vllm_job.yaml

apiVersion: batch/v1
kind: Job
metadata:
  name: vllm-job
  labels:
    app: vllm
spec:
  completions: 1  
  parallelism: 1  
  template:
    metadata:
      labels:
        app: vllm
    spec:
      containers:
      - name: vllm-container
        image: quay.io/chenw615/vllm_dra
        command:
        - "python"
        - "-m"
        - "vllm.entrypoints.openai.api_server"
        - "--model"
        - "facebook/opt-125m"
        ports:
        - containerPort: 8000
        env:
        - name: HUGGING_FACE_HUB_TOKEN
          valueFrom:
            secretKeyRef:
              name: huggingface-secret
              key: HF_TOKEN
        - name: MODEL_NAME
          value: "facebook/opt-125m"
        - name: TRANSFORMERS_CACHE
          value: "/workspace/huggingface/"
        - name: HF_HOME
          value: "/workspace/huggingface/"
        - name: NUMBA_DISABLE_JIT
          value: "1"
        - name: NUMBA_CACHE_DIR
          value: "/workspace/huggingface/"
        resources:
          limits:
            nvidia.com/mig-3g.20gb: 1
        volumeMounts:
        - name: workspace-storage
          mountPath: /workspace/huggingface/
      restartPolicy: OnFailure
      volumes:
      - name: workspace-storage
        persistentVolumeClaim:
          claimName: workspace-storage
---
apiVersion: v1
kind: Service
metadata:
  name: vllm
spec:
  type: ClusterIP
  ports:
  - port: 8000
    targetPort: 8000
    name: http
  selector:
    app: vllm
---
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: workspace-storage
spec:
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 10Gi