Carsten Aulbert wrote:
[...]
>
> The problem lies with the initramfs which is probing all found NICs
> simultaneously with DHCP (when ip=dhcp is given) and then the fastest or
> slowest reply survives in boot.log.
>
> To solve this one could
> o disable all but one MAC for DHCP, but that's not really a good
> solution when doing a full cluster reinstall, since a lot of DHCP server
> tables need to be modified and servers restarted
> o create a special hook/script taking care of this issue
> o start playing with ip=::::... again
>
> Probably we will take the latter route *sigh* and I was so happy to get
> away from the custom chboot setup we currently have :(
>
> Or is there any other way to achieve this?
if you have several nodes with identical hardware setup (like identical
motherboard with
2 onboard ports, an additional dual-port NIC always in the same slot) you could
add
an udev rules file to nfsroot and initrd (using a initramfs-tools/hooks
script), where you
label the devices with pci slot id's. You use entries like
ACTION=="add", SUBSYSTEM=="net", BUS=="pci", ID=="0000:12:04.1", NAME="eth0"
ACTION=="add", SUBSYSTEM=="net", BUS=="pci", ID=="0000:12:04.0", NAME="eth1"
...
Then you could always pxe boot via eth0.
We use these udev rules files in our FAI config space, so it was a short step
to also use them for installation -- and it did work. However, since we only
use DHCP
on our mgmt network and static ip configuration for the "data network", we
switched
back to the default config plus the live patches.
Bye, Ralf
--
Ralf Utermann
_____________________________________________________________________
Universität Augsburg, Institut für Physik -- EDV-Betreuer
Universitätsstr.1
D-86135 Augsburg Phone: +49-821-598-3231
SMTP: ralf.uterm...@physik.uni-augsburg.de Fax: -3411
