增强 8910 日志盯盘与联调可读性：本机落盘 info/error、bat 入口，出站与 echo 打 reply 预览，部署强制 UTF-8。

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-09-13 21:47:05 +08:00
parent 4d90bba99d
commit 8bdf4990a2
8 changed files with 263 additions and 45 deletions
@@ -0,0 +1,28 @@
+@echo off
+setlocal EnableExtensions
+cd /d "%~dp0"
+
+REM Double-click: fetch recent remote agent 8910 logs once, then pause.
+
+where powershell.exe >nul 2>&1
+if errorlevel 1 (
+  echo ERROR: powershell.exe not found.
+  goto :end
+)
+
+if not exist "%~dp0Watch-Agent8910.ps1" (
+  echo ERROR: Watch-Agent8910.ps1 missing next to this bat.
+  goto :end
+)
+
+echo Starting Watch-Agent8910 -Once ...
+echo.
+powershell.exe -NoLogo -NoProfile -ExecutionPolicy Bypass -File "%~dp0Watch-Agent8910.ps1" -Once %*
+set ERR=%ERRORLEVEL%
+echo.
+echo Finished. ExitCode=%ERR%
+
+:end
+echo.
+pause
+exit /b %ERR%
@@ -0,0 +1,31 @@
+@echo off
+setlocal EnableExtensions
+cd /d "%~dp0"
+
+REM Double-click: follow remote agent 8910 logs. Ctrl+C to stop.
+REM Colleague setup: see Watch-Agent8910.colleague.txt (Chinese guide also available).
+REM Pass-through args, e.g. Watch-Agent8910.bat -Once
+
+where powershell.exe >nul 2>&1
+if errorlevel 1 (
+  echo ERROR: powershell.exe not found.
+  goto :end
+)
+
+if not exist "%~dp0Watch-Agent8910.ps1" (
+  echo ERROR: Watch-Agent8910.ps1 missing next to this bat.
+  goto :end
+)
+
+echo Starting Watch-Agent8910 ...
+echo Local logs: repo\logs\YYYYMMDD_info.log and _error.log
+echo.
+powershell.exe -NoLogo -NoProfile -ExecutionPolicy Bypass -File "%~dp0Watch-Agent8910.ps1" %*
+set ERR=%ERRORLEVEL%
+echo.
+echo Finished. ExitCode=%ERR%
+
+:end
+echo.
+pause
+exit /b %ERR%
@@ -1,58 +1,120 @@
-# 只读：拉服务器新系统 8910 智能体日志中的企微相关行与错误行。
-# 用法（仓库根或任意目录）：
-#   .\deploy\Watch-Agent8910.ps1              # 持续监听（Ctrl+C 结束）
-#   .\deploy\Watch-Agent8910.ps1 -Once        # 只看最近一段
-#   .\deploy\Watch-Agent8910.ps1 -All         # 不过滤关键字，整行刷出
+# 只读：拉服务器新系统 8910 的「业务/错误」日志，并同时落到本机 logs/。
+# 用法：
+#   powershell -NoProfile -ExecutionPolicy Bypass -File .\deploy\Watch-Agent8910.ps1
+#   powershell -NoProfile -ExecutionPolicy Bypass -File .\deploy\Watch-Agent8910.ps1 -Once
+#   双击 Watch-Agent8910.bat / Watch-Agent8910-Once.bat
 #
-# 说明：
-# - 盯的是 E:\wwwroot\logs\agent-http.*.log（新系统 8910），不碰现网 8810/8811。
-# - 切流前公网 /wecom/callback 仍进现网；本脚本主要看到 inject、出站、本机调 8910 的错误。
-# - 依赖本机 SSH 密钥 ~/.ssh/id_ed25519_ytd。
+# 本机落盘（仓库根 logs/，按当天日期）：
+#   logs/20260913_info.log   — 业务信息（入队、reply=、发送成功等）
+#   logs/20260913_error.log  — 告警/异常（WARNING、ERROR、Traceback、errcode 等）
+#
+# 乱码：默认 UTF-8；仍乱码可加 -Gbk。
+# SSH：默认密钥 ~/.ssh/id_ed25519_ytd；可用 YTD_SSH_KEY / YTD_SSH_HOST 覆盖。

 [CmdletBinding()]
 param(
  [switch]$Once,
  [switch]$All,
-  [int]$Tail = 80
+  [switch]$Access,
+  [switch]$Gbk,
+  [switch]$NoLocalLog,
+  [int]$Tail = 120
 )

 $ErrorActionPreference = 'Stop'
-$SshKey = Join-Path $env:USERPROFILE '.ssh\id_ed25519_ytd'
-$SshHost = 'ytd_extdev_0903@129.211.181.80'
+$SshKey = if ($env:YTD_SSH_KEY) { $env:YTD_SSH_KEY } else { Join-Path $env:USERPROFILE '.ssh\id_ed25519_ytd' }
+$SshHost = if ($env:YTD_SSH_HOST) { $env:YTD_SSH_HOST } else { 'ytd_extdev_0903@129.211.181.80' }
 $LogOut = 'E:\wwwroot\logs\agent-http.out.log'
 $LogErr = 'E:\wwwroot\logs\agent-http.err.log'
 $IncomingRemotePath = 'C:/Users/ytd_extdev_0903/incoming/_watch_agent8910.ps1'
 $IncomingLocalRun = 'C:\Users\ytd_extdev_0903\incoming\_watch_agent8910.ps1'

-# 企微回调 / 出站 / 注入 / 失败；故意不含纯 /health 探活行，避免刷屏
-$Pattern = '/wecom/callback|inject-text|inject:|outbox|inbox|errcode|errmsg|60020|ERROR|WARNING|Traceback|Exception|ModuleNotFoundError|sender=|touser=|crypto|ChannelRuntime|dispatch intent'
+$Pattern = '回调|inject-text|inject:|outbox|inbox|errcode|errmsg|60020|ERROR|WARNING|Traceback|Exception|ModuleNotFoundError|sender=|touser=|crypto|ChannelRuntime|dispatch intent|企微发送|验签|解密|reply='

 if (-not (Test-Path $SshKey)) {
  throw "缺少 SSH 密钥: $SshKey"
 }

-Write-Host "目标: $SshHost"
-Write-Host "日志: $LogOut"
-Write-Host "      $LogErr"
-if ($All) {
-  Write-Host '过滤: 关闭（全部输出）'
-} else {
-  Write-Host "过滤: $Pattern"
+try {
+  chcp 65001 | Out-Null
+  [Console]::InputEncoding = [System.Text.UTF8Encoding]::new($false)
+  [Console]::OutputEncoding = [System.Text.UTF8Encoding]::new($false)
+  $OutputEncoding = [System.Text.UTF8Encoding]::new($false)
+} catch {
+  Write-Host 'warn: 无法设置本机控制台 UTF-8'
 }
-Write-Host '---'
+
+# 仓库根/logs（本脚本在 deploy/ 下）
+$RepoRoot = Split-Path -Parent $PSScriptRoot
+$LocalLogDir = Join-Path $RepoRoot 'logs'
+$Day = Get-Date -Format 'yyyyMMdd'
+$LocalInfoLog = Join-Path $LocalLogDir ($Day + '_info.log')
+$LocalErrorLog = Join-Path $LocalLogDir ($Day + '_error.log')
+$Utf8NoBom = New-Object System.Text.UTF8Encoding $false
+
+if (-not $NoLocalLog) {
+  New-Item -ItemType Directory -Force -Path $LocalLogDir | Out-Null
+  $sessionHdr = '===== watch start ' + (Get-Date).ToString('yyyy-MM-dd HH:mm:ss') + ' host=' + $SshHost + ' once=' + $Once.IsPresent + ' ====='
+  [System.IO.File]::AppendAllText($LocalInfoLog, $sessionHdr + [Environment]::NewLine, $Utf8NoBom)
+  [System.IO.File]::AppendAllText($LocalErrorLog, $sessionHdr + [Environment]::NewLine, $Utf8NoBom)
+}
+
+# 写入本地 *_error.log 的判定（其余进 *_info.log）
+# 注意：不能用裸 ERROR（-match 不区分大小写，会误伤 “error.log” 说明行）
+function Write-LocalLine([string]$line) {
+  Write-Host $line
+  if ($NoLocalLog) { return }
+  # 脚本自身说明行一律进 info，避免关键字误伤
+  if ($line -match '^(目标:|主日志|访问日志|日志编码|过滤:|本机落盘:|说明:|---+|=====)') {
+    $isErr = $false
+  } else {
+    $isErr = ($line -cmatch '\] WARNING |\] ERROR |Traceback|Exception|ModuleNotFoundError') -or
+             ($line -match 'errcode=\d+|errmsg=.+|health_fail|MISSING |解密失败|验签/解密失败|relation "agent_inbox"|not allow to access from your ip')
+  }
+  $target = if ($isErr) { $LocalErrorLog } else { $LocalInfoLog }
+  try {
+    [System.IO.File]::AppendAllText($target, $line + [Environment]::NewLine, $Utf8NoBom)
+  } catch {
+    Write-Host ('warn: 写本地日志失败: ' + $_.Exception.Message)
+  }
+}
+
+Write-LocalLine ("目标: $SshHost")
+Write-LocalLine ("主日志(业务/错误): $LogErr")
+if ($Access -or $All) { Write-LocalLine ("访问日志: $LogOut") }
+Write-LocalLine ("日志编码: " + $(if ($Gbk) { 'Default/GBK' } else { 'UTF-8' }))
+if ($All) {
+  Write-LocalLine '过滤: 关闭'
+} else {
+  Write-LocalLine ("过滤: $Pattern")
+}
+if (-not $NoLocalLog) {
+  Write-LocalLine ("本机落盘: $LocalInfoLog")
+  Write-LocalLine ("本机落盘: $LocalErrorLog")
+}
+Write-LocalLine '说明: 企微气泡里的 [test] 回复若已出现=发送成功；日志里看 reply= 字段'
+Write-LocalLine '---'

 $waitLiteral = if ($Once) { '$false' } else { '$true' }
 $allLiteral = if ($All) { '$true' } else { '$false' }
+$accessLiteral = if ($Access) { '$true' } else { '$false' }
+$encName = if ($Gbk) { 'Default' } else { 'UTF8' }

-# 远程执行体：先 health，再 Tail；持续模式用 -Wait
 $remoteScript = @"
 `$ErrorActionPreference = 'Continue'
+try {
+  chcp 65001 | Out-Null
+  [Console]::OutputEncoding = [System.Text.UTF8Encoding]::new(`$false)
+  `$OutputEncoding = [System.Text.UTF8Encoding]::new(`$false)
+} catch {}
 `$out = '$LogOut'
 `$err = '$LogErr'
 `$tail = $Tail
 `$wait = $waitLiteral
 `$all = $allLiteral
+`$wantAccess = $accessLiteral
 `$pat = '$Pattern'
+`$enc = '$encName'

 Write-Output '=== remote health 8910 ==='
 try {
@@ -61,41 +123,90 @@ try {
 } catch {
  Write-Output ('health_fail: ' + `$_.Exception.Message)
 }
-Write-Output '=== log tail ==='
+Write-Output '=== business/error log (err) ==='

-`$files = @()
-if (Test-Path `$out) { `$files += `$out } else { Write-Output ('MISSING ' + `$out) }
-if (Test-Path `$err) { `$files += `$err } else { Write-Output ('MISSING ' + `$err) }
-if (`$files.Count -eq 0) { throw 'no agent-http log files' }
+function Show-Err([string]`$line) {
+  if (`$all) { Write-Output ('[err] ' + `$line); return }
+  if (`$line -match `$pat) { Write-Output ('[err] ' + `$line) }
+}

-function Show-Line([string]`$line) {
-  if (`$all) {
-    Write-Output `$line
-    return
-  }
-  if (`$line -match `$pat) {
-    Write-Output `$line
+function Show-Out([string]`$line) {
+  if (-not `$wantAccess -and -not `$all) { return }
+  if (`$all) { Write-Output ('[out] ' + `$line); return }
+  if (`$line -match '/wecom/callback|inject-text') {
+    Write-Output ('[out] ' + `$line)
  }
 }

+if (-not (Test-Path `$err)) { Write-Output ('MISSING ' + `$err) }
+
 if (-not `$wait) {
-  foreach (`$f in `$files) {
-    Write-Output ('--- ' + `$f + ' ---')
-    Get-Content -Path `$f -Tail `$tail -ErrorAction SilentlyContinue | ForEach-Object { Show-Line `$_ }
+  if (Test-Path `$err) {
+    Get-Content -Path `$err -Tail `$tail -Encoding `$enc -ErrorAction SilentlyContinue | ForEach-Object { Show-Err `$_ }
+  }
+  if (`$wantAccess -or `$all) {
+    Write-Output '=== access log (out) ==='
+    if (Test-Path `$out) {
+      Get-Content -Path `$out -Tail `$tail -Encoding `$enc -ErrorAction SilentlyContinue | ForEach-Object { Show-Out `$_ }
+    }
  }
  exit 0
 }

-Write-Output 'following... (Ctrl+C to stop)'
-Get-Content -Path `$files -Tail `$tail -Wait -ErrorAction SilentlyContinue | ForEach-Object { Show-Line `$_ }
+Write-Output 'following err.log ... (Ctrl+C to stop)'
+`$watch = @()
+if (Test-Path `$err) { `$watch += `$err }
+if ((`$wantAccess -or `$all) -and (Test-Path `$out)) { `$watch += `$out }
+if (`$watch.Count -eq 0) { throw 'no log files' }
+
+Get-Content -Path `$watch -Tail `$tail -Wait -Encoding `$enc -ErrorAction SilentlyContinue | ForEach-Object {
+  `$line = `$_
+  if (`$line -match '^\s*INFO:\s+\S+\s+-\s+"(GET|POST|PUT|DELETE)') {
+    Show-Out `$line
+  } else {
+    Show-Err `$line
+  }
+}
 "@

 $localTemp = Join-Path $env:TEMP 'ytd-watch-agent8910-remote.ps1'
-Set-Content -Path $localTemp -Value $remoteScript -Encoding UTF8
+[System.IO.File]::WriteAllText($localTemp, $remoteScript, $Utf8NoBom)

 & scp.exe -o BatchMode=yes -o ConnectTimeout=12 -i $SshKey $localTemp "${SshHost}:${IncomingRemotePath}"
 if ($LASTEXITCODE -ne 0) { throw "scp failed: $LASTEXITCODE" }

-& ssh.exe -o BatchMode=yes -o ConnectTimeout=12 -i $SshKey $SshHost `
+# 用 Process + UTF-8 读 SSH 输出，避免管道把中文解错再写入本地日志
+$sshArgs = @(
+  '-o', 'BatchMode=yes',
+  '-o', 'ConnectTimeout=12',
+  '-i', $SshKey,
+  $SshHost,
  "powershell -NoProfile -ExecutionPolicy Bypass -File `"$IncomingLocalRun`""
-exit $LASTEXITCODE
+)
+$psi = New-Object System.Diagnostics.ProcessStartInfo
+$psi.FileName = 'ssh.exe'
+$psi.Arguments = ($sshArgs | ForEach-Object {
+  if ($_ -match '\s') { '"' + $_ + '"' } else { $_ }
+}) -join ' '
+$psi.UseShellExecute = $false
+$psi.RedirectStandardOutput = $true
+$psi.RedirectStandardError = $true
+$psi.CreateNoWindow = $true
+$psi.StandardOutputEncoding = [System.Text.Encoding]::UTF8
+$psi.StandardErrorEncoding = [System.Text.Encoding]::UTF8
+$proc = New-Object System.Diagnostics.Process
+$proc.StartInfo = $psi
+[void]$proc.Start()
+
+while (-not $proc.StandardOutput.EndOfStream) {
+  $line = $proc.StandardOutput.ReadLine()
+  if ($null -ne $line) { Write-LocalLine $line }
+}
+$errTail = $proc.StandardError.ReadToEnd()
+if ($errTail) {
+  foreach ($eline in ($errTail -split "`r?`n")) {
+    if ($eline) { Write-LocalLine $eline }
+  }
+}
+$proc.WaitForExit()
+exit $proc.ExitCode
@@ -0,0 +1,38 @@
+同事如何使用「Watch-Agent8910」（只读盯服务器 8910 日志）
+
+一、需要具备
+1. Windows + 已安装 OpenSSH 客户端（Win10/11 一般自带 ssh/scp）
+2. 仓库里这两个文件（或单独拷贝 deploy 目录）：
+   - Watch-Agent8910.ps1
+   - Watch-Agent8910.bat（可选 Once 版：Watch-Agent8910-Once.bat）
+3. 能登录测试机的 SSH 私钥，且账号已加入服务器授权
+
+二、本机配置密钥（二选一）
+A. 默认路径（与现网脚本一致）
+   把私钥放到：%USERPROFILE%\.ssh\id_ed25519_ytd
+   （无扩展名；权限保持仅自己可读）
+
+B. 自定义路径
+   运行前设置环境变量，例如：
+   set YTD_SSH_KEY=D:\keys\my_ytd_ed25519
+   set YTD_SSH_HOST=ytd_extdev_0903@129.211.181.80
+
+三、怎么跑
+- 双击 Watch-Agent8910.bat → 持续监听（Ctrl+C 结束）
+- 双击 Watch-Agent8910-Once.bat → 只看最近一段
+- 或命令行：
+  Watch-Agent8910.bat
+  Watch-Agent8910.bat -Once
+  Watch-Agent8910.bat -Access
+
+本机同时落盘（仓库根目录 logs/，按当天日期追加写入）：
+  logs/20260913_info.log
+  logs/20260913_error.log
+不要提交 logs/ 到 Git。
+
+四、注意
+- 这是只读拉日志，不会改 Nginx / 不会切流
+- 私钥不要发到微信群或提交 Git；向负责人单独要密钥或加自己的公钥到服务器
+- 若提示禁止运行脚本：用 bat 即可（已带 Bypass），不要直接右键用受限策略跑 ps1
+- 乱码时用：Watch-Agent8910.bat -Gbk
+- 只要屏幕不要文件：Watch-Agent8910.bat -NoLocalLog
@@ -75,6 +75,8 @@ $runHttp = @(
  ('cd /d "' + $AgentCurrent + '"'),
  'set YTD_ENV=test',
  'set AGENT_HTTP_PORT=8910',
+  'set PYTHONUTF8=1',
+  'set PYTHONIOENCODING=utf-8',
  ':loop',
  ('"' + $venvPy + '" -m agent.http_app >> "' + $logHttpOut + '" 2>> "' + $logHttpErr + '"'),
  ('echo agent exited, restart in 3s >> "' + $logHttpErr + '"'),
@@ -87,6 +89,8 @@ $runWorker = @(
  '@echo off',
  ('cd /d "' + $AgentCurrent + '"'),
  'set YTD_ENV=test',
+  'set PYTHONUTF8=1',
+  'set PYTHONIOENCODING=utf-8',
  ':loop',
  ('"' + $venvPy + '" -m agent.worker_app >> "' + $logWorkerOut + '" 2>> "' + $logWorkerErr + '"'),
  ('echo worker exited, restart in 3s >> "' + $logWorkerErr + '"'),