DGX's vicious update cycle

All, what are some tips to get out of the continuous update loop? I update from DGX Dashboard, restart, it is offering updates again.

I could potentially resolve this using apt-get update β†’ full upgrade, but my understanding is this is not the recommended way to update the Spark.

It does take 2-3 reboots until all updates are applied. It should be much better afterwards. You could try a shutdown, disconnecting the spark from power and the power brick from the power socket for 2-3 minutes, then starting again.

Thank you, I will try again when I’m back home later, even though I did go through several reboot cycles already.

I thought I read that it is recommended to disconnect the power brick to get better power, but I must say that I do not understand why?

There is some bug in USB power delivery (Spark is USB-C powered), and unplugging the brick resets the controller inside it so it could negotiate power properly. At least that’s the working theory.

Thanks again all, but none of these ideas helped.

I now followed the steps in the Manual to manually update. This is the output… and the result is still the same. DGX Dashboard keeps offering updates.

Any other ideas?

$ sudo apt dist-upgrade
[sudo] password for xxxxxx:
Reading package lists… Done
Building dependency tree… Done
Reading state information… Done
Calculating upgrade… Done
The following package was automatically installed and is no longer required:
linux-nvidia-6.11-tools-6.11.0-1014
Use β€˜sudo apt autoremove’ to remove it.
The following packages will be REMOVED:
linux-modules-nvidia-fs-6.11.0-1014-nvidia linux-tools-6.11.0-1014-nvidia
0 upgraded, 0 newly installed, 2 to remove and 0 not upgraded.
After this operation, 943 kB disk space will be freed.
Do you want to continue? [Y/n]
(Reading database … 348694 files and directories currently installed.)
Removing linux-modules-nvidia-fs-6.11.0-1014-nvidia (6.11.0-1014.14) …
Removing linux-tools-6.11.0-1014-nvidia (6.11.0-1014.14) …
$ sudo fwupdmgr refresh
Updating lvfs
Downloading…             [********************                   ]
Successfully downloaded new metadata: 3 local devices supported
$ sudo fwupdmgr upgrade
Devices with no available firmware updates:
β€’ MZALC4T0HBL1-00B07
β€’ UEFI dbx
Devices with the latest available firmware version:
β€’ Embedded Controller
β€’ UEFI Device Firmware
╔══════════════════════════════════════════════════════════════════════════════╗
β•‘ Upgrade UEFI Device Firmware from 0x00000500 to 0x00000507?                  β•‘
╠══════════════════════════════════════════════════════════════════════════════╣
β•‘ This update improves the performance and stability of the USB Power          β•‘
β•‘ Delivery Controller in DGX Spark for USB-C Ports                             β•‘
β•‘                                                                              β•‘
β•‘ NVIDIA_DGX_Spark must remain plugged into a power source for the duration    β•‘
β•‘ of the update to avoid damage.                                               β•‘
β•šβ•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•β•
Perform operation? [Y|n]: Y
Waiting…                 [***************************************]
Successfully installed firmware
Do not turn off your computer or remove the AC adapter while the update is in progress.
An update requires a reboot to complete. Restart now? [y|N]: y

Broadcast message from root@xxxxxx on pts/1 (Sun 2026-03-15 19:45:03 GMT):

The system will reboot now!

Connection to 192.168.1.50 closed by remote host.
Connection to 192.168.1.50 closed.

(Also, the above manual process, when repeated, gives exactly the same result every time. So it seems the update from 0x00000500 to 0x00000507 is actually not performed.)

Okay, got a bit more info after digging into the history of the update:

$ sudo fwupdmgr get-results
[sudo] password for xxxxxx:
0.	Cancel

xxxxx ((null))
xxxxx (Embedded Controller)
xxxxx (MZALC4T0HBL1-00B07)
xxxxx (UEFI Device Firmware)
xxxxx (UEFI Device Firmware)

xxxxx (UEFI dbx)
Choose device [0-6]: 5
UEFI Device Firmware:
Device ID:            a6c6bxxxxxxxxxxxxxxx9c4
Previous version:     0x00000500
Update State:         Failed
Update Error:         failed to run update on reboot: expected 0x00000507 and got 0x00000500
Last modified:        2026-03-15 20:00
GUID:                 dxxxxxxxxxxxxxxx6
Device Flags:         β€’ Internal device
β€’ Updatable
β€’ System requires external power source
β€’ Supported on remote server
β€’ Needs a reboot after installation
β€’ Device is usable for the duration of the update
β€’ Signed Payload

So it looks like the device, most likely USB-C controller, is not updating correctly @Nvidia, do I need a proprietary tool to update it?

Same issue. Any update on this problem?

Try a complete power off and drain of the spark. This means:

  • shutdown and power-off
  • unplug the USB-C power adapter from the Spark
  • unplug the USB-C power adapter from the wall oulet
  • press and hold the Spark’s power switch for 10 seconds
  • wait 5 minutes before reconnecting and powering on
  • check for new firmware updates

You may need to do this 2 or 3 times.

Suggestion:
sudo apt update
sudo apt dist-upgrade
sudo fwupdmgr refresh
sudo fwupdmgr upgrade
sudo reboot

It will reboot several times and take ~10 mins to come back. Always prefer apt dist-upgrade over apt upgrade