[TriLUG] CentOS 5.5: kickstart shmall and shmmax woes

[Kristopher Kane]

On Sat, Jan 8, 2011 at 11:00 AM, Lance A. Brown <lance at bearcircle.net>wrote:

>
>
> Brian Weaver said the following on 1/7/2011 17:36:
> > We use a tool to copy all the RPMS for the distribution into a single
> > directory. Once that is done the ISO is built from that single source
> > and the RPMs are served via HTTP from the single source too. Nope,
> > beyond a doubt it's using the same RPMs. If you compare the
> > %{name},%{version},%{release},%{epoch} from all the install RPMS in
> > the rpm database on an ISO and network install the sorted output is
> > identical
>
> Then it must be something anaconda is doing differently based on CDROM
> or network install.  But that doesn't make a lick of sense either.
>
> --[Lance]
>

I haven't found an answer yet but I am willing to bet there is some memory
caching going on during the network install.  The documentation doesn't tell
of any differences and  #centos at freenode.net didn't provide any help so I
will hit up the CentOS forums.

-Kristopher Kane