[Spacewalk-list] Kickstarting a RHEL5 install - better with updates included or not?

[spacewalk at epperson.homelinux.net]

I should already know this, but I can't find how to set up a kickstart
that installs with updates (rather than doing a "yum -y update" in post).

Thumbnail sketch, anyone?  I've been poking around and googling, and have
not found the magic bullet.

On Wed, November 21, 2012 11:59, Paul Robert Marino wrote:
> Actually the install from spacewalk with all the updates is cleaner
> because there is no chance an old package might have left artifacts
> behind. Although admittedly there are several schools of thought on this
> some prefer to do the updates manually others prefer the updates done in
> the install and there is still an other school of thought that if say you
> are rebuilding a host it should have the exact same rpm versions as the
> original and no additional updates. none of them are completely right or
> wrong its more of a matter of preference then any thing else.
>
>
>
> On Wed, Nov 21, 2012 at 9:47 AM, Snyder, Chris <Chris_Snyder at sra.com>
> wrote:
>> Looking for some opinions here.
>>
>>
>>
>> I’ve got SW 1.8 working with RHEL5 now (thank you, J. Pazdziora) and
>> have a kickstart profile uses three channels for initial package
>> installation: core RHEL5 packages (from the ISO), all current RHEL5
>> updates so when all is said and done, I have a host ready to roll with
>> no need for updates to be applied.
>>
>>
>>
>> Is this the best way to build a host?
>>
>>
>>
>> I don’t have any particular reason for this, but I have a gut feeling
>> that a better way to build a host might be to ONLY use the core RHEL5
>> ISO packages and the spacewalk-client packages for initial host
>> creation, then register the host with my RHEL5 update channel, and then
>> apply any needed updates (could be done in a %post section).
>>
>>
>>
>> The second option seems ‘cleaner’ from the stand point of it mimics
>> building a host from an ISO and then applying updates, whereas the
>> first does everything at once.  Theoretically the end result should be
>> the same.
>>