Best practices for setting up DeepStream file permissions and structure

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

I have been struggling getting all of the various test apps to run before breaking out and building my own app. My first question is what is the best practice for file permissions within the samples directories. If you follow the instructions DeepStream ends up getting installed into /opt/nvidia/deepstream/* and when you go to edit any config files I have to use sudo. Very annoying with vscode as it prompts to switch to sudo to save and if you don’t see this you go on to test without the file being saved. Nowhere in the instructions does it say to chown any of the tree but I did find an Nvidia response to another issue where it suggested sudo chown -R user:user /opt/nvidia/. Is this the right thing to do or can/should it be done on a smaller part of the tree?

I’m wondering if this is also leading to a problem with running deepstream-test3 where it takes several minutes to start. I’m wondering if this is because it is on-demand compiling a model and then fails to write it out for the next run.

This comes to the other part of my question. I want to start my own app but NOT have it nested down in /opt/nvidia/deepstream. If I create a git repo up in my users home directory then what do I need to do for access down into the /opt/nvidia/deepstream tree? I was kind of surprised the python samples git repo needed to be cloned down into the nvidia tree vs anywhere. What breaks and what needs to be patched if that was cloned elsewhere. It’s a very rigid structure with a lot of tentacles.

Thanks

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU) Jetson OWIN
• DeepStream Version 6.3
• JetPack Version (valid for Jetson only) 5.1.2-b104

3

Because there are many relative path issues with makefile and configuration files, some projects need to be placed in specific directories. We also have some demos that do not rely on the paths. If you develop your own demo, you can refer to the link: https://github.com/NVIDIA-AI-IOT/deepstream_tao_apps.

4

Thanks for the response. Any input on the first part of that question… what is the best practice for changing ownership/permissions on the directory structure to enable easier editing of samples and on demand caching of resources?

5

This way is OK. Or you need to add the sudo before each command.