Developing an Omniverse Kit-Based Application Tutorial

Omniverse Learning and Resources Tutorials & Guides
1

Hi, I’m new to Omniverse and I’m struggling with the Developing an Omniverse Kit-Based Application tutorial — specifically with launching my application. I only installed the packages/dependencies specified in the tutorial.

For context, I am using an Ubuntu VM (through NVIDIA Brev) with a L4 GPU (Driver Version: 550.163.01).

I was able to successfully build the default app with ./repo.sh build.

I have attached the logs with errors when I run ./repo.sh launch. Not sure if the issue is related to the headless nature of the VM: I also ran into the same issue with xvfb-run ./repo.sh launch. Any guidance here would be appreciated. Thanks!

omniverse_logs.txt (20.8 KB)

2

The issue here is that an L4 is not an RTX GPU. Therefore it cannot work. The L4 is an AI chip and basic graphics, but not an RTX. Now an L40 or L40s yes. Sorry about that.

3

Thank you for your reply, Richard.

I switched over to a L40 cloud instance but I am still unable to launch the application using the instructions from the tutorial (see logs): The CLI hangs.

Would you be able to provide guidance on how to proceed from here?

Would I need to edit the .kit file in order to run the application in a headless manner?

Otherwise, would it be possible to display the rendered output on my local machine through some VNC Viewer?

Thanks again.
omniverse_logs_2.txt (17.6 KB)

4

Well at least this time, on the second log, you now have a GPU showing up in the table. That is progress. But this time I may be several issues.

  1. You can try not using Vulkan, and use DirectX.

C:\isaacsim\apps\isaacsim.exp.full.kit
and set vulkan = false.

name = “Isaac-Sim Full”
version = “4.5.0”
file.ignoreUnsavedOnExit = false
window.title = “Isaac Sim Full”
vulkan = false # Explicitly enable Vulkan (on by default on Linux, off by default on Windows)

  1. Just try for now to run the template normally. Do not focus on headless.

I will get the log analyzed more. Have you considered using our AMIs on Amazon and Azure?

5

Hi Richard, thanks for the tips.

I tried out setting Vulkan to false in my source/apps/my_company.my_editor.kit, but it looks like I’m still running into issues with the IAppWindow :
omniverse_logs_3.txt (17.6 KB)

[package]
title = "My Editor"
version = "0.1.0"
description = """A foundation for a base editor application."""
keywords = ["app"]  # That makes it browsable in UI with "app" filter
template_name = "kit_base_editor"
vulkan = false

I have not yet worked with the AWS AMI, but I can give that a try next. :)

6

I will see if I can find out more about your logs, but I would explore our official AMIs on Amazon

NVIDIA docs on AWS vWS: Using Omniverse AMIs on the AWS Marketplace — Omniverse Developer Workstations
AWS Linux AMI for development: AWS Marketplace: NVIDIA Omniverse™ Development Workstation (Linux)
AWS Linux AMI for enterprise: AWS Marketplace: NVIDIA Omniverse™ Enterprise Workstation (Linux)
AWS Windows AMI for development: AWS Marketplace: NVIDIA Omniverse™ Development Workstation (Windows)
AWS Windows AMI for enterprise: AWS Marketplace: NVIDIA Omniverse™ Enterprise Workstation (Windows)

1 Like
7

This is from the engineers. It makes sense.
You have to either:

  1. have a physical display attached to the computer
  2. install a virtual GPU driver and display
  3. run it headless and stream it to something. For this you can also download our web streaming sample and stream it to a browser on the same machine. We do not easily support streaming it down directly to your local workstation because of NAT traversal. You can use a normal remote viewing software for that. GitHub - NVIDIA-Omniverse/web-viewer-sample: This sample demonstrates how a front-end client can present a streamed Omniverse Kit application and how to send messages between the two apps.
Investigating the logs using Claude, it appears the issues are caused because our Brev machines do not have display installed (cloud server with no desktop).Claude suggested two options:

1. Install virtual display
2. Run kit headless

Commands for Option 1:

# Install Xvfb if not already installed sudo apt-get install xvfb # Run your application with virtual display xvfb-run -a -s "-screen 0 1920x1080x24" ./repo.sh launch

Commands for Option 2:

# Add headless flags to your launch command ./repo.sh launch -- --no-window --headless --enable-kit-console
1 Like
8

I ended up going with

I launched the Omniverse app with:

Xvfb :1 -screen 0 1920x1080x24 &
export DISPLAY=:1
./repo.sh launch

And I opened up a separate terminal for the VNC server on the VM:

x11vnc -display :1

Then I was able to see the rendered graphics through a RealVNC client on my local laptop. Thanks for all your help!

9

Great. Excellent choice. Yes that makes sense.

10

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.