feat(log): H-03 日志 flush 错误监控

- 添加 ErrorMetrics 结构体记录 flush/write/drop 错误数
- 添加 ErrorHandler 回调接口用于自定义错误处理
- 在 AsyncLogWriterOptions 中支持配置错误处理器
- 在 flushBatch 中记录 flush 错误指标并回调错误处理器
- 在 enqueue fallback 路径中记录丢弃事件数
- 添加 Metrics() 方法暴露错误统计

internal/routing/logwriter.go

@@ -69,11 +69,57 @@ type routeLogSink interface {
 	Close() error
 }
 
+type ErrorMetrics struct {
+	FlushErrors  int64
+	WriteErrors  int64
+	DroppedEvents int64
+	mu           sync.RWMutex
+}
+
+func (e *ErrorMetrics) RecordFlushError() {
+	e.mu.Lock()
+	defer e.mu.Unlock()
+	e.FlushErrors++
+}
+
+func (e *ErrorMetrics) RecordWriteError() {
+	e.mu.Lock()
+	defer e.mu.Unlock()
+	e.WriteErrors++
+}
+
+func (e *ErrorMetrics) RecordDroppedEvent() {
+	e.mu.Lock()
+	defer e.mu.Unlock()
+	e.DroppedEvents++
+}
+
+func (e *ErrorMetrics) GetFlushErrors() int64 {
+	e.mu.RLock()
+	defer e.mu.RUnlock()
+	return e.FlushErrors
+}
+
+func (e *ErrorMetrics) GetWriteErrors() int64 {
+	e.mu.RLock()
+	defer e.mu.RUnlock()
+	return e.WriteErrors
+}
+
+func (e *ErrorMetrics) GetDroppedEvents() int64 {
+	e.mu.RLock()
+	defer e.mu.RUnlock()
+	return e.DroppedEvents
+}
+
+type ErrorHandler func(ctx context.Context, err error, eventType string)
+
 type AsyncLogWriterOptions struct {
 	QueueSize            int
 	FlushInterval        time.Duration
 	MaxBatchSize         int
 	FallbackWriteTimeout time.Duration
+	OnError              ErrorHandler
 }
 
 type queuedLogEvent struct {
@@ -92,6 +138,8 @@ type AsyncLogWriter struct {
 	maxBatchSize         int
 	fallbackWriteTimeout time.Duration
 	once                 sync.Once
+	onError              ErrorHandler
+	metrics              ErrorMetrics
 }
 
 func NewSQLiteLogWriter(ctx context.Context, sqliteDSN string, opts AsyncLogWriterOptions) (*AsyncLogWriter, error) {
@@ -125,6 +173,7 @@ func NewAsyncLogWriter(sink routeLogSink, opts AsyncLogWriterOptions) *AsyncLogW
 		flushInterval:        opts.FlushInterval,
 		maxBatchSize:         opts.MaxBatchSize,
 		fallbackWriteTimeout: opts.FallbackWriteTimeout,
+		onError:              opts.OnError,
 	}
 	go writer.loop()
 	return writer
@@ -213,9 +262,15 @@ func (w *AsyncLogWriter) enqueue(ctx context.Context, item queuedLogEvent) error
 	case w.queue <- item:
 		return nil
 	default:
+		// Queue is full, use fallback direct write
+		w.metrics.RecordDroppedEvent()
 		fallbackCtx, cancel := context.WithTimeout(context.Background(), w.fallbackWriteTimeout)
 		defer cancel()
-		return w.writeOne(fallbackCtx, item)
+		err := w.writeOne(fallbackCtx, item)
+		if err != nil && w.onError != nil {
+			w.onError(fallbackCtx, err, w.getEventType(item))
+		}
+		return err
 	}
 }
 
@@ -230,12 +285,43 @@ func (w *AsyncLogWriter) flushBatch(batch []queuedLogEvent) error {
 			if firstErr == nil {
 				firstErr = err
 			}
+			w.metrics.RecordFlushError()
 			log.Printf("routing: flush route log event failed: %v", err)
+
+			// Call error handler if configured
+			if w.onError != nil {
+				w.onError(context.Background(), err, w.getEventType(item))
+			}
 		}
 	}
+	if firstErr != nil {
+		w.metrics.RecordWriteError()
+	}
 	return firstErr
 }
 
+func (w *AsyncLogWriter) getEventType(item queuedLogEvent) string {
+	switch {
+	case item.decision != nil:
+		return "decision"
+	case item.failover != nil:
+		return "failover"
+	case item.sticky != nil:
+		return "sticky_audit"
+	default:
+		return "unknown"
+	}
+}
+
+// Metrics returns the error metrics for monitoring
+func (w *AsyncLogWriter) Metrics() ErrorMetrics {
+	return ErrorMetrics{
+		FlushErrors:   w.metrics.GetFlushErrors(),
+		WriteErrors:   w.metrics.GetWriteErrors(),
+		DroppedEvents: w.metrics.GetDroppedEvents(),
+	}
+}
+
 func (w *AsyncLogWriter) drainQueue(batch []queuedLogEvent) []queuedLogEvent {
 	for {
 		select {