/etc/nv_tegra_release missing from docker images

Accelerated Computing NGC GPU Cloud Docker and NVIDIA Docker
1

Hi There!

I am using the l4t base docker image from here: NVIDIA L4T Base | NVIDIA NGC

One thing I am finding is the /etc/nv_tegra_release file is missing when running the image:
image

Is there any way to determine the L4T version (of the container, not the host) when running the image?

Thanks!

2

Hi @pierce,

I hope you are doing well.

From the command you shared, 35.3.1 is the tag for the image corresponding to the l4t release.

Best regards,
Nico
Embedded Software Engineer at ProventusNova

3

Hi Nico,

Thanks for the prompt response!

Yes, I know when launching the image what L4T the version is, but I want to be able to determine it from within the container once it is running (ideally it would use the same mechanism as L4T running natively on a jetson product uses, by inspecting /etc/nv_tegra_release)

4

Hi @pierce,

I ran some test on my side. I executed the following command for running the docker on my Jetson OrinNX:

sudo docker run --runtime nvidia -it nvcr.io/nvidia/l4t-base:35.3.1 bash

In my case I do have file /etc/nv_tegra_release but it returns the L4T version of the host:

nvidia@nvidia:~$ sudo docker run --runtime nvidia -it nvcr.io/nvidia/l4t-base:35.3.1 bash
root@0ea15884e867:/# cat /etc/nv_tegra_release 
# R36 (release), REVISION: 4.4, GCID: 41062509, BOARD: generic, EABI: aarch64, DATE: Mon Jun 16 16:07:13 UTC 2025
# KERNEL_VARIANT: oot
TARGET_USERSPACE_LIB_DIR=nvidia
TARGET_USERSPACE_LIB_DIR_PATH=usr/lib/aarch64-linux-gnu/nvidia

This happens because the NVIDIA container runtime bind-mounts certain Jetson host files into the container.

However, there is no way to determine the L4T version from inside the container. The only reliable information regarding the L4T version inside the docker is the image tag.

Best regards,
Nico
Embedded Software Engineer at ProventusNova