Bug#526406: (no subject)
I can confirm this bug, as I'm seeing the same type of crashes you are
seeing for about a month now. Our server is a BackupPC machine that moves
several hundred gb of data a night over ethernet, along with a lot of disk
activity.
System is a dual cpu, Opteron 265(2x2 cores) on a Tyan K8WE motherboard.
I'm running a custom compiled 2.6.29.1 kernel, so I suspect it's not a
debian kernel specific bug. I've run memtest for a week straight, along
with cpuburn, etc... no hardware problem has shown up yet. System is a
few years old, and has been stable until the last little while. I suspect
a cpufreq bug/problem as well.
Crashes usually take a few days to happen. Once it crashes, the machine
stays responsive to ping, etc, but will not accept logins. the kernel bug
log is almost exactly the same as the one posted in the bug report.
My cpufreq is currently set to conservative, i'll set it to performance
and see if it crashes again.
I can provide crash info and PCI devices/etc, if requested.
Mike
Reply to: