Jetson Xavier NX L4T 32.4.3 Cloud-Native Demo Issues

[aplu]

Hi everyone,

First time poster here, I am currently trying to run Nvidia’s cloud native docker demo on the Jetson Xavier NX (as shown here).

GitHub - NVIDIA-AI-IOT/jetson-cloudnative-demo: Multi-container demo for...

Upon doing so, I have found that only the person object detection container works, and I think this is because the L4T version that the containers are only built for L4T 32.4.2 when i currrently only have 32.4.3. I have yet to successfully run the last 3 containers and would like to know if anyone else has any work-arounds. Thanks!

[AastaLLL]

Hi,

The container can work once recreating the TensorRT engine for the new version.
You can find an example in this comment:

Xavier NX multiple container demo stuck launching TRTIS

Hi, Sorry for the late update. This comes from the incompatible TensorRT serialized engine file. Please noticed that these docker image is built on the JetPack 4.4 DP (TRT-7.1.0). To run it on the JetPack 4.4 GA (TRT-7.1.3), please recompile the TensorRT engine first. $ sudo docker run -it --rm --net=host --runtime nvidia -e DISPLAY=$DISPLAY -v /tmp/.X11-unix/:/tmp/.X11-unix nvcr.io/nvidia/jets $ python3 >>> import torch >>> import torch2trt >>> import trt_pose.models >>> import tensorrt …

Thanks.

[Andrey1984]

the quoted command above is abrupt
it should be read as

sudo docker run -it --rm --net=host --runtime nvidia -e DISPLAY=$DISPLAY -v /tmp/.X11-unix/:/tmp/.X11-unix nvcr.io/nvidia/jetson-pose:r32.4.2

@AastaLLL
but how do we patch the other two examples[containers] of the cloud native so that they run on NX with the latest Jetpack?

[AastaLLL]

For anyone interested in the comment on Jan 13, please see Cloud_native_demo on JP4.4.1 for the following.

Thanks.