Failed to get IP VIA DHCP

[OmegaXis]

Hi all,

Currently running fog server in a ubuntu vm on a Proliant DL360p running unraid as the hypervisor. I am able to pxe boot vm’s within the server just fine. However, when I try to boot a thin client that is attached to the same cisco catalyst 2960g switch, i am getting the dreaded failed to get ip error.

Fog configuration info.
Fog V1.5.9
VM is Stock ubuntu 20.04.02

My network topology goes a bit like this
Meraki mx64 functioning as DHCP
Cisco Catalyst 3750-E as backbone
catalyst 2960g as endpoint switch where both the server and the computer I am trying to pxe are.

I have checked my configuration on the switches and STP is enabled.

Any suggestions?

right now, this is a lab environment test. The end goal is to create something that can be put into production at my company.

[Sebastian Roth]

@omegaxis said in Failed to get IP VIA DHCP:

I have checked my configuration on the switches and STP is enabled.

Either switch to the faster RSTP protocol or check of you can set the ports with clients connected to something called port fast - setting might bei called different on your switch. Essentially what this does is disable STP on particular ports because you know there is no other switch on this port that could cause a loop.

[OmegaXis]

@sebastian-roth I’ll have to check in about a week, sadly the server died today and i didnt have a backup as it was part of a home lab. The new one should arrive later this week/next

[OmegaXis]

The Plot thickens

I have 2 of these T510 Thin clients that im trying to get working. One of them works fine. The other one is still throwing the issues. I have made some configuration changes to the network, such as disabling stp.

[george1421]

@omegaxis said in Failed to get IP VIA DHCP:

I am able to pxe boot vm’s within the server just fine

when I try to boot a thin client that is attached to the same cisco catalyst 2960g switch, i am getting the dreaded failed to get ip error.

Meraki mx64 functioning as DHCP
Cisco Catalyst 3750-E as backbone
catalyst 2960g as endpoint switch where both the server and the computer I am trying to pxe are.

So lets think about what I’ve outlined. You can pxe boot a VM on the hypervisor just fine. But a physical machine attached to your 2960g can’t get ip error.

1. Is your meraki dhcp server issuing ip addresses to the VM, or is that coming from some other dhcp server?
2. If you plug a windows computer into the 2060 does it get an IP address?
3. Is the virtual switch where your FOG server is connected bridged to the physical network or is it Nat’d? (I don’t know unraid as a hypervisor to give directions to check)
4. We may want to get a pcap (packet capture) of the pxe booting process to know what is really going on. We can/should use the fog server to capture its side and then a witness computer with wireshark to capture the virtual side. I have a tutorial on using the fog server to grab a pcap. https://forums.fogproject.org/topic/9673/when-dhcp-pxe-booting-process-goes-bad-and-you-have-no-clue for wireshark you can use the capture filter of port 67 or port 68 if you don’t use a capture filter you can use a display filter of bootp. Capture both the what fog server sees and what the witness computer sees. My guess is the FOG server will not see the dhcp request.

Looking at the pcap you should see a typical DORA response.
Discovery (client sends)
Offer (server sends)
Request (client sends)
Ack (server sends)

If you look into the Offer packet, the dhcp server should set the {next-server} and {boot-file} fields in the ethernet header as well as dhcp options 66 and 67 in the options section. Both groups need to be there.

In my experience using a router or switch as a dhcp server does not give good luck because many do not handle the dhcp pxe boot settings well. Many for some reason will list the router/switch as the {next-server} even if they have a setting for a boot server. In this case you need the fog server IP configured in {next-server} and dhcp option 66.