Helm chart to deploy DeepStream-l4t cotainer in AGX Xavier


i have a jetson AGX xavier, flushed with SDK manager (Jetpack 4.6).
I am using microk8s to deploy containerized services:
CONTAINER-RUNTIME: docker://20.10.11
kubelet: V1.22
helm3 enabled.

Following Getting Started with Docker and AI workloads on NVIDIA Jetson AGX Xavier Developer Platform - Collabnix i was able to deploy the base image through docker, e.g.,

docker run --gpus all --rm -it --name human_presence /bin/bash

where i could deploy e.g., a YOLO human presence model, successfully.

However, when using helm charts, it fails with a CrashLoopBackOff.
In my version I use microk8s with helm3 (i.e., no tiller) and no gpu operator (correct me if i am wrong but the gpu operator is not compatible yet with Jetson devices)

When using microk8s with other images, it works fine (e.g.m nginx with internet access to the container etc.)

I follow DeepStream - L4T- Intelligent Video Analytics Demo | NVIDIA NGC in order to better understand how to make it work

any ideas?


Is the issue related with deepstream?


yes its is related to deepstream, and is sovled.

I just added
command: [“sleep”]
args: [“infinity”]

for the container and worked fine, with microk8s and helm3 charts for deployment.


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