Hi,
I am trying to follow the “Getting Started with AI on Jetson Nano” course and I am unable to get the docker container running.
I am attempting to run the following:
sudo docker run --runtime nvidia -it --rm --network host \
--volume ~/nvdli-data:/nvdli-nano/data \
--device /dev/video0 \
nvcr.io/nvidia/dli/dli-nano-ai:<tag>
where tag = <course_version>-<L4T_version>
I’m not sure what the course version is and I cannot seem to find it but I’m assuming it is the newest version “v2.0.2”.
My nvidia-l4t-core version is 32.7.4, which is newer than the corresponding latest docker tag “v2.0.2-r32.7.1”
Running ‘sudo apt-cache show nvidia-jetpack’ lists 4 jetpack versions: 4.6.4-b39, 4.6.3-b17, 4.6.2-b5, and 4.6.1-b110
I’m not sure why there are 4 jetpack versions but given that the oldest version, 4.6.1-b110, corresponds to the newest tag version, v2.0.2-r32.7.1, I do not understand what I am doing wrong or how to figure out what is wrong.
The full error I receive is:
Unable to find image 'nvcr.io/nvidia/dli-nano-ai:v2.0.2-r32.7.1' locally
docker: Error response from daemon: unauthorized: authentication required.
See 'docker run --help'
My internet connection is fine (ping google.com has 0% packet loss) and I tried running this both via ssh and directly on jetson nano terminal. (~30ms via ssh; ~20ms on nano)
I found a closed thread based on this same error: https://forums.developer.nvidia.com/t/jetson-nano-unable-to-find-image/169434
And ran the following based on the thread (updated the version number) :
sudo docker pull nvcr.io/nvidia/dli/dli-nano-ai:v2.0.2-r32.7.1
Re-ran the initial docker run command and received the same error as before.
Found another closed thread and tried:
sudo systemctl restart docker.service
Again, tried re-running the docker run command and still have the same error.
I’ve found one post about setting up an API key for NGC so I have one remaining troubleshooting step to go through but I have not seen that referenced in the other posts for this issue so if that does not work I’m pretty much out of ideas.
If anyone has other resolutions or considerations for correcting this I would greatly appreciate some help.