_2.2 (glibc 2.2.4 actually). Normally I would have expected it to be backwards compatible, but perhaps the Debian Maintainer does things differently, in order to keep the binaries smaller. Perhaps there is a compatibility packages in oldlibs (libc5 package) Warning: This is just an educated guess. Good luck Cameron Kerr -- [EMAIL PROTECTED] http://homepages.paradise.net.nz/~cameronk/
Re: Again: Most programs don't run. Please help!
Dimitri Maziuk [EMAIL PROTECTED] writes: I'd ask on the kernel mailing list and/or on comp.os.linux.system. Thanks. But where would I find comp.os.linux.system? My news-server doesn't provide it nor did I find it on googles groups. Andreas Goesele
Re: Again: Most programs don't run. Please help!
Cameron Kerr [EMAIL PROTECTED] writes: * Andreas Goesele ([EMAIL PROTECTED]) spake thusly: Simple description: Suddenly most programs won't start anymore but will give the error message: foo: relocation error: /lib/libnss_compat.so.2: symbol rectory, version GLIBC_2.0 not defined in file libc.so.6 with link time reference OK, GLIBC_2.0 refers to glibc5 (or 4?). Anyway, the likely reason that it is not defined in libc.so.6 is because it only supports GLIBC_2.2 (glibc 2.2.4 actually). Normally I would have expected it to be backwards compatible, but perhaps the Debian Maintainer does things differently, in order to keep the binaries smaller. Perhaps there is a compatibility packages in oldlibs (libc5 package) Thanks. But would that explanation be compatible with the fact that my system most of the time functions all-right and then suddenly for some time (one day and then a few minutes) not? If libc.so.6 wasn't backwards compatible I would expect that the problem would arise all the time. And why should the great majority of programs, in a more or less up to date woody system, depend on glibc5 or glibc4? I continue to be mystified ... Andreas Goesele
Re: Again: Most programs don't run. Please help!
On 27 Jan 2002, Andreas Goesele wrote: Thanks. But would that explanation be compatible with the fact that my system most of the time functions all-right and then suddenly for some time (one day and then a few minutes) not? If libc.so.6 wasn't backwards compatible I would expect that the problem would arise all the time. And why should the great majority of programs, in a more or less up to date woody system, depend on glibc5 or glibc4? I continue to be mystified ... Well, if the error is indeterminate, then I would start to suspect your hardware. Errors such as this often indicate bad memory, in my experience. Getting segfaults all the time is a classic example of this. Get hold of memtest86 (http://www.teresaudio.com/memtest86/) and give your memory a good going over. Running crashme off a boot floppy might also be, ah, interesting... Andreas Goesele Cameron Kerr -- [EMAIL PROTECTED] http://homepages.paradise.net.nz/~cameronk/
Re: Again: Most programs don't run. Please help!
On Saturday 26 January 2002 03:34 pm, Andreas Goesele wrote: [snip] foo: relocation error: /lib/libnss_compat.so.2: symbol rectory, version GLIBC_2.0 not defined in file libc.so.6 with link time reference OK, GLIBC_2.0 refers to glibc5 (or 4?). Anyway, the likely reason that it is not defined in libc.so.6 is because it only supports GLIBC_2.2 (glibc 2.2.4 actually). Normally I would have expected it to be backwards compatible, but perhaps the Debian Maintainer does things differently, in order to keep the binaries smaller. Perhaps there is a compatibility packages in oldlibs (libc5 package) Thanks. But would that explanation be compatible with the fact that my system most of the time functions all-right and then suddenly for some time (one day and then a few minutes) not? If libc.so.6 wasn't backwards compatible I would expect that the problem would arise all the time. And why should the great majority of programs, in a more or less up to date woody system, depend on glibc5 or glibc4? given that the disfunction is only occasional, it would seem that some app that runs only occasionally provokes the dependency issue. got anything set up in cron that might be doing this?
Re: Again: Most programs don't run. Please help!
ben [EMAIL PROTECTED] writes: Thanks. But would that explanation be compatible with the fact that my system most of the time functions all-right and then suddenly for some time (one day and then a few minutes) not? If libc.so.6 wasn't backwards compatible I would expect that the problem would arise all the time. And why should the great majority of programs, in a more or less up to date woody system, depend on glibc5 or glibc4? given that the disfunction is only occasional, it would seem that some app that runs only occasionally provokes the dependency issue. got anything set up in cron that might be doing this? I was thinking of that. But to my uneducated eye entries in cron.d, cron.daily, cron weekly and cron.monthly looked rather inoffensive. Or is there something in the following lists which could be the offender? Is there some other source of cron jobs? (anacrontab and crontab only start the jobs in these directories.) cron.daily: 0anacron, 5snort, apache, calendar, cfengine, dwww, exim, find, logrotate, man-db, mgetty, modutils, netbase, netkit-inetd, standard, sysklogd, tetex-bin, tripwire-1.2 cron.d: anacron, exim, logcheck, postgresql, shaper, tiger cron.weekly: 0anacron, apt-move, cfengine, cfingerd, cvs, dpkg-mountable, efax, htdig, lpr, man2html, man-db, sysklogd cron.monthly: 0anacron, rwhod, standard Andreas Goesele
Re: Again: Most programs don't run. Please help!
Cameron Kerr [EMAIL PROTECTED] writes: Well, if the error is indeterminate, then I would start to suspect your hardware. Errors such as this often indicate bad memory, in my experience. Getting segfaults all the time is a classic example of this. Get hold of memtest86 (http://www.teresaudio.com/memtest86/) and give your memory a good going over. Three completed runs, all 16 tests, no error. Running crashme off a boot floppy might also be, ah, interesting... Would this, run by a simple computer user - except possibly crashing my system - produce information which could help to pinpoint the problem? (To say the truth - the explanation at http://people.delphi.com/gjc/crashme.html didn't give me the impression that I could interpret the result.) The problem BTW returned again, now the third time, again only for some minutes. It started during a cron job and started and *ended* during heavy disk activity. (Actually it ended when I started memtest for a third run which seemed to cause some swapping activity.) Maybe, I should test my hard disks. What programs would you recommend for that? Andreas Goesele
Re: Again: Most programs don't run. Please help!
On Saturday 26 January 2002 04:25 pm, Andreas Goesele wrote: [snip] I was thinking of that. But to my uneducated eye entries in cron.d, cron.daily, cron weekly and cron.monthly looked rather inoffensive. Or is there something in the following lists which could be the offender? Is there some other source of cron jobs? (anacrontab and crontab only start the jobs in these directories.) cron.daily: 0anacron, 5snort, apache, calendar, cfengine, dwww, exim, find, logrotate, man-db, mgetty, modutils, netbase, netkit-inetd, standard, sysklogd, tetex-bin, tripwire-1.2 cron.d: anacron, exim, logcheck, postgresql, shaper, tiger cron.weekly: 0anacron, apt-move, cfengine, cfingerd, cvs, dpkg-mountable, efax, htdig, lpr, man2html, man-db, sysklogd cron.monthly: 0anacron, rwhod, standard it might be worth your while going through the man pages for these--or using any other means you can come up with--to check for version numbers and possibly consequent dependency conflicts. it may well be that all you need to do is force an upgrade for one or another of them. apropos cfingerd, i can't remember the details but i do remember reading of something buggy about that. if i were you, i would definitely check on everything that cron invokes, and srupulously check your logs for any clue about the times the failures occur. follow that up with a search for anything at all that might have been modified around the same time. also, there's either an apt or a dpkg option that spits out dependency info but i can't remember the fine points of that right now. dpkg -C will return info on anything that hasn't been configured properly.
Again: Most programs don't run. Please help!
Hi, my problem as described in [EMAIL PROTECTED] and with some more details in [EMAIL PROTECTED] returned, though only for some minutes. Simple description: Suddenly most programs won't start anymore but will give the error message: foo: relocation error: /lib/libnss_compat.so.2: symbol rectory, version GLIBC_2.0 not defined in file libc.so.6 with link time reference As the problem returned there is - without any doubt - something deeply wrong with my system, but I don't have any clue what and how to go about finding out. So please, if you have any suggestion about how I could locate and eventually solve the problem, let me know. Many thanks in advance! Andreas Goesele PS: This time the beginning of the problem coincided with the starting of the daily cron jobs and heavy disk activity.