On Wed, 15 Jul 2009, Troy Dawson wrote:

<snip>
>>  The problem is not related to the epoch. It occurs as soon as you install
>>  a kernel module via rpm which is 1) not in any repository (with the
>>  version you installed via rpm) 2) but have an older version in some
>>  repository.
>>  3) and then run "yum upgrade"
>>
>>  Another possibility to run into the same problem is:
>>  1) We have kernel module foo with version 1.42 in repository A and with
>>  version 0.42 in repository B. 2) We install the kernel module via yum.
>>  Version 1.42 will be installed.
>>  3) We remove version 1.42 from repository A - for some reason.
>>  4) We run "yum upgrade". Now the kernel-plugin of yum wants to install
>>  version 0.42, but 1.42 is already installed!
>
> You are correct.
> The plugin is at /usr/lib/yum-plugins/kernel-module.py
> Feel free to fix it.
> Troy
> p.s. Why in the world do you keep adding and removing all these kernel 
> modules from your repositories?  If you like a kernel module, keep using it, 
> don't just take it away.

From the description I'd expect the following to also cause a problem:

  standard sl repos have version 1.1 of the foo module

  whacky test repo has version 1.2 but isn't enabled by default (since it
  contains many whacky test things which clash with sl packages)

  yum --enablerepo=whackytest update kernel...foo...

  yum update

not that I've actually checked if this is would trigger an error 'cos and 
I tend not to install kernel modules from repos that I don't trust... :-)

-- 
/--------------------------------------------------------------------\
| "Computers are different from telephones.  Computers do not ring." |
|       -- A. Tanenbaum, "Computer Networks", p. 32                  |
---------------------------------------------------------------------|
| Jon Peatfield, _Computer_ Officer, DAMTP,  University of Cambridge |
| Mail:  [log in to unmask]     Web:  http://www.damtp.cam.ac.uk/ |
\--------------------------------------------------------------------/