[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: yum kernel/kmod update weirdness



On Sat, 2007-07-28 at 16:12 -0500, Jima wrote:
> On Sat, 28 Jul 2007, Ralf Corsepius wrote:
> > Does anybody have an explanation for this?
> *snip*
> > This would cause the system to end up with this setup:
> > kernel-2.6.22.1-33.fc7
> > kernel-2.6.21-1.3228.fc7
> > kmod-nvidia-100.14.11-1.2.6.22.1_33.fc7
> > kmod-nvidia-100.14.11-1.2.6.22.1_27.fc7
> >
> > This doesn't seem right to me.
> 
>   Out of curiosity, which kernel is running?  2.6.21-1.3228.fc7 or 
> 2.6.22.1-27.fc7?
It had been 
kernel-2.6.21-1.3228.fc7.i686
+ kmod-nvidia-100.14.11-1.2.6.21_1.3228.fc7.i686

while
kernel-2.6.22.1-27.fc7.i686
+ kmod-nvidia-100.14.11-1.2.6.22.1_27.fc7.i686
already had been installed.

>   Off the cuff, I'd guessing the former, 
You are guessing correct.

> and yum is trying 
> to make sure it stays installed, but is also trying to make sure only two 
> kmod-nvidia packages are installed, but the logic is telling it to keep 
> the two latest installed instead of the one matching the current-running 
> kernel and the latest (like it's doing with the kernel).
>   Wow, did that even make sense? :-)
>   Just a shot in the dark.  (And yes, I would agree that the behavior is 
> non-intuitive.)
I don't know if the behavior you describe is what is implemented into
yum, but your explaination would explain what I observed.

The problem with this behavior: Should a kernel update be malfunctioning
users end up with a non-functional graphics system should they want to
"switch back to the known-to-work kernel+kmods".

Ralf



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]