I don't think anything like that has been reported to us formally with the GX240  but one of the things to watch out for with the Optiplex line is that their primary design goal, right from the ground up, is to meet a manufacturing cost limit. So, there are two probable causes.

One possibility is that there's a problem in the system BIOS which is affecting things when you CD boot; it's hard to prove, and if that is the case it's not really fixable easily unless there happens to be a firmware update.  The other common problem which many of the Optiplex models have is that to save a cent or two in manufacture, they don't have controllable IRQ assignments.

The GX270's, for instance (more of which exist than probably any other model) have the IRQ lines for the SATA controller and the Network Interface card literally physically wired together. In most PCs, manufacturers have several programmable arrangements which map the PCI-bus IRQ lines down into the legacy PC-AT PICs, because classic DOS drivers have always assumed that one IRQ = one device (IBM designed a system for breaking this assumption with in the Micro Channel design for the PS/2, years before PCI took the possibility mainstream, but it wasn't for many years later that this IRQ sharing became important in the DOS world). The GX270 doesn't.

The problem here is that the initial reference drivers for the hardware Dell integrated (the SATA host adapter and a Broadcom NIC) were written by the device manufacturers without contemplating the need for IRQ sharing; Dell then made the two devices incompatible when used with the manufacturer drivers simultaneous by just wiring them together and not making them able to be programmably revectored (manufacturing cost limit).

Resolving this was not fun. GX270's with SATA drives would run like molasses when GhostCasting, because of the IRQ conflict, and for a long time no drivers were available for the NIC with the necessary magic to work around this problem, which was basically unique to the Dells and not any other chipset licensee (Broadcom's standard drivers now do, and I have to give Broadcom props generally for their driver availability).

GX745s manufactured in the US region had a similar problem (the motherboard sound shared a hardwired IRQ and the BIOS left the motherboard sound generating IRQs constantly on some GX745s), and it's possible that the GX240 has yet another instance of this style of defect.

Now, in general for PXE-bootable machines the UNDI network driver built into the BIOS tends to be a motherboard-specific driver rather than a generic one for the NIC chipset, and so if you're using an NDIS or packet driver switching to UNDI (or vice versa, in some cases) is one thing to try, and in one of the UNDI templates you have the option of using a tool called IRQCFG.EXE

This little doodad (which you can find in the Ghost Boot Wizard driver library) can as discussed here report on some of these conflicts, and if the hardware doesn't support unsticking them it'll say so.

[ If the IRQ conflict can be resolved by reprogramming the IRQ mapping from the BIOS default, then it can be loaded as a device driver instead of run as an app and it'll replumb things, which is what the UNDI (irqcfg) choice of template does. ]

The other thing you can also try if you really want to use a boot CD is to create a WinPE environment with the WAIK and use the Windows PE version of Ghost, Ghost32.exe; it's not ideal but when you have hardware that isn't really DOS-compatible it's the last resort.

El_Pescador wrote on Jun 16^th, 2008 at 6:47pm:

Heh. Considering the number of problems that particular model has, consider yourself lucky.

To give Dell their due though, their Precision-branded gear really is the mirror image of the Optiplex line - the (early build, still on BIOS rev A01) Precision M90 I use is, aside from weighing a shoulder-wrenching amount for a notebook, absolutely a mighty piece of gear. It's got every bell and whistle and every one of them works, and works well; no BIOS problems, zero incompatibilities with Ghost in DOS, the BIOS boots from all manner of devices at full USB 2.0 speed, it does everything and does it fast and does it well.

Of course it cost an eye-watering amount (my manager at the time had brought it for himself, but then IT told him it wasn't on the approved list and he wasn't allowed to use it on their network so ... hello, developer machine) but I have no hesitation in saying that those extra dollars really were buying gear of a matching level of quality.

I have had ten or twelve off-lease Dell Optiplex GX270's (a mix of Small Mini-Tower, Small Desktop, and Small Form Factor) pass through here so far this year.  Uniformly, they all have been been outfitted with the Socket 478 Pentium 2.80GHz processor, albeit some are Northwoods and others are Prescotts.

I routinely employ the 250GB Seagate FreeAgent Desktop USB external HDD on each and every GX270.  Whenever I set out to make "disk-to-image" files, I enjoy 400-700 MB/min speeds in USB2 whether using Norton Ghost Version 8.2 on a Reatogo-X-PE CD or Norton Ghost 2003 running in MS-DOS on a two-disk set of floppys.

Are you certain you are achieving operations in USB2 mode when employing an USB external HDD?

El Pescador