asked on

TFTP Oddity - Works, but doesn't

Hello,

This is strange. I have 2 subnet connected via a VPN tunnel between two Netopia R910 routers. The DHCP/TFTP server is on a Windows 2003 SBS on Network A. There is a PC on Network B that recieves its IP from the DHCP server on Network A (via DHCP relay on the R910 router). I can run tftp from the command line of the PC on Network B to the TFTP server on Network A and retrieve files all day long.

c:>tftp -i 10.0.1.10 get pxelinux.0 <--this works fine

However, if I try to boot a PC on network B using PXE, the pxelinux.0 file fails to get read. The TFTP log shows that the DHCP server tells the booting client to read the file, but then the read fails. I don't understand this. Is this a routing issue of some kind? The VPN tunnel does not NAT any traffic.

Any ideas appreciated.

Regards,
Aaron J. Wood

arnold

The issue is that your PXE booted system uses A's network IP while your test system uses B's network information.

You need to configure the DHCP on Network B to assign a local Network B IP/gateway, but set the tftp option to the Network A 10.0.1.10.

i.e. Network A has the 10.0.1.0/24
Network B has 192.168.0.0/24.
When the PXE broadcasts a BOOTP/DHCP packet, the current setup relays the requests to Network A and then relays back to the client a 10.0.1.145/24 gateway 10.0.1.1.
Network B router can not route this IP segment traffic as it is foreign to it. Additionally, access attempts by this client to a 10.0.1.x IP is attempted directly and will never be sent to 10.0.1.1 (router)

I believe normally if you have a boot server on a different segment, you have to use a local DHCP server to get the system on the network by assigning local segment IPs/gateway. You can use a DHCP relay, provided you have define the right scopes on the remote DHCP server to assign a network B IP when a request is coming from there.

aaronjwood

ASKER

Hmmm...let me explain my setup in more detail.

Network B (192.168.100.0/24) does not have a DHCP server. It consists of a Cable modem --> Netopia R910 Router (192.168.100.1) --> 2 PCs connected to router. The R910 has a VPN tunnel established to Network A's R910 router and relays DHCP requests from B to A. The VPN tunnel does not NAT or filter any traffic. Network A (10.0.1.0/24) has the DHCP/TFTP server (10.0.1.10). It consists of a ADSL modem (bridged) --> Netopia R910 Router (10.0.1.254) --> switch --> lots of PCs and DHCP/TFTP server connected to switch. The DHCP server is set up with a superscope that serves IP addresses to different subnets. For example...the PCs on network B DO IN FACT GET an IP within the 192.168.100.100 - 192.168.100.199 range. The DHCP server also defines the DHCP options 66 and 67 for boot server host (10.0.1.10) and boot file name (pxelinux.0). Therefor the traffic should not be foreign to Router B.

This issue continues to stump me. Are there other settings that need to be set in the DHCP options for the 192.168.100.0/24 scope? Is there any way to see where the packet is failing? Any other ideas?

Thanks

arnold

The only thing I can think of is that the BOOTP (options 66/67 are not making it through the relay). I think it is a two step process. Broadcast a request for IP on DHCP. Once there is an IP broadcast an ARP request for the BOOTP options. The PXE client gets the IP, but the ARP requests do not make it accross the VPN. Does your DHCP relay also relays BOOTP ARP requests to the server on Network A??

ASKER CERTIFIED SOLUTION

aaronjwood

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial