Hello,
on OpenBSD if you put ddb.panic=0 into /etc/sysctl.conf, the server won't drop into debugger on kernel panic.
Is there please a similar setting for CentOS 6.4 / 64 bit?
The background is that I have a new dedicated server with Haswell CPU and once a month it is stuck, displaying kernel trace (the other users at my hoster have similar problems).
So the users of my web site complain.
I'd prefer my web server to just reboot
Thank you Alex
I've ended up doing this (hope it's valid for CentOS 6.4):
# echo 10 > /proc/sys/kernel/panic # echo "kernel.panic=10" >> /etc/sysctl.conf
On 23/08/13 04:19, Alexander Farber wrote:
I've ended up doing this (hope it's valid for CentOS 6.4):
# echo 10 > /proc/sys/kernel/panic # echo "kernel.panic=10" >> /etc/sysctl.conf
If your servers have a watchdog timer (most with IPMI/iLO/iDRAC/etc do), you can enable it. The OS will then have to kick the timer periodically to restart it's counter. If the OS panic's, it won't be able to reset the timer. Once the timer expires, it reboots.
Thank you, I have this
# dmesg|grep -i watch NMI watchdog enabled, takes one hw-pmu counter. iTCO_wdt: Intel TCO WatchDog Timer Driver v1.07rh
do you have any tips or doc pointers?
Ok, sorry - I've found the "man watchdog" and "man watchdog.conf"
I am not sure though, if I need to start the watchdog daemon at all - because I altready have these lines in my dmesg:
iTCO_vendor_support: vendor-support=0 iTCO_wdt: Intel TCO WatchDog Timer Driver v1.07rh iTCO_wdt: Found a Lynx Point TCO device (Version=2, TCOBASE=0x1860) iTCO_wdt: initialized. heartbeat=30 sec (nowayout=0)
Does it mean there is some "hardware watchdog" active at my CentOS 6.4 / 64 bit (Haswell CPU) server already?
Thank you for any hints Alex
-
Alexander Farber -
Digimer