docker-compose.yml

version: '3.8'

services:
  llm-proxy:
    image: llm-proxy:latest
    container_name: llm-proxy
    restart: unless-stopped
    network_mode: host
    environment:
      # 必需的环境变量
      - API_KEY=sk-you-model-key
      - BASE_URL=you-model-endpoint
            
      # 模型配置
      - BIG_MODEL=qwen3-coder
      - SMALL_MODEL=qwen3-coder
      
      # 服务器配置
      - HOST=0.0.0.0
      - PORT=4000
      - LOG_LEVEL=INFO
      - MAX_TOKENS_LIMIT=16384
      
      # 连接配置
      - REQUEST_TIMEOUT=90
      - MAX_RETRIES=1
      
      # 流式配置
      - MAX_STREAMING_RETRIES=12
      - FORCE_DISABLE_STREAMING=true
      - EMERGENCY_DISABLE_STREAMING=false
      
      # Tokenizer 配置
      - TOKENIZER_FILE=tokenizers/qwen3coder30b_tokenizer.json
      
      # Worker 配置
      - WORKERS=4
    
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:4000/health"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 10s