In the event that Linux has a rare turn and decides to crash on you, you might want the machine to reboot if at all possible. The following will give you that ability. It should probably be paired with dumping any crash info out to a network server (hence the long pause – 300 secs), but very useful in any event for remote webservers.
Add the following entries to your sysctl.conf…
# vi /etc/sysctl.conf
kernel.panic_on_oops = 1
kernel.panic = 300
Then run these commands as root to enable these changes without needing to reboot:
# sysctl -w kernel.panic_on_oops=1
# sysctl -w kernel.panic=300
