From 181b338780a8be31e03f34128014aaddf7589c2c Mon Sep 17 00:00:00 2001
From: Zhou Weidong <502333560@qq.com>
Date: Wed, 28 Oct 2026 18:23:48 +0800
Subject: [PATCH] =?UTF-8?q?fix(ops):=20=E4=BF=AE=E5=A4=8D=E7=AE=A1?=
 =?UTF-8?q?=E7=90=86=E7=AB=AF=20Redis=20=E6=9C=AA=E9=85=8D=E7=BD=AE?=
 =?UTF-8?q?=E4=B8=8E=E7=9C=8B=E9=97=A8=E7=8B=97=E6=9C=AA=E7=9C=9F=E6=AD=A3?=
 =?UTF-8?q?=E5=91=A8=E6=9C=9F=E5=B7=A1=E6=A3=80?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- admin-api 的 application.yml 缺少 spring.data.redis 配置，容器内默认连
  127.0.0.1 导致 /actuator/health 返回 DOWN（redis 组件不可用）；
  补齐后与 app-api 共用 REDIS_HOST/REDIS_PORT，健康检查恢复 UP
- 计划任务原先用 BootTrigger 挂 Repetition，Windows 不生效（Next Run Time
  为 N/A，9 小时只跑过一次），看门狗形同虚设；改为 TimeTrigger + PT2M
  无限重复 + StartWhenAvailable，实测每 2 分钟稳定巡检
- watchdog.ps1 的连续失败计数改为落盘 watchdog-state.json，避免每轮进程
  重启后清零导致「连续 3 次失败告警」永不触发
---
 deploy/backend-watchdog-task.xml              | 10 ++++++++--
 deploy/watchdog.ps1                           | 19 ++++++++++++++++++-
 .../src/main/resources/application.yml        |  6 ++++++
 3 files changed, 32 insertions(+), 3 deletions(-)

diff --git a/deploy/backend-watchdog-task.xml b/deploy/backend-watchdog-task.xml
index b58a9cf..88f59d2 100644
--- a/deploy/backend-watchdog-task.xml
+++ b/deploy/backend-watchdog-task.xml
@@ -17,11 +17,17 @@
     <ExecutionTimeLimit>PT10M</ExecutionTimeLimit>
   </Settings>
   <Triggers>
-    <BootTrigger>
+    <!-- BootTrigger 上挂 <Repetition> 在 Windows 计划任务里不生效（导入后 Next Run Time 为 N/A，
+         实际只在开机跑一次）。改成 TimeTrigger + 无限重复，配合 StartWhenAvailable
+         保证开机后若错过也会立即补跑；这样才真正做到每 2 分钟巡检一次。 -->
+    <TimeTrigger>
+      <StartBoundary>2026-01-01T00:00:00</StartBoundary>
       <Repetition>
         <Interval>PT2M</Interval>
+        <Duration>P3650D</Duration>
+        <StopAtDurationEnd>false</StopAtDurationEnd>
       </Repetition>
-    </BootTrigger>
+    </TimeTrigger>
   </Triggers>
   <Actions Context="Author">
     <Exec>
diff --git a/deploy/watchdog.ps1 b/deploy/watchdog.ps1
index 404211f..3dde3d8 100644
--- a/deploy/watchdog.ps1
+++ b/deploy/watchdog.ps1
@@ -91,8 +91,18 @@ $checks = @(
     @{ Name = "Prometheus"; Type = "http";     Url = "http://127.0.0.1:9090/-/healthy"; Service = "prometheus"; Container = "fruitshorts-prometheus" }
 )
 
-# 跟踪每个服务的连续失败次数
+# 跟踪每个服务的连续失败次数。脚本由计划任务每 2 分钟拉起一次，
+# 进程内变量会跨轮丢失，因此计数持久化到状态文件，否则「连续 3 次告警」永远不触发。
+$StateFile = Join-Path $LogDir "watchdog-state.json"
 $failCounts = @{}
+if (Test-Path $StateFile) {
+    try {
+        (Get-Content $StateFile -Raw | ConvertFrom-Json).PSObject.Properties |
+            ForEach-Object { $failCounts[$_.Name] = [int]$_.Value }
+    } catch {
+        Write-Log "状态文件损坏，重置连续失败计数"
+    }
+}
 
 foreach ($check in $checks) {
     $name = $check.Name
@@ -168,3 +178,10 @@ if ($drive) {
         Send-Alert "磁盘告警" "C 盘使用率 ${pct}%，请清理日志或扩容"
     }
 }
+
+# 落盘本轮结束时的连续失败计数，供下一轮接着累计
+try {
+    $failCounts | ConvertTo-Json | Set-Content -Path $StateFile -Encoding UTF8
+} catch {
+    Write-Log "状态文件写入失败: $_"
+}
diff --git a/fruitshorts-admin-api/src/main/resources/application.yml b/fruitshorts-admin-api/src/main/resources/application.yml
index 41372e4..2930a0c 100644
--- a/fruitshorts-admin-api/src/main/resources/application.yml
+++ b/fruitshorts-admin-api/src/main/resources/application.yml
@@ -14,6 +14,12 @@ spring:
     url: ${DB_URL:jdbc:postgresql://127.0.0.1:5432/fruitshorts}
     username: ${DB_USER:fruitshorts}
     password: ${DB_PASSWORD:fruitshorts}
+  data:
+    redis:
+      # 与 app-api 共用同一个 Redis 实例；容器内必须显式指定 host，
+      # 否则 Spring Boot 默认连 127.0.0.1（容器内即自身）导致健康检查报 DOWN。
+      host: ${REDIS_HOST:127.0.0.1}
+      port: ${REDIS_PORT:6379}
   jackson:
     date-format: yyyy-MM-dd HH:mm:ss
     time-zone: Asia/Shanghai