Re: Trouble updating kernel using RHN

[Rick Stevens <rstevens vitalstream com>]

Adrian Burd wrote:
> > > Well, I've been trying to track this down for a while now. Manually doing
> > > lilo worked fine, the kernels installed no problem at all. I have no idea
> > > why RHN/up2date failed...rather scary.
> > Well, now that I think about it, the kernel may be installed (binaries
> > in /boot, modules in /lib/modules/<kernel-version>), but /etc/lilo.conf
> > may NOT be updated automatically, nor will lilo be run.  
>
> Possibly, but it certainly looks from the up2date log file that it tried
> to update lilo, or at least try and test the new configuration.
>
> [Sun Feb 24 18:15:41 2002] up2date /boot/initrd-2.4.9-21enterprise.img was
> created
> [Sun Feb 24 18:15:41 2002] up2date Running "/sbin/mkinitrd --ifneeded
> /boot/initrd-2.4.9-21smp.img 2.4.9-21smp" 
> [Sun Feb 24 18:15:41 2002] up2date /boot/initrd-2.4.9-21smp.img was
> created
> [Sun Feb 24 18:15:41 2002] up2date renaming the lilo.conf entry for linux
> to linux.bak
> [Sun Feb 24 18:15:41 2002] up2date making a backup copy of /etc/lilo.conf
> as /etc/lilo.conf.up2date-1014592916.58
> [Sun Feb 24 18:15:41 2002] up2date writing out the new /etc/lilo.conf
> [Sun Feb 24 18:15:41 2002] up2date Running " /sbin/lilo -t -r /"  (lilo
> test mode)
> [Sun Feb 24 18:15:41 2002] up2date restoring the backup file
> /etc/lilo.conf.up2date-1014592916.58 to /etc/lilo.conf[Sun Feb 24 18:15:41
> 2002] up2date Error installing lilo.conf  The message was:
> test install of lilo failed
>
> When I ran /sbin/lilo -t -r by hand, it worked fine, not even the
> slightest hint of an error, which is why I find it slightly scary.

When you ran it, was it with the new kernel in the lilo.conf or was it
the original lilo.conf?  I'd assume the new version, since you're up
to snuff on this stuff.  I sure wish the up2date thing would've used
"-t -r -v" and saved the output.  Ah, well.

> I think I'll be going back to doing everything manually 8-). At least then
> when things screw up I know who to blame 8-).

Yeah, I always do it that way.  I don't trust "automatic" updates.
Never have.  Just look at the problems M$ is having with the XP auto-
update fiasco.


> > > Of course, now the thing is that XFree doesn't work with the new kernels, 
> > > giving an error apparently because it failed to initialize the NVidia
> > > kernel driver, but that's another matter.
> > Are there modules for the card that need to be "insmod"ed?  That's
> > weird.  As far as I know, the kernel doesn't initialize any video card
> > for X--that's X's job.
>
> That's what I always thought - unless I'm misunderstanding the log
> messages (it wouldn't be the first time).

I looked at your log.  The kernel isn't loading the modules, X is:

	(II) Loading /usr/X11R6/lib/modules/libvgahw.a
	(II) Module vgahw: vendor="The XFree86 Project"
		compiled for 4.1.0, module version = 0.1.0
	(**) NVIDIA(0): Depth 24, (--) framebuffer bpp 32
	(==) NVIDIA(0): RGB weight 888
	(==) NVIDIA(0): Default visual is TrueColor
	(==) NVIDIA(0): Using gamma correction (1.0, 1.0, 1.0)
	(--) NVIDIA(0): Linear framebuffer at 0xF0000000
	(--) NVIDIA(0): MMIO registers at 0xFC000000
	(EE) NVIDIA(0): Failed to initialize the NVdriver kernel
	(EE) NxVIDIA(0):  *** Aborting ***
	(II) Unloading /usr/X11R6/lib/modules/libvgahw.a

It looks like /usr/X11R6/lib/modules/libvgahw.a may have a problem.
(Remember that _kernel_ modules are in /lib/modules/`uname -r`)
----------------------------------------------------------------------
- Rick Stevens, SSE, VitalStream, Inc.      rstevens vitalstream com -
- 949-743-2010 (Voice)                    http://www.vitalstream.com -
-                                                                    -
-   The light at the end of the tunnel is really an oncoming train.  -
----------------------------------------------------------------------