feat(panel): 数据库故障降级、健康告警与检测

- 网站统计聚合器 flush 失败也 commit（丢弃已 drain 增量），防止 DB
  长期不可写时内存无界累积导致 OOM/CPU 100%
- 势态感知同样在 upsert 失败时上报健康问题
- 新增 internal/app.Health 全局健康状态注册表，通过 /home/health 暴露
- 前端 layout 顶部新增 HealthBanner，DB 故障时红色横幅提醒并给出修复引导
- acepanel fix 检测数据库损坏改用 PRAGMA quick_check，能识别 malformed
  等仅在查询时才暴露的静默损坏
- 主库与辅助库 DSN 一律设为 synchronous(FULL)，减少宿主 write-back
  缓存丢失 fsync 导致的损坏概率
- 移除 debug.SetMemoryLimit(256MB)：软上限在活对象超限时会驱动 GC 反复
  追赶不可回收对象，反而放大 CPU 打满

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

cmd/ace/main.go

@@ -12,7 +12,6 @@ func main() {
 	}
 
 	debug.SetGCPercent(10)
-	debug.SetMemoryLimit(256 << 20)
 
 	ace, err := initAce()
 	if err != nil {

internal/app/health.go

@@ -0,0 +1,73 @@
+package app
+
+import (
+	"sort"
+	"sync"
+	"time"
+)
+
+// HealthLevelError 需要用户立即处理
+const HealthLevelError = "error"
+
+// HealthLevelWarning 提示性问题，通常已被系统自动降级处理
+const HealthLevelWarning = "warning"
+
+// HealthIssue 单条健康问题
+// Key 为稳定标识符（如 database:stat），前端据此选择翻译文案
+// Message 为原始错误详情，供诊断参考
+type HealthIssue struct {
+	Key     string    `json:"key"`
+	Level   string    `json:"level"`
+	Message string    `json:"message"`
+	Since   time.Time `json:"since"`
+}
+
+type healthRegistry struct {
+	mu     sync.RWMutex
+	issues map[string]HealthIssue
+}
+
+// Health 全局健康状态注册表，供各后台任务上报/清除故障
+var Health = &healthRegistry{issues: make(map[string]HealthIssue)}
+
+// Report 上报或更新一条健康问题
+// 同 key 重复上报时保留首次上报时间，仅更新 message 和 level
+func (h *healthRegistry) Report(key, level, message string) {
+	h.mu.Lock()
+	defer h.mu.Unlock()
+
+	if existing, ok := h.issues[key]; ok {
+		existing.Level = level
+		existing.Message = message
+		h.issues[key] = existing
+		return
+	}
+	h.issues[key] = HealthIssue{
+		Key:     key,
+		Level:   level,
+		Message: message,
+		Since:   time.Now(),
+	}
+}
+
+// Clear 清除指定 key 的健康问题
+func (h *healthRegistry) Clear(key string) {
+	h.mu.Lock()
+	defer h.mu.Unlock()
+	delete(h.issues, key)
+}
+
+// Snapshot 返回当前所有健康问题，按 Since 升序（越早发生越靠前）
+func (h *healthRegistry) Snapshot() []HealthIssue {
+	h.mu.RLock()
+	defer h.mu.RUnlock()
+
+	result := make([]HealthIssue, 0, len(h.issues))
+	for _, issue := range h.issues {
+		result = append(result, issue)
+	}
+	sort.Slice(result, func(i, j int) bool {
+		return result[i].Since.Before(result[j].Since)
+	})
+	return result
+}

internal/bootstrap/db.go

@@ -32,7 +32,7 @@ func NewDB(conf *config.Config) (*gorm.DB, error) {
 		options = append(options, sloggorm.WithTraceAll())
 	}
 
-	db, err := gorm.Open(sqlite.Open("file:"+filepath.Join(app.Root, "panel/storage/panel.db")+"?_txlock=immediate&_pragma=busy_timeout(10000)&_pragma=journal_mode(WAL)"),
+	db, err := gorm.Open(sqlite.Open("file:"+filepath.Join(app.Root, "panel/storage/panel.db")+"?_txlock=immediate&_pragma=busy_timeout(10000)&_pragma=journal_mode(WAL)&_pragma=synchronous(FULL)"),
 		&gorm.Config{
 			Logger:                                   sloggorm.New(options...),
 			SkipDefaultTransaction:                   true,

internal/data/backup.go

@@ -1211,8 +1211,8 @@ func (r *backupRepo) FixPanel() error {
 	panelBroken := !io.Exists(filepath.Join(app.Root, "panel", "ace")) ||
 		!io.Exists(filepath.Join(app.Root, "panel", "storage", "config.yml")) ||
 		!io.Exists(filepath.Join(app.Root, "panel", "storage", "panel.db"))
-	// 检查主数据库连接
-	if err := r.db.Exec("VACUUM").Error; err != nil {
+	// 检查主数据库完整性
+	if !quickCheck(r.db) {
 		panelBroken = true
 	}
 	if err := r.db.Exec("PRAGMA wal_checkpoint(TRUNCATE);").Error; err != nil {
@@ -1223,13 +1223,17 @@ func (r *backupRepo) FixPanel() error {
 	var brokenAuxDBs []string
 	for _, name := range []string{"stat", "scan"} {
 		auxDB, err := openDB(name)
-		if err == nil {
-			if sqlDB, dbErr := auxDB.DB(); dbErr == nil {
-				_ = sqlDB.Close()
-			}
+		if err != nil {
+			brokenAuxDBs = append(brokenAuxDBs, name)
 			continue
 		}
-		brokenAuxDBs = append(brokenAuxDBs, name)
+		ok := quickCheck(auxDB)
+		if sqlDB, dbErr := auxDB.DB(); dbErr == nil {
+			_ = sqlDB.Close()
+		}
+		if !ok {
+			brokenAuxDBs = append(brokenAuxDBs, name)
+		}
 	}
 
 	// 一切正常，无需修复

internal/data/helper.go

@@ -55,7 +55,7 @@ func getOperatorID(ctx context.Context) uint64 {
 // openDB 打开数据库
 func openDB(name string) (*gorm.DB, error) {
 	dsn := "file:" + filepath.Join(app.Root, fmt.Sprintf("panel/storage/%s.db", name)) +
-		"?_txlock=immediate&_pragma=busy_timeout(10000)&_pragma=journal_mode(WAL)&_pragma=synchronous(NORMAL)"
+		"?_txlock=immediate&_pragma=busy_timeout(10000)&_pragma=journal_mode(WAL)&_pragma=synchronous(FULL)"
 	db, err := gorm.Open(sqlite.Open(dsn), &gorm.Config{
 		SkipDefaultTransaction:                   true,
 		DisableForeignKeyConstraintWhenMigrating: true,
@@ -69,6 +69,15 @@ func openDB(name string) (*gorm.DB, error) {
 	return db, nil
 }
 
+// quickCheck 用 PRAGMA quick_check 检测 SQLite 数据库是否损坏
+func quickCheck(db *gorm.DB) bool {
+	var result string
+	if err := db.Raw("PRAGMA quick_check").Row().Scan(&result); err != nil {
+		return false
+	}
+	return result == "ok"
+}
+
 // upsert 分批大小，避免超出 SQLite 变量数限制
 const upsertBatchSize = 100
 

internal/job/firewall_scan.go

@@ -17,6 +17,9 @@ import (
 	"github.com/acepanel/panel/v3/pkg/geoip"
 )
 
+// healthKeyScanDB 扫描事件辅助数据库健康问题上报 key
+const healthKeyScanDB = "database:scan"
+
 // ipCounter 单个 IP 的扫描计数器
 type ipCounter struct {
 	count     uint
@@ -187,7 +190,10 @@ func (r *FirewallScan) flush() {
 
 	if err := r.scanRepo.Upsert(events); err != nil {
 		r.log.Warn("failed to upsert scan events", slog.Any("err", err))
+		app.Health.Report(healthKeyScanDB, app.HealthLevelError, err.Error())
+		return
 	}
+	app.Health.Clear(healthKeyScanDB)
 }
 
 // ensureFirewall 懒加载防火墙实例

internal/job/website_stat.go

@@ -15,6 +15,9 @@ import (
 	"github.com/acepanel/panel/v3/pkg/websitestat"
 )
 
+// healthKeyStatDB 网站统计辅助数据库健康问题上报 key
+const healthKeyStatDB = "database:stat"
+
 // WebsiteStat 网站统计后台任务
 type WebsiteStat struct {
 	log          *slog.Logger
@@ -177,9 +180,13 @@ func (r *WebsiteStat) flush() {
 
 	if err := r.statRepo.Upsert(stats); err != nil {
 		r.log.Warn("failed to upsert website stats", slog.Any("err", err))
+		app.Health.Report(healthKeyStatDB, app.HealthLevelError, err.Error())
+		// 直接丢弃已 drain 的增量，避免 DB 长期不可写时内存无界累积
+		commit()
 		return
 	}
 	commit()
+	app.Health.Clear(healthKeyStatDB)
 }
 
 // flushErrors 将错误日志缓冲写入数据库
@@ -205,9 +212,13 @@ func (r *WebsiteStat) flushErrors() {
 
 	if err := r.statRepo.InsertErrors(errors); err != nil {
 		r.log.Warn("failed to insert website error logs", slog.Any("err", err))
+		app.Health.Report(healthKeyStatDB, app.HealthLevelError, err.Error())
+		// 直接丢弃已 snapshot 的错误缓冲，避免内存无界累积
+		commit()
 		return
 	}
 	commit()
+	app.Health.Clear(healthKeyStatDB)
 }
 
 // flushDetails 将详细统计增量写入数据库（蜘蛛/客户端/IP/URI）
@@ -286,25 +297,29 @@ func (r *WebsiteStat) flushDetails() {
 		}
 	}
 
-	failed := false
+	var lastErr error
 	if err := r.statRepo.UpsertSpiders(spiders); err != nil {
 		r.log.Warn("failed to upsert spider stats", slog.Any("err", err))
-		failed = true
+		lastErr = err
 	}
 	if err := r.statRepo.UpsertClients(clients); err != nil {
 		r.log.Warn("failed to upsert client stats", slog.Any("err", err))
-		failed = true
+		lastErr = err
 	}
 	if err := r.statRepo.UpsertIPs(ips); err != nil {
 		r.log.Warn("failed to upsert ip stats", slog.Any("err", err))
-		failed = true
+		lastErr = err
 	}
 	if err := r.statRepo.UpsertURIs(uris); err != nil {
 		r.log.Warn("failed to upsert uri stats", slog.Any("err", err))
-		failed = true
+		lastErr = err
 	}
-	if !failed {
-		commit()
+	// 无论成败都 commit：成功→减去已入库增量；失败→直接丢弃，避免内存无界累积
+	commit()
+	if lastErr != nil {
+		app.Health.Report(healthKeyStatDB, app.HealthLevelError, lastErr.Error())
+	} else {
+		app.Health.Clear(healthKeyStatDB)
 	}
 }
 

internal/route/http.go

@@ -245,6 +245,7 @@ func (route *Http) Register(r *chi.Mux) {
 			r.Post("/restart_server", route.home.RestartServer)
 			r.Get("/runtime_info", route.home.RuntimeInfo)
 			r.Get("/goroutines", route.home.Goroutines)
+			r.Get("/health", route.home.Health)
 		})
 
 		r.Route("/task", func(r chi.Router) {

internal/service/home.go

@@ -522,3 +522,8 @@ func (s *HomeService) Goroutines(w http.ResponseWriter, r *http.Request) {
 
 	Success(w, goroutines)
 }
+
+// Health 返回当前所有健康问题，供前端全局横幅展示
+func (s *HomeService) Health(w http.ResponseWriter, r *http.Request) {
+	Success(w, app.Health.Snapshot())
+}

web/src/api/panel/home/index.ts

@@ -30,4 +30,6 @@ export default {
   runtimeInfo: (): any => http.Get('/home/runtime_info'),
   // Goroutine 列表
   goroutines: (): any => http.Get('/home/goroutines'),
+  // 面板健康问题列表（全局横幅使用）
+  health: (): any => http.Get('/home/health', { meta: { noAlert: true } }),
 }

web/src/components/system/HealthBanner.vue

@@ -0,0 +1,72 @@
+
+<script lang="ts" setup>
+import { useGettext } from 'vue3-gettext'
+
+import home from '@/api/panel/home'
+
+interface HealthIssue {
+  key: string
+  level: 'error' | 'warning'
+  message: string
+  since: string
+}
+
+const { $gettext } = useGettext()
+
+const { data } = useAutoRequest(() => home.health(), { initialData: [] })
+
+// 前端根据后端稳定 key 选择本地化文案，避免依赖后端错误详情
+const titleFor = (issue: HealthIssue): string => {
+  switch (issue.key) {
+    case 'database:panel':
+      return $gettext('Main database write failed, panel may not work properly')
+    case 'database:stat':
+      return $gettext('Website statistics database write failed, statistics may be inaccurate')
+    case 'database:scan':
+      return $gettext('Scan events database write failed, situation awareness may be inaccurate')
+    default:
+      return $gettext('Panel encountered a fault: %{key}', { key: issue.key })
+  }
+}
+
+const hintFor = (issue: HealthIssue): string => {
+  switch (issue.key) {
+    case 'database:panel':
+    case 'database:stat':
+    case 'database:scan':
+      return $gettext(
+        'Try running "acepanel fix" on the server to detect and rebuild the broken database file.'
+      )
+    default:
+      return ''
+  }
+}
+
+// 一次性显示所有问题，最严重的靠前
+const sorted = computed<HealthIssue[]>(() => {
+  const order: Record<string, number> = { error: 0, warning: 1 }
+  const list = (data.value ?? []) as HealthIssue[]
+  return [...list].sort((a, b) => (order[a.level] ?? 9) - (order[b.level] ?? 9))
+})
+</script>
+
+<template>
+  <div v-if="sorted.length" class="flex flex-col gap-1 border-b border-border-default">
+    <n-alert
+      v-for="issue in sorted"
+      :key="issue.key"
+      :type="issue.level"
+      :show-icon="true"
+      :bordered="false"
+      class="rounded-none"
+    >
+      <div class="flex flex-col">
+        <span class="font-medium">{{ titleFor(issue) }}</span>
+        <span v-if="hintFor(issue)" class="text-xs opacity-80">{{ hintFor(issue) }}</span>
+        <span v-if="issue.message" class="text-xs opacity-60 font-mono">
+          {{ issue.message }}
+        </span>
+      </div>
+    </n-alert>
+  </div>
+</template>

web/src/layouts/IndexView.vue

@@ -1,6 +1,8 @@
 <script lang="ts" setup>
 import { useThemeStore } from '@/stores'
 
+import HealthBanner from '@/components/system/HealthBanner.vue'
+
 import AppMain from './AppMain.vue'
 import AppHeader from './header/IndexView.vue'
 import SideBar from './sidebar/IndexView.vue'
@@ -55,6 +57,7 @@ onBeforeUnmount(() => window.removeEventListener('resize', handleResize))
       >
         <app-header />
       </header>
+      <health-banner />
       <section class="bg-bg-base flex flex-col flex-1 overflow-hidden">
         <app-main />
       </section>