Omniverse Leonardo Preview crashes

I’ve been trying to run NVIDIA Omniverse Leonardo Preview simulator for robotic manipulation.

After configuring AWS EC2 instance of type g4dn.4xlarge with VPC in North California and running Omniverse Client with provided bash script as follows:

./ -s <public_aws_ec2_ip>

Application (client) runs fine and loads fine, but after clicking play in all of the provided scenarios Omniverse Client freezes and crashes with following error messages:

Bash where nvidia docker container for leonardo preview is running

/usr/local/bin/ line 29:     6 Segmentation fault      (core dumped) /omniverse-kit/_build/linux-x86_64/release/omniverse-kit --allow-root --no-window --carb/app/livestream/enabled=true --carb/app/window/width=1440 --carb/app/window/height=900 --carb/app/renderer/resolution/width=1280 --carb/app/renderer/resolution/height=720

Stack trace from local bash where omniverse client is running:

Wed Mar  4 15:32:24:272   ERROR [NVST:RtspCurlBase] {FECCD700} -  curl_easy_perform() failed: 85
Wed Mar  4 15:32:24:272   ERROR [NVST:RtspCurlEvent] {FECCD700} -  RTSP server returns with errror code: 85, continuing
Wed Mar  4 15:32:24:442   ERROR [NVST:RtspCurlBase] {FECCD700} -  curl_easy_perform() failed: 7
Wed Mar  4 15:32:24:442   ERROR [NVST:RtspCurlEvent] {FECCD700} -  RTSP polling failed: 7
Wed Mar  4 15:32:32:173   ERROR [NVST:ClientSession] {FCCC9700} -  No decodable unit received in over 8215.960410ms. Disconnecting...
Wed Mar  4 15:32:32:173   ERROR [NVST:ClientSession] {FCCC9700} -  session time 171559.681817ms, last decodable unit:4961 arrived at 163343.721407ms, timeout at 8000.000000ms
Wed Mar  4 15:32:32:173   ERROR [NVST:ClientSession] {FCCC9700} -  last RTP PING:0.000000ms, last payload:163.343674ms, failedPingBackCount:0
Wed Mar  4 15:32:32:177   ERROR [NVST:UdpRtpSource] {FD4CA700} -  UDP RTP Source: failed to receive data (Error: 0x80000013)
Wed Mar  4 15:32:32:177   ERROR [NVST:RtpSourceQueue] {FD4CA700} -  RtpSourceQueue: failed to read RTP packet (Result: 0X80000013)
Wed Mar  4 15:32:32:177   ERROR [BifrostClient: NvscWrapper] {FCCC9700} - Received old frame - current 4961 received 0
Wed Mar  4 15:32:32:178   ERROR [RemoteGridApp] {21A03700} - STREAMING: FAILED: NVB_SN_STREAMER_NETWORK_ERROR (80040017)
Wed Mar  4 15:32:32:178   ERROR [NVST:ClientSession] {FCCC9700} -  Failed to handle VideoTimeout for streamer instance 0

Resolution is set up on recommended 1080p(1920x1080), but my ISP is based in EU and that could possibly cause a problem.

Help or feedback would be highly appreciated. Thank you very much.


If you force close the client and reconnect again, do you see the scenario running?
If you are unable to connect after that, it sounds like IsaacSim has crash but not the client.

Please provide the logs at /var/lib/isaac-sim/data/omniverse-kit.log on your EC2 instance.


Unfortunately I’ve terminated EC2 instance yesterday, therefore I don’t have access to those logs anymore and I’ve completely foreseen that file. I’m pretty sure that IsaacSim crashed yesterday, but now I can’t know why. Sorry.

Today I’ve repeated all the steps, and now I’m able to run all existing scenarios. The only difference is in starting resolution, yesterday I changed resolution after first crash, today I’ve set it on the start on 1080p.

Thank you for your fast reply, I’ll continue to test this preview because it seems quite powerful tool. If anything occurs, I’ll provide log with brief explanation.