Run isaacsim in remote headless container and use livestream by rtc , but i seems not work ,

[fzg7919]

i follow this distruction to do, when i execute on step 4 , run the command to start the isaacsim container,
i takes a long time to triggered, and seems not successfully, here is the log , and the browser shows nothing. by the way, I have another queston that do I need the command to start isaacsim evety time ? I will take long time to downloads the python package? Is there another way to start the isaacsim container when i has been downloaded this docker after first time ?
https://docs.omniverse.nvidia.com/app_isaacsim/app_isaacsim/setup.html#isaac-sim-headless-container

[fzg7919]

rtc_log (64.0 KB)

[fzg7919]

[fzg7919]

nothing shows on browser

[fzg7919]

the console has no more logs to outout , just static

[fzg7919]

and when i run by runheadless.websocket.sh , core dump occurswebsocket.txt (65.2 KB)

[Sheikh_Dawood]

Hi. Looks like your first run is using the WebRTC mode. There are some known issues with this mode.

Try running the websocket mode again. Make sure there are no other containers currently running on the machine before running this:

sudo docker run --entrypoint ./runheadless.websocket.sh --gpus all -e “ACCEPT_EULA=Y” --rm --network=host nvcr.io/nvidia/isaac-sim:2021.1.1

When you see the message below in the Isaac Sim logs, you are ready to be connected.

Isaac Sim Headless Websocket App is loaded.

Connect using your browser at http://:8211/streaming/client

[fzg7919]

yes, It is okay now , thanks

[fzg7919]

I have another question, can we run several containers using different port at the same time ? so we can from different uri on browser to run several issasim applications?

[Sheikh_Dawood]

Yes, technically it is possible to remap ports for each container and select a specific GPU for each container to use. The WebSocket livestreaming is using ports 8899 and 8211.

[fzg7919]

thank you very much, I will try!!

[wangzr1116]

I use webrtc in remote headless but I got this, what should I do about this?

[Sheikh_Dawood]

Hi. Please provide more details like the output of nvidi-smi and describe your setup.

[wangzr1116]

Thanks for your replay.

The reason has been found, that some ports (such as 49100) are hard-coded in the container. However, the ports in the container are mapped randomly to the outside user in the cloud resource I used, resulting in the inability to access the webrtc url bounded to these hard-coded ports.

Is there any way to modify or configure these hard-coded ports?

[Sheikh_Dawood]

Please see WebRTC Browser Client — Omniverse Extensions documentation for ports used for WebRTC.
These port are currently hardcoded.