PXE boot fails on some subnets

[Irman Surani]

Hi,

I am having issues with PXE boot. The client and server are on different subnets. 'ip helper-address' have been configured on all the router interfaces to point to the server. The problem is, the client is able to boot and tftp the image from server successfully on some subnets but not others. On those that does not work, client is able to get ip address from DHCP server (which is different from PXE server) but error message "PXE-E32: TFTP open timeout" received. I suspect that the server is not receiving the broadcast when client is on the 'problem' subnet.

There are switches connected to the router and the client is connected to the switches. I suspect it may have something to do with STP but not exactly sure what it is.

Anyone can advise ? Thanks in advance.

Best Regards,
Irman

[DaGuru]

Hi Irman,

Can you provide any more information about your architecture? Such as, what kind of PXE server/solution, how you have your IP helpers configured, sample router/switch configurations, dhcp scope options for both working and non-working subnets, network topoolgy (including ST topology), and any details about the clients themselves, such as port settings for both the edge ports and client NICs would be very helpfull in visualizing your problem. Also, have you tried looking at the network traffic on the failing subnet with a sniffer to see what's going on?

What makes you suspect its a problem with spanning tree? What kind of routing are you doing; on what equipment, and what firmware?

When I've seen PXE timeout issues before it had to do with trying to push persistent route information down with DHCP when the PXE client didn't support it. I have also seen this problem before when edge ports were set to 100/full; PXE will tend to negotiate 100/half because it can only auto negotiate... (Just some food for thought.)


Lots and lots of questions, I know. :)
Dennis

[Andrew_291]

Lets try next solution

At the global "configure" context -

(config)# ip broadcast-forward

And at vlan-config context (by example vlan 5)
where the PXE clients are -

(config)#vlan 5
(config-vlan-5)#ip forward-protocol udp tftp
(config-vlan-5)#ip forward-protocol udp bootp
(config-vlan-5)# wri me

Hope, this helps.

[Irman Surani]

Hi Guys,

Thanks for responding to my query and my apologies for not replying earlier. I managed to find the cause of the problem.

It was due to the "broadcast limit" statement configured on the HP ProCurve 5300 series switch that are used for the subnet. Once I removed the statement on all the switches, the client manage to contact the server and boot successfully.

Best Regards,
Irman

[Carsten M]

Hi
I had the same problem. DHCP-server in the client vlan and Empirum-server in the server Vlan.
There has functioned the following configuration:
ip routing
ip directed-broadcast
ip udp bcast forward
vlan 1 (server Vlan)
ip forward-protocol udp 192.168.15.1 tftp (Empirum server)
ip forward-protocol udp 192.168.15.1 10042 (Empirum servers)
Vlan 11 (clients Vlan)
ip helper-address 192.168.15.255
Good Days

[Irman Surani]

Solution as per previous post.