View Full Version : Removing a new kernel

8th March 2008, 07:50 PM
I upgraded my F8 box today, and found that with the new kernel ( I think), I'm unable to boot my machine.
I can boot in the previous kernel, but this gives me all sorts of problems with the X server crashing on startup, which I assume is because of the kmod-nvidia driver being for the new kernel (hopefully).

My question is though, is it pretty much safe to remove the newest kernel, and if so, should I do it through yum or rpm, or doesn't it make any difference? Hopefully, it should sort out dependency issues for itself with the nvidia driver, so I can get the previous kernel running as well as it did this morning.

I've tended to shy away from kernel stuff in my linux time, and only removed an old 1 manually once, using rpm -e. Apologies if it's a dumb question, but I'd rather be safe than sorry..


8th March 2008, 10:09 PM
You've got it right with the 'rpm -e'. That's the correct procedure for removing a kernel. Not sure if you'll be able to set up the kmod-nvidia with the old kernel though. Firewing1's the man who creates them and perhaps he can get you set up if you find there's problems.

9th March 2008, 11:47 PM
Nigel, I updated 3 fc8 boxes last night and two failed to boot with the new kernel. To go back to the previous one I did the following:

(Do make sure still have the previous kernel installed (verify this in your /etc/grub.conf))

yum remove kernel- kernel-devel-

This was enough on my plain vanilla VIA EPIA-M1000 box. On my game machine with the NVIDIA drivers installed it was a bit more complicated. The above command also removed the nvidia drivers that went along with this kernel, so manually go to the livna.org repository and downloaded the drivers for your kernel. My previous kernel was, so I downloaded:


and installed these using yum. Now I am back up and running on the older kernel, and I am trying to find out why I had a 66.7% failure rate, and why noone else is complaining more loudly.

My only guess is, one machine was a via epia-m1000, and the significance about the other is an encrypted swap partition.

The epia machine had a hard crash while booting (spitting registers on the screen) and the nvidia machine locks up after the [udev] message when its analyzing the partitions.


Flak Pyro
10th March 2008, 05:44 AM
haha well ill complain, my sound card no longer works on this new kernel ICH8 (965 chipset) so now i either have to wait or roll back to the old kernel :(

So your not the only one haha

10th March 2008, 09:33 AM
Thanks a lot for the help guys.
My machine is now free of that dratted kernel.

Will just wait for a newer one to arrive via yum I think, and in the mean time, make do without 3D acceleration.


10th March 2008, 06:15 PM
Maybe the Fedora users are a patient bunch - well that's not a bad thing.

I was a bit frantic on Saturday night when I urgently needed a PC and neither Fedora machine would boot. (had to rush to a Windows pc - lol). Although I do always keep a Fedora live and a Puppy linux disc to hand! Anyway, on my laptop, Network manager is now completely broke as is wireless networking.

Reminds me of a pact I made with myself some time ago not to automatically accept updates - and certainly not to update all machines at the same time. Just broke my own rule!!

At the moment I am booting into the old kernel and that works fine - really waiting to see what the Fedora developers have to say. I'm sort of assuming a repair to the situation will be forthcoming?

Hopefully - Tonypm

10th March 2008, 06:57 PM
Hi Guys,

I too had have this problem - see my post in the installation section.
I too use Epia Via an ML 6000ea board.
Is there a theme


12th March 2008, 02:57 AM
Yeah, I made the same mistake, I usually update one at the time, not all machines at once.
Anyway, feel free to add to the fedora kernel test log, I already did...

kernel- Test Log (https://admin.fedoraproject.org/updates/F8/FEDORA-2008-2103)

Makes you wonder how come this isn't addressed a bit more forcefully. Oh well, I updated my yum.conf with the line:

exclude=kernel* kmod-nvidia* xorg-x11-drv-nvidia*

12th March 2008, 06:50 AM
You should have a /boot/grub/grub.conf file

If you want to boot "Linux - new kernel" change the number after "default" at the top of the page to 1. Should be 1 and none other else.

If you want to get rid of "Linux - old kernel" remove the stanza with it from the file -- but remember that to boot the new kernel automatically then you'll have to have have 0 after "default".

Hope this helps !

"Technology Will Make Things Possible"

27th March 2008, 12:42 PM
Hello Everyone,

Have now upgraded to kernel and everything works perfectly.
Well done you boffins