https://bugzilla.novell.com/show_bug.cgi?id=680687 https://bugzilla.novell.com/show_bug.cgi?id=680687#c2 Matthias Andree <matthias.andree@gmx.de> changed: What |Removed |Added ---------------------------------------------------------------------------- Priority|P5 - None |P0 - Crit Sit Component|Other |Kernel AssignedTo|mhrusecky@novell.com |kernel-maintainers@forge.pr | |ovo.novell.com --- Comment #2 from Matthias Andree <matthias.andree@gmx.de> 2011-03-29 15:02:37 UTC --- Resetting Component to kernel - as this is likely a kernel memory or buffer or cache or I/O handling bug. dmesg given above. I have removed md from the picture to be sure it's not the culprit, and mount /dev/sd?5 directly as /. Evidence: - memtest86+ as well as a memtest=3 kernel command line option pass memory checks without complaints - booting into vanilla 2.6.36, I can run mysqlrepair -A and get "OK" on all tables, whereas with SUSE's 2.6.37.1-1.2-default, I consistently get complaints that there are rows in bacula.File where the auto_increment column is 0, duplicate key records, and even after mysqlrepair -A, mysqlcheck errors persist. With 2.6.36, mysqlrepair succeeds on the first run. - compiling a 2.6.37.6 vanilla kernel with SuSE's 2.6.37.1-1.2-default (from an XFS file system mounted off a sata_via driven SATA disk drive), I sometimes get intermittent failures, and upon inspection with vim, I see binary garbage in a file. Rebooting fixes this condition. - in-memory file corruption happens on XFS and ext4 file systems. It is usually cleared on a reboot. - openSUSE up to and including 11.3 have been rock solid on the same hardware, issues correlate in time with the upgrade to 11.4. Apparently it's not mySQL's fault. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.