On 10/25/2012 10:01 PM, Mike Burger wrote:
Then there's kernel updates, which require reboots, but which do not happen automatically upon installation...and if the system does not reboot, properly after that update, someone is going to have to lay hands on the keyboard.
I have machines with /etc/cron.weekly/newkernel with:
/usr/bin/yum list recent | /bin/fgrep -q kernel EXITVALUE=$?
if [ $EXITVALUE == 0 ]; then /sbin/reboot fi
exit 0
for automatic reboot Sunday morning if a new kernel was installed during the week.
It has worked for years.
On the other hand, I've seen other machines with CentOS 6 where the network had stopped working after the reboot. Physical access - or in my case HP's iLo - saved it by rebooting the previous kernel (it was a problem with some NIC's and VLAN).
I had done a test by booting the new kernel on the machine which is a mirror of this machine, without problems. They are identical - except for the network setup (no VLAN on the mirror).
Mogens