Kubernetes를 사용하여 Nvidia Jetson Nano에 객체 감지 앱 배포

준비

YAML 파일

apiVersion: apps/v1
kind: Deployment
metadata:
  creationTimestamp: null
  labels:
    app: jetson-inference
  name: jetson-inference
  namespace: jetson-inference
spec:
  replicas: 1
  selector:
    matchLabels:
      app: jetson-inference
  strategy: {}
  template:
    metadata:
      creationTimestamp: null
      labels:
        app: jetson-inference
    spec:
      hostNetwork: true
      containers:
      - image: my-harbor.io/public/jetson-inference:r32.4.3
        name: jetson-inference
        command: ["/jetson-inference/build/aarch64/bin/detectnet-camera.py"]
        args: ["--network=ssd-mobilenet-v2", "/dev/video0", "display://0"]
        env:
        - name: DISPLAY
          value: ":0.0"
        securityContext:
          privileged: true
        resources: {}
      imagePullSecrets:
        - name: harbor-secret
      nodeSelector:
        kubernetes.io/hostname: jetson-nano-01  # deploy the app to the specific device
status: {}

apiVersion: v1
kind: Namespace
metadata:
  creationTimestamp: null
  name: jetson-inference
spec: {}
status: {}

apiVersion: v1
data:
  .dockerconfigjson: <YOUR-DOCKER-AUTH-STRING>
kind: Secret
metadata:
  creationTimestamp: null
  name: harbor-secret
  namespace: jetson-inference
type: kubernetes.io/dockerconfigjson

참조