Ubuntu repo signature invalid: EXPKEYSIG CD63F8B21266DE3C svc-workbench <svc-workbench@nvidia.com>

AI & Data Science NVIDIA AI Workbench
1

Please describe your issue or request:

The Ubuntu GPG package signing key is expired.

root@vroomfondel:~# gpg --show-keys ai-workbench-desktop-key.gpg 
pub   rsa3072 2024-12-12 [SC] [expired: 2025-12-12]
      52ECDF6597CFBD86D5BF4DB3CD63F8B21266DE3C
uid                      svc-workbench <svc-workbench@nvidia.com>

I checked also the key mentioned in the official documentation for a newer version:

pointing to https://workbench.download.nvidia.com/stable/linux/gpgkey, but that’s the same as the one already installed (pre-installed with DGX Spark/ASUS Ascent GX10).

Couldn’t find anything in here or elsewhere that mentioned a way to update.

Please tick the appropriate box to help us categorize your post
Bug or Error
Feature Request
Documentation Issue
Other

2 Likes
2

@cosinus - see post here for more information: DGX Spark apt update failure (EXPKEYSIG)

1 Like
3

Thank you for the fast fix! 😎

1 Like
4

The fix did not work for me - it still says the key is invalid.

5

Proposed fix doesn’t work for me either

6

Could you run a check via gpg --show-keys?

root@vroomfondel:~# gpg --show-keys /usr/share/keyrings/ai-workbench-desktop-key.gpg
pub   rsa3072 2024-12-12 [SC] [expires: 2027-12-13]
      52ECDF6597CFBD86D5BF4DB3CD63F8B21266DE3C
uid                      svc-workbench <svc-workbench@nvidia.com>

And just to make sure that it is not some kind of CDN problem:

root@vroomfondel:/tmp# curl -o newkey -fsSL https://workbench.download.nvidia.com/stable/linux/gpgkey
root@vroomfondel:/tmp# gpg --show-keys newkey
pub   rsa3072 2024-12-12 [SC] [expires: 2027-12-13]
      52ECDF6597CFBD86D5BF4DB3CD63F8B21266DE3C
uid                      svc-workbench <svc-workbench@nvidia.com>

To check what you get from that URL?

7

I have the same key. It looks like the proper key is there, it just isn’t recognized as valid despite it having the 2027-12-13 expire date.

8

@andrew.sheahen @youp

The linked forum post marked as the solution addresses the AI Workbench installation in DGX Spark and GB10 systems, which is configured slightly different.

If you installed the AI Workbench apt repo following the User Guide, you need to run a slightly different command:

curl -fsSL  https://workbench.download.nvidia.com/stable/linux/gpgkey  |  sudo tee /etc/apt/trusted.gpg.d/ai-workbench-desktop-key.asc

sudo apt update
1 Like
9

That worked, thank!