Docker not executing in the jetson_inference directory on JetPack 6.2

Robotics & Edge Computing Jetson & Embedded Systems Jetson Orin Nano
,
1

Hey everyone, I’m having trouble opening Docker in the Jetson Inference folder, and this issue started after I updated my JetPack to 6.2 on my Jetson Orin Nano. I’ve been looking through the forums, but I’m having a hard time understanding what they’re saying. Is it possible that the Docker files are now in a different location? I’m a bit worried because I confidently recommended to my professor to buy 8 Jetson Supers for an AI workshop I’ve been working on. Any help would be greatly appreciated! Jetpack 6 & jetson inference

2

Hi ssdontletmed
Have you got a dustynv/jetson-inference docker image ?
sudo docker images
the tag will be r36.3.0
Because your jetpack is 6.2 it will look for image r36.4.0 (This image does not exist)
to run container
cd jetson-inference
docker/run.sh --container IMAGEID

e.g. (docker/run.sh --container 48d8c1f46455)
this will take you to the container

if you dont have jetson-inference image you need to

docker pull dustynv/jetson-inference

Hope this helps!

3

So, if I understand correctly, the Docker images I’m looking for may not yet exist for JetPack 6.2 but running docker/run.sh guides me to this unbuilt path, and I would need to use the Docker image that works with JetPack 6.0 if I decide to proceed. Also, I’m wondering if I can use a Docker image for JetPack 5.1.1 on JetPack 6.2, or would there be compatibility issues with some files?

Apologies for all these questions—I’m an electrical engineering student, so I’m still learning about this process.

4

The issue that persists is that i cant run docker/run.sh like before as shown "uafs@uafs-desktop:~/Downloads/jetson-inference$ docker/run.sh
ARCH: aarch64
reading L4T version from /etc/nv_tegra_release
L4T BSP Version: L4T R36.4.3
localuser:root being added to access control list
CONTAINER_IMAGE: dustynv/jetson-inference:r36.4.3
DATA_VOLUME: --volume /home/uafs/Downloads/jetson-inference/data:/jetson-inference/data --volume /home/uafs/Downloads/jetson-inference/python/training/classification/data:/jetson-inference/python/training/classification/data --volume /home/uafs/Downloads/jetson-inference/python/training/classification/models:/jetson-inference/python/training/classification/models --volume /home/uafs/Downloads/jetson-inference/python/training/detection/ssd/data:/jetson-inference/python/training/detection/ssd/data --volume /home/uafs/Downloads/jetson-inference/python/training/detection/ssd/models:/jetson-inference/python/training/detection/ssd/models --volume /home/uafs/Downloads/jetson-inference/python/www/recognizer/data:/jetson-inference/python/www/recognizer/data
DISPLAY_DEVICE: -e DISPLAY=:1 -v /tmp/.X11-unix/:/tmp/.X11-unix
docker: Cannot connect to the Docker daemon at unix:///var/run/docker.sock. Is the docker daemon running?.
See ‘docker run --help’.
" something i tried was to run “sudo docker pull dustynv/jetson-inference:r36.3.0” but after training ai i noticed that the program a hard time strating and closing up “python3 train_ssd.py --dataset-type=voc --data=data/recMAIN --model-dir=models/test_aone --batch-size=4 --workers=4 --epochs=100” which is part of the hello AI world tutorial

6

Hi,

Could you follow the below command to set up the docker?

Thanks.

7

image
image1920×1080 315 KB

After running the command and getting a positive result with the docker commands it still wouldnt run with docker/run.sh for some reason is it that the jetson inference tutorial is still not compatible with jetpack 6.2? and is there a way to revert my jetson nano to jetpack 5.2

8

Hi,

Jetpack 6.1/6.2 upgrade TensorRT from 8.x to 10.3 and remove several support that is required in jetson-inference.
If you want to it, you can try to setup the system with JetPack 6.0 / r36.3.0.

Here are the steps:

  1. Please download the r36.3.0 from below link:
    Jetson Linux Archive | NVIDIA Developer

  2. Reflash the device with the command according to your environment (sdcard/NVMe/USB)
    Quick Start — NVIDIA Jetson Linux Developer Guide 1 documentation

  3. Login to the device and install JetPack component with:

$ sudo apt-get update
$ sudo apt-get install nvidia-jetpack

Then try to setup the jetson-inference again.

Thanks.

9

Hey thank you so much I found that jetpack 6.0 is the better jetpack and everything works for me now