gpu-workload/t5/kubernetes/serving-cpu.yaml

---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: t5-inference
  labels:
    model: t5
    version: v1.0
    machine: cpu
spec:
  replicas: 1
  selector:
    matchLabels:
      model: t5
      version: v1.0
      machine: cpu
  template:
    metadata:
      labels:
        model: t5
        version: v1.0
        machine: cpu
    spec:
      securityContext:
        fsGroup: 1000
        runAsUser: 1000
        runAsGroup: 1000
      containers:
        - name: inference
          image: gcr.io/lustrous-baton-363720/models/t5-small:1.0-cpu
          # image: IMAGE_NAME
          imagePullPolicy: IfNotPresent
          args: ["torchserve", "--start", "--foreground"]
          resources:
            limits:
              cpu: "3000m"
              memory: 16Gi
              ephemeral-storage: 10Gi
            requests:
              cpu: "3000m"
              memory: 16Gi
              ephemeral-storage: 10Gi
          ports:
            - containerPort: 8080
              name: http
            - containerPort: 8081
              name: management
            - containerPort: 8082
              name: metrics
          readinessProbe:
            httpGet:
              path: /ping
              port: http
            initialDelaySeconds: 120
            failureThreshold: 10
          livenessProbe:
            httpGet:
              path: /models/t5-small
              port: management
            initialDelaySeconds: 150
            periodSeconds: 5
---
apiVersion: v1
kind: Service
metadata:
  name: t5-inference
  labels:
    model: t5
    version: v1.0
    machine: cpu
spec:
  type: ClusterIP
  selector:
    model: t5
    version: v1.0
    machine: cpu
  ports:
    - port: 8080
      name: http
      targetPort: http
    - port: 8081
      name: management
      targetPort: management
    - port: 8082
      name: metrics
      targetPort: metrics