One of my systems have two onboard NICs which uses the e100 and e1000 drivers (yes, the interfaces are not the same). This system kickstart fine with CentOS 3.x. I recently tried to rekick it with CentOS 4.x but was unsuccessful in doing so. When kicking CentOS 4.1, the interfaces are swapped around, i.e. eth0 becomes eth1 and eth1 becomes eth0 (as described at http://marc.theaimsgroup.com/?l=linux-kernel&m=104888957013124&w=2). Is there a way to fix the interface during kickstart?
Unfortunately, because we have so many different types of systems, simply swapping the cables is not a good idea because it complicates management.
On Mon, 1 Aug 2005, sudo Yang wrote:
One of my systems have two onboard NICs which uses the e100 and e1000 drivers (yes, the interfaces are not the same). This system kickstart fine with CentOS 3.x. I recently tried to rekick it with CentOS 4.x but was unsuccessful in doing so. When kicking CentOS 4.1, the interfaces are swapped around, i.e. eth0 becomes eth1 and eth1 becomes eth0 (as described at http://marc.theaimsgroup.com/?l=linux-kernel&m=104888957013124&w=2). Is there a way to fix the interface during kickstart?
For real fun, add a PCI based dual port card. Then the first onboard NIC will probably be (but not for sure) eth2!!!
The short answer is there is no guaranteed way. The long answer is you may be able to add module loading commands to the kernel invocation line in your kickstart file and get the old behavior. Does not work for all servers.
I've got notes at work about it, but can't seem to find them from home.
------------------------------------------------------------------------ Jim Wildman, CISSP, RHCE jim@rossberry.com http://www.rossberry.com "Society in every state is a blessing, but Government, even in its best state, is a necessary evil; in its worst state, an intolerable one." Thomas Paine
I've seen that as well, even with RedHat 9.0. With Linux, because it's so configurable and flexible, I'm surprise there isn't an easy way to do this. I could swap the cable on a few systems to do the installation, but we have thousands of systems this is not feasible.
Anyway, thank you for the response. At least I know I'm not alone :)
On 8/1/05, Jim Wildman jim@rossberry.com wrote:
On Mon, 1 Aug 2005, sudo Yang wrote:
One of my systems have two onboard NICs which uses the e100 and e1000 drivers (yes, the interfaces are not the same). This system kickstart fine with CentOS 3.x. I recently tried to rekick it with CentOS 4.x but was unsuccessful in doing so. When kicking CentOS 4.1, the interfaces are swapped around, i.e. eth0 becomes eth1 and eth1 becomes eth0 (as described at http://marc.theaimsgroup.com/?l=linux-kernel&m=104888957013124&w=2). Is there a way to fix the interface during kickstart?
For real fun, add a PCI based dual port card. Then the first onboard NIC will probably be (but not for sure) eth2!!!
The short answer is there is no guaranteed way. The long answer is you may be able to add module loading commands to the kernel invocation line in your kickstart file and get the old behavior. Does not work for all servers.
I've got notes at work about it, but can't seem to find them from home.
Jim Wildman, CISSP, RHCE jim@rossberry.com http://www.rossberry.com "Society in every state is a blessing, but Government, even in its best state, is a necessary evil; in its worst state, an intolerable one." Thomas Paine _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Why not edit the config files in /etc/sysconfig/network-scripts?
Rename the two files to match the interfaces you want them to be, i.e. ifcfg-eth0 & ifcfg-eth1, and edit their contents to match the new file names. Then do a simple "service network restart" and they should be swapped around.
HTH
On 8/2/05, sudo Yang sudoyang@gmail.com wrote:
I've seen that as well, even with RedHat 9.0. With Linux, because it's so configurable and flexible, I'm surprise there isn't an easy way to do this. I could swap the cable on a few systems to do the installation, but we have thousands of systems this is not feasible.
Anyway, thank you for the response. At least I know I'm not alone :)
On 8/1/05, Jim Wildman jim@rossberry.com wrote:
On Mon, 1 Aug 2005, sudo Yang wrote:
One of my systems have two onboard NICs which uses the e100 and e1000 drivers (yes, the interfaces are not the same). This system kickstart fine with CentOS 3.x. I recently tried to rekick it with CentOS 4.x but was unsuccessful in doing so. When kicking CentOS 4.1, the interfaces are swapped around, i.e. eth0 becomes eth1 and eth1 becomes eth0 (as described at http://marc.theaimsgroup.com/?l=linux-kernel&m=104888957013124&w=2). Is there a way to fix the interface during kickstart?
For real fun, add a PCI based dual port card. Then the first onboard NIC will probably be (but not for sure) eth2!!!
The short answer is there is no guaranteed way. The long answer is you may be able to add module loading commands to the kernel invocation line in your kickstart file and get the old behavior. Does not work for all servers.
I've got notes at work about it, but can't seem to find them from home.
Jim Wildman, CISSP, RHCE jim@rossberry.com http://www.rossberry.com "Society in every state is a blessing, but Government, even in its best state, is a necessary evil; in its worst state, an intolerable one." Thomas Paine _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
On Tue, 2 Aug 2005, Kenneth Kalmer wrote:
Why not edit the config files in /etc/sysconfig/network-scripts?
Rename the two files to match the interfaces you want them to be, i.e. ifcfg-eth0 & ifcfg-eth1, and edit their contents to match the new file names. Then do a simple "service network restart" and they should be swapped around.
The issue is a bit more complicated. In many data centers, you have production lans and support lans. For security reasons, you are not allowed to do installations on the production lans. And for ease of maintenance, the jack labeled '1' is ALWAYS production and the jack labeled '2' is ALWAYS support. So you have a machine (or tons of them) that try to install one way with one rev of the OS and the other way with another rev of the same OS. And if they have additional network cards, (say 2 or 4 port), there are still more variants. It gets complicated trying to script it all out.
------------------------------------------------------------------------ Jim Wildman, CISSP, RHCE jim@rossberry.com http://www.rossberry.com "Society in every state is a blessing, but Government, even in its best state, is a necessary evil; in its worst state, an intolerable one." Thomas Paine
you can use netdevice=(eth*) and/or ksdevice=(eth*) in the kernel command line when kickstarting.... you can also set the MAC in /etc/sysconfig/network-scripts/ifcfg-eth* to whatever port you want... ... thanks
- doug
************************************************************************ Doug Zeman \ \ | __ \ ____ | Phone: 408 7747674 CAD Sys Eng II / \ | _/ | | | | Cell : 408 7187466 CA MicroProc Div ____ \ | | | | | | Fax : 408 7747811 Sunnyvale, CA _/ __| _| _____/ ____/ | doug.zeman@amd.com ************************************************************************
sudo Yang wrote:
One of my systems have two onboard NICs which uses the e100 and e1000 drivers (yes, the interfaces are not the same). This system kickstart fine with CentOS 3.x. I recently tried to rekick it with CentOS 4.x but was unsuccessful in doing so. When kicking CentOS 4.1, the interfaces are swapped around, i.e. eth0 becomes eth1 and eth1 becomes eth0 (as described at http://marc.theaimsgroup.com/?l=linux-kernel&m=104888957013124&w=2). Is there a way to fix the interface during kickstart?
Unfortunately, because we have so many different types of systems, simply swapping the cables is not a good idea because it complicates management. _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
You can specify the Hardware (MAC) address in the network config (ifcfg-ethn) for each card to prevent this from happening, especially if the cards use the same driver....
P.
Doug Zeman wrote:
you can use netdevice=(eth*) and/or ksdevice=(eth*) in the kernel command line when kickstarting.... you can also set the MAC in /etc/sysconfig/network-scripts/ifcfg-eth* to whatever port you want... ... thanks
- doug
Doug Zeman \ \ | __ \ ____ | Phone: 408 7747674 CAD Sys Eng II / \ | _/ | | | | Cell : 408 7187466 CA MicroProc Div ____ \ | | | | | | Fax : 408 7747811 Sunnyvale, CA _/ __| _| _____/ ____/ | doug.zeman@amd.com
sudo Yang wrote:
One of my systems have two onboard NICs which uses the e100 and e1000 drivers (yes, the interfaces are not the same). This system kickstart fine with CentOS 3.x. I recently tried to rekick it with CentOS 4.x but was unsuccessful in doing so. When kicking CentOS 4.1, the interfaces are swapped around, i.e. eth0 becomes eth1 and eth1 becomes eth0 (as described at http://marc.theaimsgroup.com/?l=linux-kernel&m=104888957013124&w=2). Is there a way to fix the interface during kickstart?
Unfortunately, because we have so many different types of systems, simply swapping the cables is not a good idea because it complicates management. _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
-
Doug Zeman -
Jim Wildman -
Kenneth Kalmer -
Peter Farrow -
sudo Yang