OpenClaw Gateway 服务启动、停止、监控实战指南

# OpenClaw Gateway 完整配置示例
# 全局配置
openclaw:
  # Gateway 服务配置
  gateway:
    port: 18789                    # 服务监听端口
    host: "0.0.0.0"                # 绑定地址，0.0.0.0 表示所有网卡
    auth_token: "your-secret-token" # 认证令牌（必填，建议32位以上）
    max_connections: 1000          # 最大并发连接数
    request_timeout: 30000         # 请求超时时间（毫秒）
    enable_https: false            # 是否启用 HTTPS
    ssl_cert: "/path/to/cert.pem"  # SSL 证书路径
    ssl_key: "/path/to/key.pem"    # SSL 私钥路径
  # 会话管理配置
  session:
    storage: "memory"              # 存储类型：memory/redis/sqlite
    ttl: 3600                      # 会话超时时间（秒）
    max_sessions: 10000            # 最大会话数
    cleanup_interval: 300          # 清理间隔（秒）
  # 日志配置
  logging:
    level: "info"                  # 日志级别：debug/info/warn/error
    format: "json"                 # 日志格式：json/text
    output: "/var/log/openclaw/gateway.log"  # 日志文件路径
    max_size: 100                  # 单文件最大大小（MB）
    max_backups: 10                # 最大备份文件数
    max_age: 30                    # 最大保留天数
  # 监控配置
  monitoring:
    enabled: true                  # 是否启用监控
    metrics_port: 9090             # 指标暴露端口
    health_check_path: "/health"   # 健康检查路径
    prometheus: true               # 是否启用 Prometheus 格式

# 设置 Gateway 端口
export OPENCLAW_GATEWAY_PORT=8080
# 设置认证令牌
export OPENCLAW_GATEWAY_AUTH_TOKEN="production-token-xxx"
# 设置日志级别
export OPENCLAW_LOGGING_LEVEL=debug
# 设置会话存储类型
export OPENCLAW_SESSION_STORAGE=redis

# 查看 Gateway 服务状态
openclaw gateway status
# 前台启动 Gateway（调试用）
openclaw gateway start
# 后台启动 Gateway（生产环境推荐）
openclaw gateway start --daemon
# 使用指定配置文件启动
openclaw gateway start --config /path/to/config.yaml
# 停止 Gateway 服务
openclaw gateway stop
# 重启 Gateway 服务
openclaw gateway restart
# 查看 Gateway 帮助信息
openclaw gateway --help

# Gateway 优雅关闭核心逻辑（伪代码示意）
import signal
import asyncio
from typing import Set
class GatewayServer:
    def __init__(self):
        self.active_requests: Set[asyncio.Task] = set()
        self.shutdown_event = asyncio.Event()
        self.graceful_timeout = 30  # 优雅关闭超时时间（秒）
    def setup_signal_handlers(self):
        """注册信号处理器"""
        loop = asyncio.get_event_loop()
        for sig in (signal.SIGTERM, signal.SIGINT):
            loop.add_signal_handler(
                sig,
                lambda: asyncio.create_task(self.graceful_shutdown())
            )
    async def graceful_shutdown(self):
        """优雅关闭主流程"""
        self.logger.info("开始优雅关闭...")
        # 1. 停止接收新请求
        self.server.close()
        self.logger.info("已停止接收新请求")
        # 2. 等待现有请求完成
        if self.active_requests:
            self.logger.info(f"等待 {len(self.active_requests)} 个请求完成...")
            try:
                await asyncio.wait_for(
                    asyncio.gather(*self.active_requests, return_exceptions=True),
                    timeout=self.graceful_timeout
                )
            except asyncio.TimeoutError:
                self.logger.warning("优雅关闭超时，强制终止剩余请求")
        # 3. 持久化会话状态
        await self.session_manager.persist_sessions()
        self.logger.info("会话状态已持久化")
        # 4. 释放资源
        await self.resource_pool.close_all()
        self.logger.info("资源已释放")
        # 5. 设置关闭事件
        self.shutdown_event.set()
        self.logger.info("Gateway 已安全关闭")

# 正常停止（优雅关闭）
openclaw gateway stop
# 强制停止（立即终止）
openclaw gateway stop --force
# 设置优雅关闭超时时间
openclaw gateway stop --timeout 60
# 停止并保存会话状态
openclaw gateway stop --save-session

# 健康检查端点
GET /health
# 返回示例
{
  "status": "healthy",
  "timestamp": "2024-01-15T10:30:00Z",
  "version": "1.2.0",
  "uptime": 86400,
  "components": {
    "database": "healthy",
    "redis": "healthy",
    "ai_engine": "healthy"
  }
}
# 就绪检查端点（Kubernetes Readiness Probe）
GET /ready
# 存活检查端点（Kubernetes Liveness Probe）
GET /live

# Prometheus 抓取配置示例
scrape_configs:
  - job_name: 'openclaw-gateway'
    static_configs:
      - targets: ['localhost:9090']
    metrics_path: '/metrics'
    scrape_interval: 15s

{
  "dashboard": {
    "title": "OpenClaw Gateway 监控",
    "panels": [
      {
        "title": "请求 QPS",
        "type": "graph",
        "targets": [
          {
            "expr": "rate(openclaw_requests_total[5m])",
            "legendFormat": "{{method}} - {{channel}}"
          }
        ]
      },
      {
        "title": "请求延迟 P99",
        "type": "stat",
        "targets": [
          {
            "expr": "histogram_quantile(0.99, rate(openclaw_request_duration_seconds_bucket[5m]))"
          }
        ]
      },
      {
        "title": "错误率",
        "type": "gauge",
        "targets": [
          {
            "expr": "rate(openclaw_errors_total[5m]) / rate(openclaw_requests_total[5m]) * 100"
          }
        ]
      }
    ]
  }
}

# 1. 查看详细日志
openclaw gateway start --log-level debug
# 2. 检查端口占用
lsof -i :18789
# 3. 检查配置文件语法
openclaw gateway config validate
# 4. 检查权限
ls -la ~/.openclaw/

# 1. 检查 Gateway 日志
tail -f /var/log/openclaw/gateway.log | grep timeout
# 2. 检查 AI 引擎状态
curl http://localhost:18789/health
# 3. 检查网络连通性
ping your-ai-service.com
# 4. 查看当前连接数
netstat -an | grep 18789 | wc -l

# 1. 监控内存使用
watch -n 1 'ps aux | grep openclaw'
# 2. 分析内存分布
curl http://localhost:9090/metrics | grep memory
# 3. 检查会话数量
curl http://localhost:18789/admin/sessions/count
# 4. 开启 pprof 分析
openclaw gateway start --enable-pprof

# 查看最近错误日志
cat gateway.log | jq 'select(.level=="error")'
# 统计各渠道请求量
cat gateway.log | jq -r '.channel' | sort | uniq -c
# 分析慢请求（超过 5 秒）
cat gateway.log | jq 'select(.duration > 5000)'
# 按时间范围过滤
cat gateway.log | jq 'select(.timestamp >= "2024-01-15T10:00:00" and .timestamp < "2024-01-15T11:00:00")'

# Nginx 负载均衡配置
upstream openclaw_gateway {
    # 使用最少连接算法
    least_conn;
    # Gateway 实例列表
    server 192.168.1.101:18789 weight=1 max_fails=3 fail_timeout=30s;
    server 192.168.1.102:18789 weight=1 max_fails=3 fail_timeout=30s;
    server 192.168.1.103:18789 weight=1 max_fails=3 fail_timeout=30s;
    # 健康检查（需要 nginx-plus 或 tengine）
    check interval=3000 rise=2 fall=3 timeout=1000 type=http;
    check_http_send "GET /health HTTP/1.0\r\n\r\n";
    check_http_expect_alive http_2xx http_3xx;
}
server {
    listen 80;
    server_name gateway.openclaw.ai;
    # 请求体大小限制
    client_max_body_size 10m;
    # 超时配置
    proxy_connect_timeout 60s;
    proxy_send_timeout 60s;
    proxy_read_timeout 60s;
    location / {
        proxy_pass http://openclaw_gateway;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header X-Forwarded-Proto $scheme;
    }
}

# 会话 Redis 存储配置
session:
  storage: "redis"
  redis:
    host: "redis-cluster.openclaw.internal"
    port: 6379
    password: "${REDIS_PASSWORD}"
    db: 0
    pool_size: 50
    key_prefix: "openclaw:session:"
    ttl: 3600

# 安全配置示例
security:
  # 认证配置
  auth:
    enabled: true
    token_rotation_days: 30  # 令牌轮换周期
  # 限流配置
  rate_limit:
    enabled: true
    requests_per_minute: 100
    burst: 20
  # IP 白名单
  ip_whitelist:
    enabled: true
    allowed:
      - "10.0.0.0/8"
      - "172.16.0.0/12"
  # HTTPS 配置
  tls:
    enabled: true
    cert_path: "/etc/ssl/certs/gateway.pem"
    key_path: "/etc/ssl/private/gateway.key"
    min_version: "TLS1.2"

# 性能优化配置
performance:
  # 连接池配置
  connection_pool:
    max_idle: 100
    max_open: 200
    idle_timeout: 300
  # 缓存配置
  cache:
    enabled: true
    type: "redis"
    ttl: 300
  # 并发配置
  concurrency:
    max_workers: 100
    queue_size: 1000