From: <pe...@ho...> - 2004-03-30 21:08:37
|
My machine here gets constant lockups when using smartmontools. It will normally run anywhere between 10 minutes and 2 hours before locking up SOLID. Have attached output from lspci -v, dmidecode, dmesg and tree /sys as well as .config which I hope will be useful for you guys. Linux kernel version: 2.6.5-rc2 (same thing happens with 2.6.4) Smartmontools version: 5.30-2 I am a bit hesitant to install smartmontools again on this pc (for home production use) but if you need the output from smartctl then I will of course arrange that. |
From: Bruce A. <ba...@gr...> - 2004-03-31 04:22:58
|
Hi Peter, > My machine here gets constant lockups when using smartmontools. It > will normally run anywhere between 10 minutes and 2 hours before > locking up SOLID. Have attached output from lspci -v, dmidecode, dmesg > and tree /sys as well as .config which I hope will be useful for you > guys. Am I right that you don't have a promise controller? (See the smartmontools WARNINGS file). In which case this is an unreported problem: did it appear with the 2.6.4 and on kernels? > Linux kernel version: 2.6.5-rc2 (same thing happens with 2.6.4) > Smartmontools version: 5.30-2 > > I am a bit hesitant to install smartmontools again on this pc (for home > production use) but if you need the output from smartctl then I will of > course arrange that. I'm not sure quite where to start. Is smartd responsible (in other words, turning it off fixes the lockups?). Could you please send SYSLOG (usually: /var/log/messages) showing smartd output before and around one of the lockups? Cheers, Bruce |
From: Peter H. <pe...@ho...> - 2004-03-31 11:04:12
|
Quoting Bruce Allen <ba...@gr...>: > Am I right that you don't have a promise controller? (See the > smartmontools WARNINGS file). In which case this is an unreported > problem: did it appear with the 2.6.4 and on kernels? Same problem with 2.6.4 and the machine does not have a promise controller (intel i believe - will check when I get home). The machine is a dell L600. > I'm not sure quite where to start. Is smartd responsible (in other words, > turning it off fixes the lockups?). Well, as soon as I did deleted it again (dpkg -P smartmontools) the problem went away even with SMART enabled for the drive. Some hours later the "ksoftirqd" started eating CPU like crazy but I havent tried restarting the machine when SMART (not the daemon) was not enabled. Will try that tonight. > Could you please send SYSLOG (usually: /var/log/messages) showing smartd > output before and around one of the lockups? I of course can - but there is nothing there - I checked myself. One moment running happily - the next a solid lockup. /peter |
From: Bruce A. <ba...@gr...> - 2004-03-31 14:24:41
|
> > Could you please send SYSLOG (usually: /var/log/messages) showing smartd > > output before and around one of the lockups? > > I of course can - but there is nothing there - I checked myself. One moment > running happily - the next a solid lockup. By default smartd polls the devices every thirty minutes. Could you please look at the smartd timestamps in SYSLOG the next time you run the system, and confirm that the lockup takes place at the instant when smartd polls the devices? Cheers, Bruce |
From: <li...@pe...> - 2004-04-01 02:23:16
|
I don't see your .config in the e-mail you sent, only the other outputs. Also, send your lsmod. And better test it with "init 1" just running smartd. On Wed, 31 Mar 2004, Peter Hoeg wrote: > Quoting Bruce Allen <ba...@gr...>: > > > Am I right that you don't have a promise controller? (See the > > smartmontools WARNINGS file). In which case this is an unreported > > problem: did it appear with the 2.6.4 and on kernels? > > Same problem with 2.6.4 and the machine does not have a promise controller > (intel i believe - will check when I get home). The machine is a dell L600. > > > I'm not sure quite where to start. Is smartd responsible (in other words, > > turning it off fixes the lockups?). > > Well, as soon as I did deleted it again (dpkg -P smartmontools) the problem went > away even with SMART enabled for the drive. Some hours later the "ksoftirqd" > started eating CPU like crazy but I havent tried restarting the machine when > SMART (not the daemon) was not enabled. Will try that tonight. > > > Could you please send SYSLOG (usually: /var/log/messages) showing smartd > > output before and around one of the lockups? > > I of course can - but there is nothing there - I checked myself. One moment > running happily - the next a solid lockup. -- http://www.pervalidus.net/contact.html |
From: Andy I. <ad...@he...> - 2004-04-02 06:08:08
|
On Wed, Mar 31, 2004 at 01:04:05PM +0200, Peter Hoeg wrote: > Well, as soon as I did deleted it again (dpkg -P smartmontools) the > problem went away even with SMART enabled for the drive. Some hours > later the "ksoftirqd" started eating CPU like crazy but I havent tried > restarting the machine when SMART (not the daemon) was not enabled. > Will try that tonight. Are you quite sure the machine doesn't just have bad hardware? Please run memtest86 on it for, say, 12 hours and verify that there are no errors. -andy |