Hi:
I'm trying to setup an LTSP test enviroment and can not succeed not mather what.
In my network I have a dhcp box serving from 100 to 199 in a 192.168.2.0 subnet and acting as a a router as well. In a laptop running ubuntu, I have installed the LTSP daemon. The dhcp server running on ubuntu serves dynamically from 10 to 15 in the same subnet. The dhcp server in Ubuntu works fine to the point that a pc running windows gets the ip address and network information from the ubuntu dhcp server though it can't ping by hostnames( hostnmes are in the file /etc/hosts in the laptop where ubuntu is running 192.168.2.100) , in the ltsp server,the thin client running on the same subnet doesn't even try to boot ( Boot from LAN support for PXE), saying "system file not found or something". In yu Ubuntu server all the required services are running. See below:


-------------------------------------
This is my ifconfig:
------------------------------------
eth0 Link encap:Ethernet HWaddr 00:0A:E4:A2:63:F6
inet addr:192.168.2.100 Bcast:192.168.2.255 Mask:255.255.255.0
inet6 addr: fe80::20a:e4ff:fea2:63f6/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:4362 errors:0 dropped:0 overruns:0 frame:0
TX packets:3501 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:3036835 (2.8 MiB) TX bytes:490448 (478.9 KiB)
Interrupt:23 Base address:0x1800

lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
inet6 addr: ::1/128 Scope:Host
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:171 errors:0 dropped:0 overruns:0 frame:0
TX packets:171 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:14441 (14.1 KiB) TX bytes:14441 (14.1 KiB)

sit0 Link encap:IPv6-in-IPv4
NOARP MTU:1480 Metric:1
RX packets:0 errors:0 dropped:0 overruns:0 frame:0
TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:0 (0.0 b) TX bytes:0 (0.0 b)





------------------------------------------------------------------------------


This is my dhcp.conf:
--------------------------------------------------------------------------------
ddns-update-style none;
option domain-name "habana";
option domain-name-servers 192.168.2.1;
option subnet-mask 255.255.255.0;
option broadcast-address 192.168.2.255;
option routers 192.168.2.1;
ddns-update-style none;
default-lease-time 7600;
max-lease-time 7600;
option root-path "192.168.2.100:/opt/ltsp/i386";
filename "/2.6.9-ltsp-3/bzImage-2.6.9-ltsp-3";
lease-file-name "/etc/dhcp3/dhcpd.leases";
next-server 192.168.2.100;
use-host-decl-names on;


subnet 192.168.2.0 netmask 255.255.255.0 {
range 192.168.2.15 192.168.2.30;
option domain-name-servers 192.168.2.100;
max-lease-time 600;
default-lease-time 600;
}

# This is a very basic subnet declaration.





shared-network WORKSTATIONS{
subnet 192.168.2.0 netmask 255.255.255.0{
range dynamic-bootp 192.168.2.10 192.168.2.14;
}
}

group {

use-host-decl-names on;
option log-servers 192.168.2.100 ;

host ws015{
hardware ethernet 00:04:5F:82:75:9e;
fixed-address 192.168.2.15;
filename "/var/lib/tftpboot/2.6.9-ltsp-3/pxelinux.cfg";
}
}

if substring (option vendor-class-identifier, 0, 9) = "PXEClient"
{
filename "/2.6.9-ltsp-3/pxelinux.0";
}
-----------------------------------------------------------------------------------------

Can somebody take a look and notice if there is something I can't notice.

Any help greatly apreciated since I'm out of ideas and this is my first LTSP installation.

I've been running pxe booting on my network for a couple of years with no problems...Its not specifically ltsp but uses the same principals. I originally had problems and found it was all in my dhcpd.conf file. Been so long since I set it up I cant remember much but I'll post my dhcpd.conf so you can take a look. Maybe it will help....I've x'ed out any real ip info. I've hacked up my .conf file alot over time so please excuse any mess

Also, have you watched your log files as you try to get a network boot?
Not sure if I read your post correctly, but do you have one or two dhcp servers running? I'll assume you have one because you cannot have two dhcp servers on the same subnet....


shared-network Net0 {
option subnet-mask 255.255.255.0;
default-lease-time 7200;
max-lease-time 14400;

subnet 192.168.1.0 netmask 255.255.255.0 {
range dynamic-bootp 192.168.1.101 192.168.1.253;
option broadcast-address 192.168.1.254;
option routers 192.168.1.1;
option netbios-name-servers 192.168.1.14;
option netbios-dd-server 192.168.1.14;
option netbios-node-type 8;
option domain-name-servers x.x.x.x,x.x.x.x;
next-server 192.168.1.14;
# filename "pxegrub.tulip";
filename "nbgrub.rtl8139";
use-host-decl-names on;
option root-path "/opt/ltsp/i386";
# option dhcp-class-identifier "PXEClient";
option vendor-encapsulated-options 01:04:00:00:00:00;
option option-150 "/tftpboot/lts/grub.mnu";
}

subnet 192.168.2.0 netmask 255.255.255.0 {
range dynamic-bootp 192.168.2.101 192.168.2.253;
option broadcast-address 192.168.2.254;
option routers 192.168.2.1;
option netbios-name-servers 192.168.2.14;
option netbios-dd-server 192.168.2.14;
option netbios-node-type 8;
option domain-name-servers x.x.x.x,x.x.x.x;
next-server 192.168.2.14;
# filename "pxegrub.tulip";
filename "nbgrub.rtl8139";
use-host-decl-names on;
option root-path "/opt/ltsp/i386";
# option dhcp-class-identifier "PXEClient";
option vendor-encapsulated-options 01:04:00:00:00:00;
option option-150 "/tftpboot/lts/grub.mnu";
}
}
group {
use-host-decl-names on;
option root-path "/opt/ltsp/i386";
# filename "Ximage";
filename "nbgrub.rtl8139";
# filename "pxegrub.tulip";
# option dhcp-class-identifier "PXEClient";
option vendor-encapsulated-options 01:04:00:00:00:00;
option option-150 "/tftpboot/lts/grub.mnu";
next-server 192.168.1.14;

host 192.168.1.100 {
hardware ethernet 00:48:54:3D:B4:F2;
filename "nbgrub.rtl8139";
fixed-address 192.168.1.100;
}

host 192.168.1.51 {
hardware ethernet 00:48:54:3e:e0:f0;
fixed-address 192.168.1.51;
}

#host 192.168.1.103 {
# hardware ethernet 00:10:b5:0e:fc:99;
# filename "nbgrub.ne";
# fixed-address 192.168.1.103;
# }


}

Hi,

Actually I have two dhcp running on the same subnet. One serving from 99 to 200 and the dhcp server in the ltsp machine serving as seen in my posted dhcp.conf ( 10 to 50 ). At first it yielded "There is another dhcp running" but with the latest dhcp.conf both can run at the same time without conflicts(for testing my thin client the main dhcpd is dissabled).I have to rechek but I believe this is so. The dhcpd logs when booting the thin client remains still, notihing happens, that's why I'm suspecting something cuold be wrong with the pxe card.

What bothers me alot is the fact that the pc box receives all the ip information from the ltsp's machine dhcp server (including domain-name-server) but still it can not ping by hostname (ws001, ws002 and so on).
The other thougth that occured to me culd be the problem is the fact that i have the firewall in the router turned on.

I will take a detailed look at your post.
Thanks a lot for your help

Cheers

two quick suggestions:

1) mirror the good stuff from the dhcpd.conf on the existing DHCP server or preferrably set it up to give out the PXE info (can you make one server give out the info and tell the client to get the boot files off a different server?)

2) run a packet sniff (tcpdump, ethereal) on the network when the PXE client boots and see if it does the whole DHCP and PXE thing (request server, request address, accept address, gimme boot file, etc.)

Hi:
I did tcpdump on the dhcp machine and there is no request, nada, whatsoever form the thin client, may be definitely an NIC's issue. Firewall is not an issue either the data's comunication is inside the subnet.

No luck.I'll try to compile a new kernel for that box (Zevo's fanless pc, usb boot) and boot from there to examine further. I'll need help on that one surely, so I'l be back

Thanks a lot for your help

Take a look here for two dhcp-servers on the same subnet
http://wiki.etherboot.org/pmwiki.php...TwoDhcpServers

Have you tried tcpdump on the other DHCP server to see if it's trying to give out an address instead, or do you just take it completely offline during the thin client boot? If you are really not seeing anything in terms of PXE requests on the network I'd suggest sticking it on a crossover cable right into your LTSP/PXE/DHCP server and running tcpdump again to absolutly pin down that the client NIC isn't doing what it's supposed to.

FYI about the two DHCP servers on same network: As a rule of thumb the advice is that you can't have two DHCP servers on the same subnet. This is mainly because [when it's being done by someone who posts here, or being asked by someone less knowledgable in networking] it's usually implimented wrong and breaks something. This is not to say it can't be done and can't be done right, just that since most of the time when it shows up it's from someone who isn't knowledgeable enough in networking to get it right the first time.

It appears that you have yours set up so they won't conflict. Since something isn't working right though, it isn't a bad idea to eliminate the other DHCP server from the equation during troubleshooting.

Hi:
The primary dhcp server is dissabled.
The fanless pc is configured to boot first from LAN, then from an usb-flash memory.
It just booted from the usb-flash memory bypassing the LAN boot, it nicely asked for the ip and it got it from the ltsp-dhcp server, along with the router info, that made me think this is definitly an NIC's issue.
Where else to look for?

Also I have configured the dhcp server with the option:

option domain-name-server 192.168.2.100 ( ltsp-dhcp macchine, where /etc/hosts/ contains all the hostnames to ip maping)

However the linux-box doesn't recognize ws100 or any other hostname which are entried in the /etc/hosts in the ltsp-dhcp server. It can ping to any machine on the network so obviously routing is working OK. What will be the command for finding out which dns server the linux-box is using.

Thanks for your help
Cheers

I wish it was that simple. Unfortunatly, one system won't share name resolution from it's hosts file with another, you would have to set up a DNS server and put hosts in it's DNS config. You could specify another nameserver (the router for example if it does DNS forwarding or even the ISP's public nameservers) in the DHCP options on the ltsp server though.
Did you try a tcpdump while it was doing this? Did it do whatever PXE clients are supposed to do (sorry, I'm familiar with what PXE is, just not what it actually does on the network/tcpdump level) and ask for a boot file? Does the client show the PXE banner during POST, or immediatly go into USB-boot?

Hi:
I did tcpdump while booting bbut it goes straigth to usb booting, not banner or message else. tcpdump shows a request from the linux after it booted the from usb-flash,

Thanks for the tips about the dns issue. Good to know.

The nic card is a via chip, integrated in the mainboard, so maybe I could try to make an etherboot usb-boot kind of if I find the dirver?

Cheers

So if it's going straight to USB booting, then it sounds like an issue between the BIOS and the NIC. Similiar to selecting to boot off a floppy or CD or hard drive, the BIOS should support an option to boot from LAN. This should activate the PXE agent on the NIC to do it's thing. If this is it, the sad part would be that after all the server troubleshooting it comes down to the BIOS on the client being the culprit

The usb-boot is another solution, going a different route than a straight network boot. The way I understand it is that it's as simple as taking a bootable disk image and storing it on the right spot on the USB media (since the system BIOS obviously already knows how to boot off the USB device.)