[CentOS] kickstart problems

Thu Sep 4 12:58:16 UTC 2008
Les Mikesell <lesmikesell at gmail.com>

John R Pierce wrote:
> Paolo Supino wrote:
>>   On the other hand if you were right about it than RHEL/CentOS/Fedora 
>> installation would be unsuitable in any multihome configuration 
>> because it would map ETH devices differently (albeit once in a while) 
>> which means one whould have to swtich the cables because of network 
>> device remapping!!! and that isn't something users and corporations 
>> that use REHL (and there are many of those) would be willing to live 
>> with :-)
> 
> 
> (please PLEASE trim quoted articles to just what you're commenting on, 
> like I have above).
> 
> 
> I've /never/ seen RHEL/CentOS or any of its predecessors renumber 
> ethernet ports on a working system..   I've seen it number them 
> backwards, such that eth0 was the port labeled '1' outside the chassis 
> and eth1 was port '0', but it was extremely consistent about this (one 
> specific case of this I remember is  RHEL2.1 or 3 on a Intel SE7501WV2 
> dual xeon board).   I've had a pile of different RH linux configurations 
> running on various servers for 10 or more years.

The behavior changed when the system started using udev.  Devices are 
detected in parallel in more or less random order now.  However, the MAC 
address of each NIC is normally stored in the corresponding 
/etc/sysconfig/network.scripts/ifcfg-ethxx file and they are renamed to 
match the device specified in the files as they are activated. 
Kickstarting is a special case since these files don't exist yet, but 
you can specify ksdevice either by mac address or as bootif, meaning the 
interface where the pxe boot happened, according to:
http://kbase.redhat.com/faq/FAQ_80_531.shtm

-- 
   Les Mikesell
    lesmikesell at gmail.com