Red Hat Knowledgebase

Release Found: Red Hat Enterprise Linux 3 and 4

A machine that has frozen does not trigger netdump. With the help of other tools, a crash dump can still be captured. This article assumes that netdump has been configured properly.

I. SysRq

1. Enable sysrq by executing the command:

   echo 1 > /proc/sys/kernel/sysrq
   

2. An alternative way of doing this is to edit the /etc/sysctl.conf file. Change the kernel.sysrq and setting to:

   kernel.sysrq = 1
   

   In order to load the changes made to the /etc/sysctl.conf file without a reboot, execute the command:

   sysctl -p

Assuming that the machine is able to respond to keyboard interrupts when it freezes, a system crash can be forced by hitting the key combination: Alt-SysRq-c

This should initiate the crash dump.

II. nmi_watchdog

Sometimes, a machine can be so badly frozen that it will ignore keyboard interrupts. When this happens, the SysRq key combination is useless. The solution is to use nmi_watchdog.

Edit /boot/grub/grub.conf and add the option "nmi_watchdog=1" on the kernel line. For example:

kernel /vmlinuz-2.6.9-42.ELsmp ro root=LABEL=/ rhgb quiet nmi_watchdog=1

Reboot the machine for the changes to take effect.

When the machine freezes, nmi_watchdog will automatically trigger netdump. This is also recommended for remotely managed servers where keyboard access is not possible.