Optiplex 3020 TFTP Issue

[A Former User]

Hello All,

We’ve been using FOG fairly successfully for a while now until the past two days where we cannot get some fresh Dell Optiplex 3020’s to get to PXE boot. We’ve tried many different things, but the 3020’s we tried all time out at:

tftp://10.1.0.1/default.ipxe…Connection Timed Out

(Doing this from memory, so that may not be the exact wording on the error)

I should point out that 10.1.0.1 is not the IP for our FOG server, but rather it is the router for our network, which does not do DHCP, nor is it specified anywhere in the FOG configs that I’ve checked. FOG is also working just fine with all other computers including a new Dell Latitude laptop we got last week.

[B]System Details:[/B]
FOG version - 1.2.0 Rev 2993
OS - Ubuntu 12.04 LTS desktop
This is a virtual machine running on VMWare ESX 5.5 Update 2 with 4 GB memory and 100GB disk space allocated
DHCP - DNSMASQ, using undionly.kpxe
FOG kernel - I tried 3.19.2, 3.19.1, and 3.18.2 (? can’t remember the exact numbering for the kernel below 3.19.1)

[B]Optiplex 3020[/B]:
OS - Windows 7 64-bit
BIOS - A06 (Also tried A05 and A07)
Secure boot - disabled
UEFI - disabled

I tired to update/downgrade the BIOS on the computer, we tried plugging directly into the switch, I’ve tried several different kernels and I tried to update to the latest revision (3186 I think?), but the install hung at “Downloading New FOG Client”, so I had to revert back to a snapshot I took yesterday. I’ve started and restarted services to the end of the Earth and I’ve hunted through every config file I could think of, with no luck.

I will not be able to test with the workstations until tomorrow, but I can remote into the server and try to grab some information if needed. I also am open to the fact that this could just be a hardware issue with the batch of desktops we got, so I’ll just have to work around that if that turns out to be the case.

Thanks very much for any help!

[Junkhacker]

have you had any other computers work while connected to the same switch as these new computers that are failing. I’ve never had any problems with Optiplex 3020’s and fog.

[A Former User]

Yes, we’ve imaged at least two other computers on that same switch although they were both Dell Latitude laptops. The switch in question is a crappy Dell 8 port, that has a Cisco VoIP phone and a couple of other things plugged into it. Usually we just use the port on the back of the phone, but we tried using the switch directly with no dice.The Dell switch is plugged into a Cisco 3560 switch in our closet.

[Sebastian Roth]

I’d guess it would be best to get into the details (packet capturing) before I start guessing about what could be wrong. Could you please connect a hub in between the client and the FOG server (e.g. cisco, crappy dell switch, hub, optiplex) and capture the traffic using another computer or laptop on the hub? Display filter in wireshark ‘bootp’. You are welcome to upload the packet capture file if you want us to look at it…

[A Former User]

Well unfortunately, we do not have any hubs on hand to do this, but I’d welcome some guesses!

[Tom Elliott]

The fact that it’s getting tftp://10.1.0.1/default.ipxe tells me that the tftp server is located at 10.1.0.1 (or the dhcp server thinks it is).

This sounds similar to sourceminer’s issue with a Kerio dhcp server. What’s the contents of the Option 66 for your Dhcp server. This may change, but essentially which DHCP server is handing out 10.1.0.1.

[A Former User]

We are using DNSMASQ for the DHCP with FOG and it is set to use undionly.kpxe and the FOG server IP at 10.1.0.53

[Sebastian Roth]

[quote=“Jack S., post: 44818, member: 29223”]Well unfortunately, we do not have any hubs on hand to do this, but I’d welcome some guesses![/quote]
Guessing with this is like finding drugs in a house. Not very efficient doing this without having one of these trained dogs with you… My trained dog in network hickups is wireshark/tcpdump.

If you don’t have a hub you can either just connect your capture laptop to the switch as some DHCP servers talk broadcast all the time (so you see all the traffic anyway) or use tcpdump on the for server where your dnsmasq DHCP is running:

[CODE]sudo tcpdump -i eth0 -w dhcp_dump.pcap udp[/CODE]

You’ll only see one message from tcpdump. All the packets captured are written to a file. After you started up one of your faulting clients you can stop tcpdump (ctrl+c), get that file from your server (e.g. winscp) and check it out with wireshark or even upload it to let us do that.

Are you absolutely sure that there is no other DHCP server or DHCP proxy on the net???

[A Former User]

Okay I can give tcpdump a shot when I get in the office later today.

[QUOTE]Are you absolutely sure that there is no other DHCP server or DHCP proxy on the net??? [/QUOTE]
I am fairly certain that there isn’t any DHCP server that can hand out PXE information. We do have DHCP running in our Windows domain, but it was never touched while we setup FOG, nor has the Windows DHCP ever been configured for PXE

[A Former User]

[SIZE=3][COLOR=#000000][FONT=Times New Roman]Okay I have some captures to take a look at and I’ll upload them here as well. I have a capture from a machine that can successfully boot into FOG (a Dell Latitude E5540) and one that does not (one of the Optiplex desktops in question). Its very curious because from just looking over the captures both seem to TFTP from the right server (10.1.0.53), but the Optiplex times out. [/FONT][/COLOR][/SIZE]

[SIZE=3][COLOR=#000000][FONT=Times New Roman]Virustotal link for my ZIP file: [url]https://www.virustotal.com/cs/file/e8529017b59fa10f49b4005fbe84432977b3e3a32163cc15e39b024f3604f227/analysis/1427906958/[/url][/FONT][/COLOR][/SIZE]

[url=“/_imported_xf_attachments/1/1843_DHCP_DUMP.zip?:”]DHCP_DUMP.zip[/url]

[Wayne Workman]

Does the Optiplex 3020 have solid state drives?

And, out of curiosity, can you enable the long POST check on one? Like, a full memory count before booting… or something similar?
I read something the other day that claims this can help newer/faster systems to PXE boot right… Again, Just a curiosity.

[A Former User]

The hard drives are standard HDD, any ideas on how I would enable a “long” post check?

[davido38]

Hi,

we have some optiplex 380, 390, 3010 and 3020 and no problem with PXE. Just the 3020 is very slow to start.

[Wayne Workman]

I just checked on my Optiplex 7010,

My particular BIOS does not allow Long-POST, or Through-POST.

I know older systems used to allow configuring fast POSTing and full POSTing…
Maybe that’s gone to the wayside???

Either way, you don’t know till you just get in there and look.

[A Former User]

This post is deleted!

[A Former User]

[quote=“davido38, post: 44849, member: 29185”]Hi,

we have some optiplex 380, 390, 3010 and 3020 and no problem with PXE. Just the 3020 is very slow to start.[/quote]

Do you happen to know which version of FOG/which kernel you are running?

[quote=“Wayne Workman, post: 44850, member: 28155”]I just checked on my Optiplex 7010,

My particular BIOS does not allow Long-POST, or Through-POST.

I know older systems used to allow configuring fast POSTing and full POSTing…
Maybe that’s gone to the wayside???

Either way, you don’t know till you just get in there and look.[/quote]

I just checked my BIOS option and there was barely anything there for POST options so I think I’m SOL with that path

I’m thinking this is definitely a hardware issue with this particular batch of computers (8 in total), but I’m holding on to hope

[Tom Elliott]

So the Cisco VoIP phone is an indicator to me. The TFTP requests that VoIP phones require for their configuration is a know potential problem. What’s more concerning is the crappy 8 port switch with this VoIP phone is only seeming to affect the 3020s. Does this somewhat match what you’re seeing? Is there a way to find out what the VoIP system is looking at for their configuration?

[A Former User]

[quote=“Tom Elliott, post: 44855, member: 7271”]So the Cisco VoIP phone is an indicator to me. The TFTP requests that VoIP phones require for their configuration is a know potential problem. What’s more concerning is the crappy 8 port switch with this VoIP phone is only seeming to affect the 3020s. Does this somewhat match what you’re seeing? Is there a way to find out what the VoIP system is looking at for their configuration?[/quote]

The VoIP system gets its configuration information from a separate subnet. All end devices (PCs, laptops, servers) are on 10.x.x.x while the VoIP system is on 172.16.x.x. What gets even weirder is we’ve been able to FOG without issue for months now through the VoIP phones built in switch. Now not all desks have the 8-port switch to worry about, so maybe tomorrow I’ll try a port straight from the wall and see where that gets us.

[A Former User]

Okay, so bad news: using a cable straight from a wall jack into the computer made no difference. The OptiPlex 3020 still tries to grab default.ipxe from 10.1.0.1 instead of 10.1.0.53. Even worse news: I just tried to image a Dell OptiPlex 380 and while it could boot the FOG menu (using the same wall jack as the OptiPlex 3020), PartClone did not start and it just sat at a black screen for about 10 minutes. I also tried to create a task through the FOG web console and the task just sat at “Queued” even if I tried to force it to start. I’ve restored the FOG VM to a what I thought was a stable configuration before we even tried to mess with the 3020’s, but it looks like I need to go back further.

[Wayne Workman]

What is running DHCP in your environment? Do you have access to it?