OpenClaw Gateway 卡死假死问题诊断与预防方案

✅ 步骤 2：检查端口占用

# 检查 18789 端口
Get-NetTCPConnection -LocalPort 18789 -ErrorAction SilentlyContinue | Select-Object LocalAddress, LocalPort, OwningProcess, State

✅ 步骤 4：清理临时文件

# 清理锁文件
Remove-Item -Path "$env:TEMP\openclaw\*.lock" -Force -ErrorAction SilentlyContinue
# 清理旧日志
Get-ChildItem "$env:TEMP\openclaw" | Where-Object {$_.LastWriteTime -lt (Get-Date).AddDays(-7)} | Remove-Item -Force

✅ 预防性维护方案

方案一：PowerShell 维护脚本

创建 preventive-maintenance.ps1：

# OpenClaw 预防性维护脚本
# 安全执行，不影响当前运行的服务
Write-Host "🔧 OpenClaw 预防性维护" -ForegroundColor Cyan
Write-Host "========================" -ForegroundColor Cyan

# 1. 清理 7 天前的日志
$LogDir = "$env:TEMP\openclaw"
if (Test-Path $LogDir) {
    $OldLogs = Get-ChildItem $LogDir | Where-Object {$_.LastWriteTime -lt (Get-Date).AddDays(-7)}
    $OldLogs | Remove-Item -Force
    Write-Host "✅ 已清理 $($OldLogs.Count) 个旧日志文件"
}

# 2. 清理残留锁文件
$LockFiles = @("$env:TEMP\openclaw\*.lock", "$env:TEMP\openclaw\gateway.pid")
foreach ($lock in $LockFiles) {
    if (Test-Path $lock) {
        Remove-Item $lock -Force -ErrorAction SilentlyContinue
        Write-Host "✅ 已清理锁文件：$lock"
    }
}

# 3. 检查端口占用
$Port = Get-NetTCPConnection -LocalPort 18789 -ErrorAction SilentlyContinue
if ($Port) {
    Write-Host "✅ 端口 18789 正常占用 (PID: $($Port.OwningProcess))"
} else {
    Write-Host "⚠️ 端口 18789 未被占用，Gateway 可能未运行"
}

# 4. 检查 Gateway 状态
try {
    $Status = openclaw gateway status 2>&1
    if ($Status -match "Listening") {
        Write-Host "✅ Gateway 运行正常"
    } else {
        Write-Host "⚠️ Gateway 状态异常，建议运行 openclaw doctor --repair"
    }
} catch {
    Write-Host "❌ Gateway 状态检查失败：$_"
}

# 5. 配置文件备份
$ConfigPath = "$env:USERPROFILE\.openclaw\openclaw.json"
if (Test-Path $ConfigPath) {
    $BackupPath = "$ConfigPath.backup.$(Get-Date -Format 'yyyyMMdd-HHmmss')"
    Copy-Item $ConfigPath $BackupPath
    Write-Host "✅ 配置已备份：$BackupPath"
}

Write-Host ""
Write-Host "🎉 预防性维护完成！" -ForegroundColor Green

方案二：健康检查配置

创建 gateway-healthcheck.json：

{
  "healthcheck": {
    "enabled": true,
    "intervalMinutes": 15,
    "timeoutSeconds": 30,
    "maxRetries": 3,
    "autoRestart": true,
    "alerts": {
      "onFailure": true,
      "onRestart": true
    }
  },
  "resourceLimits": {
    "maxSessionMemoryMb": 512,
    "maxConcurrentSessions": 10,
    "sessionTimeoutMinutes": 120,
    "idleTimeoutMinutes": 30
  },
  "gateway": {
    "connectionTimeout": 30,
    "requestTimeout": 120,
    "keepAliveInterval": 60,
    "maxRequestBodySize": "50mb"
  },
  "logging": {
    "level": "warn",
    "rotateDaily": true,
    "maxBackups": 7
  }
}

端口占用处理

# 查找占用进程
$Port = Get-NetTCPConnection -LocalPort 18789 -ErrorAction SilentlyContinue
if ($Port) {
    Write-Host "占用进程 PID: $($Port.OwningProcess)"
    # 释放端口
    Stop-Process -Id $Port.OwningProcess -Force
    Write-Host "✅ 端口已释放"
    # 等待端口释放
    Start-Sleep -Seconds 5
}

📈 长期稳定方案

短期 (立即执行)

1. ✅ 添加看门狗进程（任务计划）
2. ✅ 配置日志轮转
3. ✅ 设置资源限制

中期 (1 周内)

1. 升级 OpenClaw 到最新版本
2. 配置远程监控 webhook
3. 建立备份策略

长期 (架构优化)

1. 多实例部署（主备模式）
2. 容器化部署（Docker）
3. 服务网格（熔断降级）