Steve Traylen
2004-Aug-02 10:50 UTC
[syslinux] pxelinux.0 fails when ip address matchs nnn.nnn.nnn.n0n.
Hi
First we are using pxelinux.0 from syslinux-2.09 and out hardware is
from lspci
03:01.0 Ethernet controller: Intel Corp.
82541EI Gigabit Ethernet Controller (Copper)
We booted some 200 machines successfully over the last few days
using pxelinux.0 and a root over NFS that requires kernel level
IP configuration.
The only failures were hosts with IP address
> lcg0469.gridpp.rl.ac.uk 130.246.182.100
> lcg0470.gridpp.rl.ac.uk 130.246.182.101
> lcg0471.gridpp.rl.ac.uk 130.246.182.102
> lcg0472.gridpp.rl.ac.uk 130.246.182.103
> lcg0473.gridpp.rl.ac.uk 130.246.182.104
> lcg0474.gridpp.rl.ac.uk 130.246.182.105
> lcg0475.gridpp.rl.ac.uk 130.246.182.106
> lcg0476.gridpp.rl.ac.uk 130.246.182.107
> lcg0477.gridpp.rl.ac.uk 130.246.182.108
> lcg0478.gridpp.rl.ac.uk 130.246.182.109
>
> lcg0569.gridpp.rl.ac.uk 130.246.182.200
> lcg0570.gridpp.rl.ac.uk 130.246.182.201
> lcg0571.gridpp.rl.ac.uk 130.246.182.202
> lcg0572.gridpp.rl.ac.uk 130.246.182.203
> lcg0573.gridpp.rl.ac.uk 130.246.182.204
> lcg0574.gridpp.rl.ac.uk 130.246.182.205
> lcg0575.gridpp.rl.ac.uk 130.246.182.206
> lcg0576.gridpp.rl.ac.uk 130.246.182.207
> lcg0577.gridpp.rl.ac.uk 130.246.182.208
> lcg0578.gridpp.rl.ac.uk 130.246.182.209
if we take the second one above as an example what happened
was that the NIC requested the correct IP address (130.246.182.101)
from the DHCP.
pxelinux.0 printed out that the hosts IP was
82F6B665, 130.246.182.11
While the hex number is correct the IP address is wrong and 0 was
stripped out. This happened for all other hosts matching nnn.nnn.nnn.n0n.
We were using
default linux
label linux
kernel lcfg/kernelboot-2.4.20-20.7
ipappend 1
append root=/dev/nfs nfsroot=....
This then caused a failure since the ipappend put a ip= with the wrong
address on the kernel.
Changing to
label linux
kernel lcfg/kernelboot-2.4.20-20.7
append root=/dev/nfs nfsroot=.... ip=dhcp
and now everything is working for fine for us. We have a work around but
I thought I would report the problem we saw as I'm interested in what is
going on. I guess pxelinux.0 interrogating the NIC is going wrong somewhere.
Steve
--
Steve Traylen
s.traylen at rl.ac.uk
http://www.gridpp.ac.uk/
Apparently Analagous Threads
- CESA-YYYY:NNN <level> CentOS 4 axp <package> - <short explanation>
- Sorta of "quota" needed -> "max file size = nnn ( bytes )"
- DO NOT REPLY [Bug 6597] New: Add Kib/s Mib/s indicators to last line message: sent NNN bytes...
- Using firewall module, how do I clear iptables before rules are applied?
- SteveTraylen - new member of the configmanagementsig