training-job.yaml

apiVersion: batch/v1
kind: Job
metadata:
  name: ml-training-v1
  namespace: ml-platform
spec:
  backoffLimit: 3
  template:
    spec:
      serviceAccountName: ml-trainer
      containers:
      - name: trainer
        image: 107.148.176.193:30180/ml-platform/training:v1
        env:
        - name: MODEL_VERSION
          value: "v1"
        - name: MODEL_OUTPUT_PATH
          value: "/models"
        - name: MLMODEL_NAME
          value: "housing-model"
        - name: MLMODEL_NAMESPACE
          value: "ml-platform"
        - name: CREATE_MLMODEL_CR
          value: "true"
        volumeMounts:
        - name: models
          mountPath: /models
        resources:
          requests:
            cpu: "1"
            memory: 512Mi
          limits:
            cpu: "4"
            memory: 2Gi
      volumes:
      - name: models
        persistentVolumeClaim:
          claimName: ml-models
      restartPolicy: Never