Core24 track request for DCGM

store-requests
1

Hi :wave:

I would like to ask to create a new track named core24 for the dcgm snap.

In the future, when we change the base to core26, it might break for users using Ubuntu series lower than 26.04, so we want to have channels that points to which core the snap is using. Moreover this will avoid users breaking because it was using latest/stable.

Thanks

2

Hi, Gabriel!

Per Process for aliases, auto-connections and tracks 2, we need a 1-week voting/discussion period, so I’ll check back on the discussion and votes in a few days.

I have two questions before casting my vote.

  1. What’s SNAP’s release cadence, how often is a new major version (potentially requiring a new track) released? is this documented somewhere by upstream?
  2. Is there some commitment from upstream on the maintenance of old versions? e.g. is core24 will be supported with security updates? will it continue to be supported now that core26 is out, and for how long?

Thanks for your request!

3

Hi Alex.

  1. What’s SNAP’s release cadence, how often is a new major version (potentially requiring a new track) released? is this documented somewhere by upstream?

We think that new tracks will be necessary every two years and it’s highly correlated with the Ubuntu kernel cycle

snap DCGM ships the datacenter-gpu-manager package that needs NVIDIA drivers to properly work. NVIDIA drivers are closely related to the Ubuntu kernel because they interact directly with the kernel to enable hardware acceleration and manage NVIDIA GPUs.

Each version of the NVIDIA driver is compatible with specific versions of the Linux kernel. When the Ubuntu kernel is updated, the NVIDIA driver module must also be recompiled or updated to ensure compatibility. If there’s a mismatch between the kernel and the driver version, the GPU may not function properly, leading to issues like lack of graphics acceleration or even system crashes.

Our concern is that if users from the dcgm SNAP do a series upgrade from Noble → Next Lts Release that the snap can stop working.

  1. Is there some commitment from upstream on the maintenance of old versions? e.g. is core24 will be supported with security updates? will it continue to be supported now that core26 is out, and for how long?

I don’t know if I understood this correctly. The DCGM has their own package release. In case a new security update lands, we would need to change the git tag for the latest release and do some functional tests to see if still works for the GPUs in a certain Ubuntu release. This is a little bit agnostic of the Ubuntu core that is used. So for example a new security patch lands on DCGM, we would need to change the git tag on all tracks (core24, core26 and etc).