Re: memories
I wrote: | Until recently I had just 32MB of RAM. I added 64 more on | Saturday. Everything seemed fine to begin with---the 96MB was | detected in BIOS and by the kernel; I had much less disk-thrashing | in long Netscape sessions and so on. But | | If I leave the machine up overnight when I log on again in the | morning, I get constant segmentation faults (especially in | resource-heavy applications like emacs, TeX, X ...) and sometimes | a kernel-panic. Rebooting `fixes' the problem. I'm responding to my own message in case the information might be useful to others who could face similar problems. The difficulties were caused, as a number of people suggested, by a corrupt memory module. The memory problem was not detected in BIOS nor by the memtest program that is in the package `sysutils' (it reported `no errors'). However, massive problems were detected and reported by the memtest-86 program which is part of the hwtools package. I highly recommend this program. You copy it to a floppy and boot from that floppy. It then does a (very!) lengthy and detailed check of every nook and cranny of your memory modules. No home should be without one, Jim
memories
I have a Debian box which has been rock-solid in the three years I've been using it. Currently it's slink with the 2.0.38 kernel (custom-compiled) and just a few extras in /usr/local. No other OS. Until recently it had just 32MB of RAM. I added 64 more on Saturday. Everything seemed fine to begin with---the 96MB was detected in BIOS and by the kernel; I had much less disk-thrashing in long Netscape sessions and so on. But If I leave the machine up overnight (as has been my habit) with nobody logged on and only cron jobs running, when I log on again in the morning, `top' tells me that almost all of the memory is in use, and when I try to work, I get constant segmentation faults (especially in resource-heavy applications like emacs, TeX, X ...) and sometimes a kernel-panic. Rebooting `fixes' the problem. The hardware: Pentium 2 (233 with 512K cache), an Asus P2L97 AGP Motherboard, Quantum 4.3GB SCSI Hard Drive. Are there tools available that would help me diagnose the problem and hopefully solve it? Thanks in advance for any advice, Jim McCloskey
Re: memories
*- On 10 Jan, [EMAIL PROTECTED] wrote about memories I have a Debian box which has been rock-solid in the three years I've been using it. Currently it's slink with the 2.0.38 kernel (custom-compiled) and just a few extras in /usr/local. No other OS. Until recently it had just 32MB of RAM. I added 64 more on Saturday. Everything seemed fine to begin with---the 96MB was detected in BIOS and by the kernel; I had much less disk-thrashing in long Netscape sessions and so on. But If I leave the machine up overnight (as has been my habit) with nobody logged on and only cron jobs running, when I log on again in the morning, `top' tells me that almost all of the memory is in use, and when I try to work, I get constant segmentation faults (especially in resource-heavy applications like emacs, TeX, X ...) and sometimes a kernel-panic. Rebooting `fixes' the problem. The hardware: Pentium 2 (233 with 512K cache), an Asus P2L97 AGP Motherboard, Quantum 4.3GB SCSI Hard Drive. Are there tools available that would help me diagnose the problem and hopefully solve it? Thanks in advance for any advice, Which netscape are you using? Netscape 4.7 is much tighter on its memory leaks than previous versions. I have also found that X seems to have a memory leak somewhere. My solution to this is to restart the window manager, not logging out of X but just restarting the window manager. It is amazing but I can reclaim 128M/256M of swap by doing this sometimes. Brian Servis -- Mechanical Engineering | Never criticize anybody until you Purdue University | have walked a mile in their shoes, [EMAIL PROTECTED] | because by that time you will be a http://www.ecn.purdue.edu/~servis | mile away and have their shoes.
Re: memories
[EMAIL PROTECTED] writes: I have a Debian box which has been rock-solid in the three years I've been using it. Currently it's slink with the 2.0.38 kernel (custom-compiled) and just a few extras in /usr/local. No other OS. Until recently it had just 32MB of RAM. I added 64 more on Saturday. Everything seemed fine to begin with---the 96MB was detected in BIOS and by the kernel; I had much less disk-thrashing in long Netscape sessions and so on. But If I leave the machine up overnight (as has been my habit) with nobody logged on and only cron jobs running, when I log on again in the morning, `top' tells me that almost all of the memory is in use This would necessarily indicate a problem. Linux uses any memory that isn't currently being used as disk cache. Overnight I believe the updatedb command is run which accesses all of you hard drive and thus it's likely Linux allocates all your free memory to disk cache. When an application requests memory Linux will kindly reduce the amount of memory being used for cache. , and when I try to work, I get constant segmentation faults (especially in resource-heavy applications like emacs, TeX, X ...) and sometimes a kernel-panic. Rebooting `fixes' the problem. The hardware: Pentium 2 (233 with 512K cache), an Asus P2L97 AGP Motherboard, Quantum 4.3GB SCSI Hard Drive. Are there tools available that would help me diagnose the problem and hopefully solve it? Did these symptoms you're seeing only begin after you installed the new memory? If so then that might indicate a bad memory chip. There's a little utility called memtest in the sysutils package that might be able to detect it. There's an even more thorough test in the hwtools package (memtest86) that you actually boot into, via floppy. I haven't used these in a LONG time so maybe someone else can give you more details on them. Don't rely on your BIOS memory test. It isn't very thorough. Gary
Re: memories
id be willing to bet its bad memory. i would take the old 32MB out and keep the new 64MB in and try some tests.. http://www.freshmeat.net/search.php3?query=memory+test I haven't had experience with those programs, but they may show some results, i reccomend Microscope 7, but it is about $300 or something.. also a good test is runnign multiple copies of [EMAIL PROTECTED] (or distributed.net although ive never tried d.net) if its still flakey then id bet bad ram. make sure the ram sticks(if its 72pin) are the same speed/type(EDO/FPM), check the speed in the bios, if its 60ns try putting it at 70, play around with the timings ..also make sure you are not mixing SDRAM with EDO, 95% of mainboards can't handle that properly(EDO is 5V and SDRAM is typically 3.3V). on my machines with 64MB i usually run 4x copies of seti, with 128MB i run 9-10 copies, let it go for 24-48 hours if its still alive its declared stable. nate On Mon, 10 Jan 2000 [EMAIL PROTECTED] wrote: mcclos mcclos I have a Debian box which has been rock-solid in the three years I've mcclos been using it. Currently it's slink with the 2.0.38 kernel mcclos (custom-compiled) and just a few extras in /usr/local. No other OS. mcclos mcclos Until recently it had just 32MB of RAM. I added 64 more on mcclos Saturday. Everything seemed fine to begin with---the 96MB was detected mcclos in BIOS and by the kernel; I had much less disk-thrashing in long mcclos Netscape sessions and so on. But mcclos mcclos If I leave the machine up overnight (as has been my habit) with nobody mcclos logged on and only cron jobs running, when I log on again in the mcclos morning, `top' tells me that almost all of the memory is in use, and mcclos when I try to work, I get constant segmentation faults (especially in mcclos resource-heavy applications like emacs, TeX, X ...) and sometimes a mcclos kernel-panic. Rebooting `fixes' the problem. mcclos mcclos The hardware: Pentium 2 (233 with 512K cache), an Asus P2L97 AGP mcclos Motherboard, Quantum 4.3GB SCSI Hard Drive. mcclos mcclos Are there tools available that would help me diagnose the problem and mcclos hopefully solve it? mcclos mcclos Thanks in advance for any advice, mcclos mcclos Jim McCloskey mcclos mcclos mcclos -- mcclos Unsubscribe? mail -s unsubscribe [EMAIL PROTECTED] /dev/null mcclos [mailto:[EMAIL PROTECTED] ]-- Vice President Network Operations http://www.firetrail.com/ Firetrail Internet Services Limited http://www.aphroland.org/ Everett, WA 425-348-7336http://www.linuxpowered.net/ Powered By:http://comedy.aphroland.org/ Debian 2.1 Linux 2.0.36 SMPhttp://yahoo.aphroland.org/ -[mailto:[EMAIL PROTECTED] ]-- 9:07am up 143 days, 21:06, 2 users, load average: 1.81, 1.89, 1.72
Re: memories
Gary Hennigan [EMAIL PROTECTED] writes: [EMAIL PROTECTED] writes: I have a Debian box which has been rock-solid in the three years I've been using it. Currently it's slink with the 2.0.38 kernel (custom-compiled) and just a few extras in /usr/local. No other OS. Until recently it had just 32MB of RAM. I added 64 more on Saturday. Everything seemed fine to begin with---the 96MB was detected in BIOS and by the kernel; I had much less disk-thrashing in long Netscape sessions and so on. But If I leave the machine up overnight (as has been my habit) with nobody logged on and only cron jobs running, when I log on again in the morning, `top' tells me that almost all of the memory is in use This would necessarily indicate a problem. Linux uses any memory that Typo alert! Should have been would not. Sorry about that. Gary