From Bugzilla Helper: User-Agent: Mozilla/4.0 (compatible; MSIE 5.01; Windows NT 5.0) Description of problem: Using the kernel-smp-2.4.7-0.12.1.i686.rpm. And I receive the following panic under high load. Here is the following information that is displayed on the console before the machine reboots from the crash. >Process mysqld (pid:30064, stackpage=d1f45000) > >Code: 81 7b 04 ed 1e af de 74 13 68 af 00 00 00 68 c0 be 29 c0 e8 > >kernel BUG at /tmp/31322-i686/BUILD/kernel->2.4.7/linux/include/akm/rknl >spiBUoc > >kth:11mp<4>/3invlaid operand:0000 > >CPU:1 >EIP: 0010:[<c011eba7>] >EFLAGS:00010082 > >Process klogd (pid:480, stackpage=f5fdd000) > >Code: 0f 0b 58 5a c6 05 f8 6b 2b c0 01 57 9d 5b 89 f0 5e 5f c3 8d > This machine is a high load machine. This machine prior to upgrading the kernel was crashing approximately once a day, after a call to tech support, I switched my ethernet module from the e100.o module to the eepro100.o, via their suggestion. After using this module my machine stays up for roughly a week and then crashes again. I then called tech support back and it was suggested that I attempt to use the development kernel listed above, at the beginning of this message. I installed the development kernel on 8.11.2001 and the server has crashed approximately 6 times between now and then, unfortunately this is the only message I was able to get from the console. Mysql seems to run fine for a few days and then begins spawning many mysql child processes until the load reaches a max if we restart the mysqld this will temporarily fix the problems and then mysql will run fine for a few days and then the child processes will start to creep up again. I am not certain though that this is the source of the crash, due to problems with nic's earlier. You can examine recent messages on my machine by visting the following urls. real-time tail -4000 of zeus log with a grep -v for "running" and "starting" to not show the logging of the restarting of the virtual servers which are about 95% of the log. http://www.softwaresidewalk.com/zeuslog.php B info on php, mysql and kernel versions http://www.softwaresidewalk.com/phpinfo.php B real-time echo of dmesg http://www.softwaresidewalk.com/dmesg.php B real-time tail -200 of mysql's hosts.err log http://www.softwaresidewalk.com/srv12.err.php B real-time tail -1000 of /var/log/messages http://www.softwaresidewalk.com/messages.php B cpu info from cat /proc/cpuinfo http://www.softwaresidewalk.com/cpuinfo.php netstat -asl http://www.softwaresidewalk.com/netstat1.php netstat -a http://www.softwaresidewalk.com/netstat2.php If there is any other information I can provide please let me know, immediately. Version-Release number of selected component (if applicable): How reproducible: Didn't try Additional info: Haven't tried to reproduce it, due to it reproducing itself frequently enough.
It would be interesting if the successor of the 2.4.7 kernel you are running will fix this: http://people.redhat.com/arjanv/testkernels/i686/ has the 2.4.7-2.18 kernel which has a few eepro100 bugs fixed (and several other bugs), and has survived a lot of stress-testing in our lab so far.