DGX Spark July 2026 software update issue

Hi team,

I am trying to use DGX Dashboard to install system update, but it stucks here (I have clickupdate and auto reboot for more than 7 times)

any issue here?

@andybchen131 instead of the dashboard open a terminal and do the updates manually:

sudo apt update followed by sudo apt upgrade and if there are any held packages, finish the upgrade later or do sudo apt full-upgrade to force it. The dashboard will show available updates as long as there are packages on hold.

Thanks Elsaco

I tried your commands and verifying versions:

looks like nv-docker-options not install

can I install it manually?

dpkg -l | grep -E ā€œnv-docker-options|nvidia-system-extra|nvidia-system-station|nvidia-system-station-utilsā€
ii nvidia-system-station-apps 2404.26.01-1 all Meta-package for DGX Station apps packages
ii nvidia-system-station-common 2404.26.01-1 all Meta-package for DGX Station common packages
ii nvidia-system-station-core 2404.26.01-1 all Meta-package for DGX Station core packages
ii nvidia-system-station-games 2404.26.01-1 all Meta-package for DGX Station games packages
ii nvidia-system-station-localization 2404.26.01-1 all Meta-package for DGX Station localization packages

dpkg -s nv-docker-options

dpkg-query: package ā€˜nv-docker-options’ is not installed and no information is available
Use dpkg --info (= dpkg-deb --info) to examine archive files.

can you collect DGX dashboard logs for engineering to review? Please collect these logs, found in /var/log. Please collect as many as you see (e.g. dgx-dashboard-admin.log.1)

  • dgx-dashboard-admin.log
  • dgx-dashboard-admin.err.log
  • dgx-dashboard-reboot.log
  • dgx-dashboard-service.log
  • dgx-dashboard-service.err.log

Hi
I did the update today, here are logs (admin.error and service.err are empty)

dgx-dashboard-admin.log (12.9 KB)

dgx-dashboard-reboot.log (1.0 KB)

dgx-dashboard-service.log (7.7 KB)

and those still can’t get install

@andybchen131 Thank you for the logs. Do you have any other saved logs from previous boots. These would be files named dgx-dashboard-admin.log.1 and similar for other logs. This will help see your update history rather than just your current state.

This ask is for anyone else experiencing this issue as well

For me once I installed the new drivers from ASUS the final updates would take. I think there is a dependency in the software update on the vendor drivers that the software update does not take into account

Great that you found a fix!
Can we still get the logs so we can look into this from our end? Also can you link which ASUS drivers you needed to install?

Hi robert, when you mean drivers, do you mean the GX10 7.5.0-2 system package?

I have two ASUS DGX Sparks also. During a dashboard update, one shuts down while the other reboots, even though both have the exact same firmware and configuration. The one that shuts down is the head node. Any idea why?

I was setting up 2 new ASUS GX10s during the weekend. They both ended up in this state. I tried everything suggested in this thread, but Dashboard still shows the July 2026 update to be installed.

My older GX10s were updated successfully a while ago (they’d been incrementally updated over a few months). It’s just these 2 new ones going through the initial setup that’s stuck at this point.

At least some (or even maybe all) of the July 2026 update was actually installed, I think. 2 new systems show ā€œuname -rā€ as ā€œ6.17.0-1029-nvidiaā€ like the older ones.

For everyone experiencing this issue, can you please share with me the above logs, as well as any past logs that are numbered e.g. dgx-dashboard-admin.log.1

Thank you for checking in.

system-2.tar.gz (36.6 KB)

system-1.tar.gz (60.7 KB)

One of the systems after ā€œupdateā€ and reboot just now:

The systems otherwise seem to work fine so far. I ran them together with older systems in TP=4 for the latest DeepSeek V4 Flash and Nemotron 3 Ultra over the weekend. They were doing some evaluation work for me for a full day using Claude Code as harness.

Has anyone with the problem tried reinstalling using the OS image ASUS has on the support site and restarting from there? That was my next thing to try this weekend.

I will try that method today and let you know.

I take that back. My USB drives are at the office…

Thank you, engineering is working on the issue. I will report back when there is more info.

I had the same issue: no matter how many times I clicked on ā€œUpdateā€ on the Dashboard, the nvidia-firmware-580-580.159.03 kept showing up.

I suspect the problem is due to the update script trying to install package nvidia-firmware-580-580.159.03, but I already had nvidia-firmware-580-580.173.02 installed:

$ sudo apt list | grep nvidia-firmware | grep installed

WARNING: apt does not have a stable CLI interface. Use with caution in scripts.

nvidia-firmware-580-580.173.02/noble-updates,noble-security,now 580.173.02-0ubuntu0.24.04.1 arm64 [installed,automatic]
nvidia-firmware-580-580.95.05/noble-updates,now 580.95.05-0ubuntu0.24.04.3 arm64 [installed]

I resolved by manually installing nvidia-firmware-580-580.159.03 (with sudo apt install nvidia-firmware-580-580.159.03), rebooted, and now the ā€œUpdatesā€ tab shows the ā€œNo Available Updatesā€ message.

$ sudo apt list | grep nvidia-firmware | grep installed

WARNING: apt does not have a stable CLI interface. Use with caution in scripts.

nvidia-firmware-580-580.159.03/noble-updates,noble-security,now 580.159.03-0ubuntu0.24.04.1 arm64 [installed]
nvidia-firmware-580-580.173.02/noble-updates,noble-security,now 580.173.02-0ubuntu0.24.04.1 arm64 [installed,automatic]
nvidia-firmware-580-580.95.05/noble-updates,now 580.95.05-0ubuntu0.24.04.3 arm64 [installed]

Thanks! That did the trick for me, too.

I am seeing the same issue on an ASUS Ascent GX10, but my system has now progressed considerably beyond the versions discussed earlier in this thread.

The DGX Dashboard still offers the July 2026 update and specifically nvidia-firmware-580-580.159.03.

Current system:

dgx-release:                  7.6.0
kernel:                       7.0.0-1019-nvidia
NVIDIA driver:                580.178.04
nvidia-driver-580-open:       580.178.04
nvidia-kernel-common-580:     580.178.04
nvidia-firmware-580-580.178.04: installed
nvidia-spark-ota-check:       1.0.16-1
dgx-spark-ota-update-meta:    26.09.2

apt list --upgradable is empty.

I also completed the available ASUS/LVFS platform firmware updates. fwupdmgr get-updates now reports No updates available.

Current DMI firmware:

BIOS Firmware: 5.36_GX10DGX
FLASH:         SBP:R:2.152.15
EC Firmware:   3.3.2
PD0:           FW1 5.22 / FW2 5.7
PD1:           FW1 5.22 / FW2 5.7

However:

nvidia-spark-ota-check installed-name
-> OTA2604

nvidia-spark-ota-check torn-score
-> OTA2604, torn: 1

nvidia-spark-ota-check is-ota-available
-> available: true
-> name: July 2026
-> releaseDate: 2026-07-15
-> metadata: spark-ota-2607.json

The OTA2607 recipe is therefore still being offered even though the installed NVIDIA driver/firmware package is now 580.178.04, which is newer than the 580.159.03 package being offered by Dashboard.

I saw the earlier workaround in this thread of manually installing nvidia-firmware-580-580.159.03. I have not done that because the machine is now running the newer 580.178.04 stack.

Before installing an older version-specific firmware package simply to satisfy the OTA recipe, could NVIDIA engineering please confirm whether 580.159.03 should still be installed alongside 580.178.04, or whether this is now a stale OTA2607/Dashboard detection issue?

I can provide the DGX Dashboard logs requested earlier in this thread.

nvidia-dgx-dashboard-report-20261004-035115-REDACTED-V2.tar.gz (358.8 KB)
Attached is a redacted DGX Dashboard diagnostic bundle including the requested admin/service/reboot logs, OTA checker results, fwupd status/history, and current system/package information.

The system is currently on DGX 7.6.0 with NVIDIA driver 580.178.04, while the Dashboard/OTA checker continues to identify the installed OTA as OTA2604 and reports the July 2026 update as available.

Please let me know if you need any additional logs or diagnostics.