apiVersion: apps/v1
kind: Deployment
metadata:
  name: vllm-sim-demo
spec:
  replicas: 1
  selector:
    matchLabels:
      app: vllm-sim-demo
  template:
    metadata:
      labels:
        app: vllm-sim-demo
    spec:
      initContainers:
      # vLLM render sidecar - handles tokenization via HTTP (vllm launch render)
      # restartPolicy: Always makes this a native sidecar (K8s 1.29+):
      # the simulator will not start until this container's readinessProbe passes
      - name: vllm-render
        image: vllm/vllm-openai-cpu:v0.21.0
        imagePullPolicy: IfNotPresent
        restartPolicy: Always
        command: ["vllm"]
        args: ["launch", "render", "Qwen/Qwen3.5-397B-A17B", "--port=8082"]
        env:
        - name: HF_TOKEN
          value: "<your HuggingFace token, if this is for a real model>"
        ports:
        - name: http
          containerPort: 8082
        livenessProbe:
          httpGet:
            path: /health
            port: 8082
          initialDelaySeconds: 10
          periodSeconds: 10
        readinessProbe:
          httpGet:
            path: /health
            port: 8082
          initialDelaySeconds: 5
          periodSeconds: 5
      containers:
      - name: vllm-sim
        image: ghcr.io/llm-d/llm-d-inference-sim:latest
        imagePullPolicy: IfNotPresent
        args:
        - "--model"
        - "Qwen/Qwen3.5-397B-A17B"
        - "--port"
        - "8000"
        - "--render-url"
        - "http://localhost:8082"
        env:
          - name: POD_NAME
            valueFrom:
              fieldRef:
                apiVersion: v1
                fieldPath: metadata.name
          - name: POD_NAMESPACE
            valueFrom:
              fieldRef:
                apiVersion: v1
                fieldPath: metadata.namespace
          - name: POD_IP
            valueFrom:
              fieldRef:
                fieldPath: status.podIP
        ports:
        - containerPort: 8000
          name: http
          protocol: TCP
---
apiVersion: v1
kind: Service
metadata:
  name: vllm-sim-demo-svc
  labels:
    app: vllm-sim-demo
spec:
  selector:
    app: vllm-sim-demo
  ports:
    - protocol: TCP
      port: 8000
      targetPort: 8000
      name: http
  type: ClusterIP