Seems this kernel is once more a headache version
Last night I had to upgrade manually a server from a friend. He was experiencing heavy load, many processes and no reason for this. The server was pretty new installed. After some digging on several forums I ended on a Bugzilla from Redhat. . Basically it had the same symptoms as kernel 2.6.9-34.. I decided to jump to a newer version.
I went to this nice kernel vault, choose me the one I needed for the cpu, a little wget for kernel and kernel-devel (don’t forget this one or you have no compilers to upgrade other parts). Then it is simple if you know
I downloaded the parts to root level, but any place is ok.
to install :
>rpm -ivh kernel……. (that long name with a lot of numbers) and let it finish, might take some time
once finished
>rpm -ivh kernel-devel-……. and again let it finish
Your kernel is now installed but… nope not yet finished. It is not loaded and won’t till you reboot.
Mostly you use grub to set the default kernel at startup, but if you leave it as is and reboot, and kernel doesn’t work, you’re stuck unless you can push the button manually.. This might be the way to follow. Next time you reboot, the actual kernel reloads… else you’ll need restore cd to get in and reset grub to another kernel and that might be long for a data house for example.
go to /boot/grub/, open grub.conf…
somewhere you’ll see a line with default =0 (this is the top mentioned kernel and this should be the latest installed one)
This is the one we don’t want as default in case it isn’t working properly, we set default=x (x=number of the actual working kernel, just read a little down and count which one it is in the list, most probably 1)
save with ctrl x
now we go into grub, type following
grub
press enter
it’ll open another level of terminal within grub, now we’ll tell grub to boot once with first kernel, just once
type
>savedefault –default=0 –once
>quit
and you’re back in normal terminal
You’ve done the hard part. You’re now ready to reboot the box. It will load the new kernel and if you’re lucky, it is all ok and it comes live after a few minutes.
If not, push the reboot button or send in a ticket to dc to reboot. Normally next reboot will load previous kernel again. You can go back to square one and start over with another kernel.
I do advise to search around on several forum to find a stable kernel. Latest isn’t always the best one. And like with all software, from time to time a version is a misser.
And an advice, if you don’t feel comfortable, don’t do it, certainly not with a production server. A crashed machine can take some time to recover. And if you decide to go for it, make complete notes of what you do, in case it can help the person if it fails.
Always be careful, you’re changing lowest level of the box, all the rest depends on kernel, if you miss it… you have a dead duck. This is a guide on how I did and this doesn’t mean it is perfect for each setup. For me this worked on 4 server upgrades I did, both Pentium and AMD machines….

great info, I think I’ll digg it for ya
thanks, I hope it will be succesfull for others since there isn’t to much about upgrading a kernel on a live box.. yum etc are easy but you can’t always use it.