apiVersion: apps/v1
kind: Deployment
metadata:
  name: vllm
  namespace: vllm-service
  labels:
    app: vllm
spec:
  replicas: 1
  strategy:
    type: Recreate
  selector:
    matchLabels:
      app: vllm
  template:
    metadata:
      labels:
        app: vllm
    spec:
      runtimeClassName: nvidia
      enableServiceLinks: false
      nodeSelector:
        kubernetes.io/hostname: gremlin-1
      containers:
        - name: vllm
          image: registry.celestium.life/stonks-oracle/vllm-openai:latest
          imagePullPolicy: Always
          args:
            - "--model"
            - "numind/NuExtract3"
            - "--served-model-name"
            - "numind/NuExtract3"
            - "--host"
            - "0.0.0.0"
            - "--port"
            - "8000"
            - "--gpu-memory-utilization"
            - "0.80"
            - "--max-model-len"
            - "8192"
            - "--max-num-seqs"
            - "8"
          env:
            - name: HF_TOKEN
              valueFrom:
                secretKeyRef:
                  name: vllm-secrets
                  key: HF_TOKEN
            - name: VLLM_ATTENTION_BACKEND
              value: "FLASHINFER"
          ports:
            - containerPort: 8000
              name: http
          resources:
            limits:
              nvidia.com/gpu: "1"
            requests:
              cpu: "2"
              memory: "8Gi"
          readinessProbe:
            httpGet:
              path: /health
              port: 8000
            initialDelaySeconds: 120
            periodSeconds: 10
          livenessProbe:
            httpGet:
              path: /health
              port: 8000
            initialDelaySeconds: 300
            periodSeconds: 30
          volumeMounts:
            - name: hf-cache
              mountPath: /root/.cache/huggingface
      volumes:
        - name: hf-cache
          persistentVolumeClaim:
            claimName: vllm-hf-cache-pvc