MSI Edge Expert Nvme devices not found

Accelerated Computing DGX Spark / GB10 User Forum DGX Spark / GB10
1

My firm has bought an MSI Edge Expert which is based on NVIDIA DGX Spark it seems and it was working fine until it didn’t. The device just booted up to a screen stating

Checking media presence…
Media present…
Start PXE over IPv4 on MAC: xxx

Then timed out to the BIOS
and when I checked there in the NVMe configuration, there were no NVMe devices found. Can you help us?

2

Hi @santhosh.kumar3,

Thanks for reaching out. Since the BIOS is not detecting any NVMe devices, the system has lost visibility of its storage drives entirely — hence the PXE boot fallback.

Because the MSI Edge Expert is an MSI OEM product, NVIDIA’s usual remote diagnostics path isn’t available here. Here’s what we’d recommend:

Step 1 — Run an NVMe self-test from a USB live environment

Since the system can’t boot from the internal drives, you’ll need to boot a Linux live USB (e.g. Ubuntu) to reach the drives if they’re still electrically present:

  1. Boot from a USB live image.

  2. Install nvme-cli:

    sudo apt install nvme-cli

  3. List detected NVMe devices:

    sudo nvme list

  4. If drives appear, run a short self-test on each (e.g. /dev/nvme0):

    sudo nvme device-self-test /dev/nvme0 -s 1

    sudo nvme self-test-log /dev/nvme0

  5. Check SMART data for errors:

    sudo nvme smart-log /dev/nvme0

If no drives appear under nvme list even from the live USB, the drives have either physically failed or have lost their physical connection — this is not recoverable without hardware intervention.

Step 2 — Contact MSI for RMA

If the drives are invisible to the OS as well, please contact MSI support directly for hardware service or RMA. As this is an MSI-branded product, MSI is the responsible party for hardware warranty and repair:

When contacting MSI, share:

  • The full BIOS screenshot showing no NVMe devices detected
  • The output of nvme list from the live USB (even if empty)
  • Any events that preceded the failure (power loss, OS update, etc.)

Let us know what you find in Step 1 — if the drives do appear from the live USB, there may be more we can do on the firmware/BIOS side.

3

Hey there, quick update, there’s an easier way to do this than my last message suggested. There’s an NVME test within the UEFI on the DGX Spark version (Docs: Advanced Tab — DGX Spark UEFI User Guide), there should be something similar or identical in the MSI version.

Thanks!

4 
Checking Media Presence...
Media Present...
Start PXE over IPv4 on MAC xxx

I just received my replacement Spark, and on first power on, I got this message too. It didn’t give me an option to do anything. As I was searching and found this post, I saw it rebooted (or continued), and then went into the initial setup.

I’m no wondering whether this was normal expected behaviour on first boot, or whether this might suggest something flaky with the internal nvme drive? (the “Media Present” line sounds like it was there, but then I don’t know why it would’ve stalled on this screen for so long?).

(Mine is an nvidia FE)

Edit: After initial setup, it restarted and has been showing an nvidia logo with a green progress bar that’s been moving very slowly, and has jumped back to the start at least once.

Edit2: Now it’s showing that message again. I think I’ve someone traded a mostly faulty one, for a totally faulty one.

Edit3: It has finally booted into the desktop. Maybe it just needed 10 reboots during first setup? I’ll see how it goes.

Edit4: The original message seems to be just that th enetwork was set as highest boot priority (and I’m using a cable). The constant restarts have also stopped now, so I’m assuming it was just the first-setup installing multiple rounds of updates. Think it’s good now. But I do need to run the field diagnostics to ensure it doesn’t overheat like the last one!

5

Have you checked the boot order in BIOS, two of my four FE modeld were set the PXE boot first from factory.

6

lol yeah, I edited that in to my post afterwards. I don’t remember that being the case for the previous one, but I can see that it could be handy for companies. I’ve swapped that around and no longer see this. And the boot loops stopped - I think it was more likely just first-time-setup stuff rather than issue. Running the field-diagnostics now just to ensure I don’t have the issue from the previous one before I start setting it up again.

Thanks :)

7

Glad it worked, in my case it didn’t break anything, just added 60s to the bootup time and as I was using the kit headless from the start it took a month or so before I noticed.