servingEngineSpec:
  modelSpec:
    - name: "qwen"
      repository: "vllm/vllm-openai"
      tag: "latest"
      modelURL: "Qwen/Qwen2.5-Coder-14B-Instruct"
      replicaCount: 1

      requestCPU: 16
      requestMemory: "32Gi"
      requestGPU: 1

      pvcStorage: "10Gi"
      pvcAccessMode:
        - ReadWriteMany
      storageClass: "nfs"

      vllmConfig:
        enableChunkedPrefill: false
        enablePrefixCaching: false
        maxModelLen: 4096
        dtype: "float16"
        extraArgs:
          [
            "--disable-log-requests",
            "--gpu-memory-utilization",
            "0.95",
            "--download-dir",
            "/data/qwen",
          ]
      env:
        - name: HF_HUB_OFFLINE
          value: "1"
routerSpec:
  repository: "lmcache/lmstack-router"
  tag: "latest"
  enableRouter: true
  resources:
    requests:
      cpu: "2"
      memory: "8G"
    limits:
      cpu: "4"
      memory: "16G"
  ingress:
    enabled: true
    className: ""
    annotations:
      cert-manager.io/cluster-issuer: "selfsigned-issuer"
      cert-manager.io/duration: 8760h
      cert-manager.io/renew-before: 720h
    hosts:
      - host: vllm.example.com
        paths:
          - path: /
            pathType: Prefix
    tls:
      - secretName: vllm-tls-secret
        hosts:
          - vllm.example.com