配置文件

通过配置文件管理 GPT-Load 的详细设置，支持 JSON 和 YAML 格式。

配置文件位置

GPT-Load 按以下优先级查找配置文件：

  1. ./config.json
  2. ./config.yaml
  3. ~/.gpt-load/config.json
  4. /etc/gpt-load/config.json

JSON 配置示例

{
  "server": {
    "host": "0.0.0.0",
    "port": 3000,
    "cors": {
      "origin": "*",
      "credentials": true
    }
  },
  "providers": {
    "openai": {
      "enabled": true,
      "apiKeys": [
        "sk-key1",
        "sk-key2",
        "sk-key3"
      ],
      "baseURL": "https://api.openai.com",
      "timeout": 30000
    },
    "azure": {
      "enabled": false,
      "endpoint": "https://your-resource.openai.azure.com",
      "apiKey": "your-azure-key",
      "apiVersion": "2023-12-01-preview"
    }
  },
  "loadBalancing": {
    "strategy": "round-robin",
    "healthCheck": {
      "enabled": true,
      "interval": 60000,
      "timeout": 5000
    },
    "retryConfig": {
      "maxRetries": 3,
      "retryDelay": 1000,
      "backoffMultiplier": 2
    }
  },
  "rateLimit": {
    "enabled": true,
    "windowMs": 60000,
    "maxRequests": 100,
    "keyGenerator": "ip"
  },
  "monitoring": {
    "enabled": true,
    "endpoints": {
      "health": "/health",
      "stats": "/stats",
      "metrics": "/metrics"
    }
  },
  "logging": {
    "level": "info",
    "format": "json",
    "file": {
      "enabled": true,
      "path": "./logs/gpt-load.log",
      "maxSize": "10MB",
      "maxFiles": 5
    }
  }
}

YAML 配置示例

server:
  host: "0.0.0.0"
  port: 3000
  cors:
    origin: "*"
    credentials: true

providers:
  openai:
    enabled: true
    apiKeys:
      - "sk-key1"
      - "sk-key2"
      - "sk-key3"
    baseURL: "https://api.openai.com"
    timeout: 30000
  
  azure:
    enabled: false
    endpoint: "https://your-resource.openai.azure.com"
    apiKey: "your-azure-key"
    apiVersion: "2023-12-01-preview"

loadBalancing:
  strategy: "round-robin"
  healthCheck:
    enabled: true
    interval: 60000
    timeout: 5000
  retryConfig:
    maxRetries: 3
    retryDelay: 1000
    backoffMultiplier: 2

rateLimit:
  enabled: true
  windowMs: 60000
  maxRequests: 100
  keyGenerator: "ip"

monitoring:
  enabled: true
  endpoints:
    health: "/health"
    stats: "/stats"
    metrics: "/metrics"

logging:
  level: "info"
  format: "json"
  file:
    enabled: true
    path: "./logs/gpt-load.log"
    maxSize: "10MB"
    maxFiles: 5

主要配置项说明

Server 配置

  • • host: 服务器监听地址
  • • port: 服务器端口
  • • cors: CORS 跨域配置

Providers 配置

  • • openai.apiKeys: OpenAI API 密钥数组
  • • azure.endpoint: Azure OpenAI 端点
  • • timeout: 请求超时时间（毫秒）

Load Balancing 配置

  • • strategy: 负载均衡策略（round-robin, random, least-connections）
  • • healthCheck: 健康检查配置
  • • retryConfig: 重试配置

动态重载配置

GPT-Load 支持动态重载配置文件，无需重启服务：

# 发送 SIGHUP 信号重载配置
kill -HUP $(pgrep gpt-load)

# 或通过管理接口
curl -X POST http://localhost:3000/admin/reload