Setting Jupyter as a service to autostart

I’m following the getting started with AI on Jetson nano course and having difficulty connecting to Jupyter while my nano is running under in headless machine mode. I can access jupyter locally by running

jupyter notebook

and after that is running I can access jupyter remotely by going to http://ip:8888 but I need to have started jupyter locally. According to the guide, JupyterLab server should be running automatically but doesn’t appear to be. Is there anyway to modify this?

I also do not have a classification folder as referenced [url]Sign In

Hi reedshorton,

My best guess is that you may be using the standard Jetson Nano image, rather than the special image created for the DLI course.

Did you use the SD card image provided in the DLI instructions under “Setting up your Jetson Nano”? This should include the Jupyter Lab server that spawns on boot.

Best,
John

Yes I did use the standard image, but then followed the steps here [url]https://github.com/NVIDIA-AI-IOT/jetbot/wiki/Create-SD-Card-Image-From-Scratch[/url]. Is there any way to download parts of the notebook?

If you did these things:

...
python3 create_jupyter_service.py
sudo mv jetbot_jupyter.service /etc/systemd/system/jetbot_jupyter.service
sudo systemctl enable jetbot_jupyter
sudo systemctl start jetbot_jupyter
...

Then jupyter should start on boot. You can then connect to it with your browser via (ip):(port) or localhost:(port) (from the nano) and open any .ipynb notebooks you choose. I followed those instructions to create my own JetBot image so I’m pretty sure they’re correct, although I haven’t tested since 4.2.1.

I might’ve just needed a reboot or something as when I logged in today I can now access remotely through my IP. I still don’t have the example notebooks references un the tutorial though. Is there a repository somewhere I could download them?

Huh. That sounds like a problem with the paths. I think the notebooks are in the jetbot repo you clone (git clone GitHub - NVIDIA-AI-IOT/jetbot: An educational AI robot based on NVIDIA Jetson Nano. ), but don’t quote me on that. Somebody from Nvidia should probably answer. Once you find them you can modify how jupyter starts so it points there, rather than whatever empty folder it’s starting at now.

Thanks for the replies mdegans. Yes I see the notebooks from jetbot “basic_motion”, “collision_avoidance”, etc… but I don’t see the notebook referenced in the “Thumbs Project” of the Getting started with AI on Jetson Nano course.

Hi reedshorton,

The SD card image used for the DLI course is linked to in the DLI instructions. In the course instructions it is under Getting Started with AI on Jetson Nano > Setting up your Jetson Nano > Write image to the SD card.

It is labeled “NVIDIA DLI Jetson Nano SD Card Image”

Please let me know if this works for you.

Best,
John

Hi John,

Is there anyway to get these notebooks without making a new image? I’d like to keep my old image if possible, and obviously could clone the DLI image onto a new SD card but ideally could just download the relevant notebooks from somewhere. Is that possible?

Thanks
Reed

3 Likes