Hello,
today one of the servers I administer crashed for the third time due to
kernel panic. Folks at colocation site provided me with the following
"screenshot": http://miha.krneki.org/netfilter/kernelpanic.jpg
I'm not sure whether this is related with netfilter/iptables, tg3 or
something else, but the screenshot is full of netfilter messagess, so I
tried this list first.
Funny thing is that /proc/sys/kernel/panic was set to 60 seconds, yet
the machine did not reboot itself after the crash and colocation people
said machine was not responding to keyboard, so they could not provide
the whole trace.
Another interesting thing - machine is heavily loaded (apache+php+mysql)
- pushing around 35-45mbit/s througout the whole day. After midnight
utilization gets very low and yet all 3 crashes occured at around 2AM
local time. There are no cron scripts at that time that would cause any
cpu/network hammering, system is virtually idle.
Server is Dual Opteron 250 with 4gb ECC RAM and Adaptec 2015S SCSI
controller (2x RAID1).
At the time of crash, system was using:
OS: (32bit) slackware-based with:
glibc 2.3.6 (NPTL), gcc 3.3.6,
iptables 1.3.4 (upgraded to 1.3.5 after crash)
kernel 2.6.17.8 with the following patches:
- grsecurity 2.1.9 (200608012135)
- dpt_i2o-2.5.0-2426 driver from adaptec
SMP and PAE are enabled, config file is located at:
http://miha.krneki.org/netfilter/config-2.6.17.8+d1
If anyone has any clue, don't hesitate to contact me.
Cheers
--
Miha
