优化生产日志体积：访问降噪、压缩清理与退款告警节流

跳过轮询/封面等热路径成功请求，按日 gzip 并保留可配天数，退款 max-retry 汇总告警。
2026-07-13 22:02:20 +08:00
parent f51aed1f0f
commit 2efcede800
8 changed files with 361 additions and 41 deletions
@@ -225,14 +225,25 @@ func (j *Job) warnMaxRetryRefunds(ctx context.Context, now time.Time) {
 		j.logger.Warn("refund retry max count scan failed", zap.Error(err))
 		return
 	}
+	if len(rows) == 0 {
+		return
+	}
+	// 汇总一条告警，避免每 2 分钟对同一批单刷屏；更新 last_retry_at 实现 24h 节流
+	ids := make([]uint64, 0, len(rows))
+	orderIDs := make([]uint64, 0, len(rows))
 	for _, row := range rows {
-		j.logger.Warn("refund retry reached max count, need manual check",
-			zap.Uint64("payment_id", row.ID),
-			zap.Uint64("order_id", row.OrderID),
-			zap.String("biz_type", row.BizType),
-			zap.String("status", row.Status),
-			zap.Int("retry_count", row.RetryCount),
-		)
+		ids = append(ids, row.ID)
+		orderIDs = append(orderIDs, row.OrderID)
+	}
+	j.logger.Warn("refund retry reached max count, need manual check",
+		zap.Int("count", len(rows)),
+		zap.Uint64s("payment_ids", ids),
+		zap.Uint64s("order_ids", orderIDs),
+	)
+	if err := j.db.WithContext(ctx).Model(&model.PaymentOrder{}).
+		Where("id IN ?", ids).
+		Update("last_retry_at", now).Error; err != nil {
+		j.logger.Warn("refund retry bump last_retry_at failed", zap.Error(err))
 	}
 }

@@ -248,6 +259,8 @@ func (j *Job) warnMissingRefundOrders(ctx context.Context, now time.Time) (int,
 	}
 	missing := 0
 	rebuilt := 0
+	var missingOrderIDs []uint64
+	var rebuildFailOrderIDs []uint64
 	for _, row := range rows {
 		var count int64
 		if err := j.db.WithContext(ctx).Model(&model.PaymentOrder{}).
@@ -264,18 +277,11 @@ func (j *Job) warnMissingRefundOrders(ctx context.Context, now time.Time) (int,
 		// 非仲裁孤儿 biz_type 无法从订单状态可靠区分，维持只告警。
 		if row.SettlementStatus == "arbitrated" {
 			if j.payments == nil {
-				j.logger.Warn("auto rebuild arbitration refund skipped, payment repository unavailable",
-					zap.Uint64("order_id", row.ID),
-					zap.Int64("refund_amount_cent", row.RefundAmountCent),
-				)
+				rebuildFailOrderIDs = append(rebuildFailOrderIDs, row.ID)
 				continue
 			}
 			if _, err := j.payments.StartRefund(ctx, row.ID, row.RefundAmountCent, "arbitration_refund", "仲裁退款补偿补建"); err != nil {
-				j.logger.Warn("auto rebuild arbitration refund failed, need manual check",
-					zap.Uint64("order_id", row.ID),
-					zap.Int64("refund_amount_cent", row.RefundAmountCent),
-					zap.Error(err),
-				)
+				rebuildFailOrderIDs = append(rebuildFailOrderIDs, row.ID)
 			} else {
 				rebuilt++
 				j.logger.Info("auto rebuild arbitration refund submitted",
@@ -285,13 +291,36 @@ func (j *Job) warnMissingRefundOrders(ctx context.Context, now time.Time) (int,
 			}
 			continue
 		}
+		missingOrderIDs = append(missingOrderIDs, row.ID)
+	}
+	// 汇总告警 + Redis 24h 去重，避免每轮扫描刷屏
+	if len(missingOrderIDs) > 0 && j.shouldWarn(ctx, "missing_payment", 0) {
 		j.logger.Warn("refund order missing payment record, need manual check",
-			zap.Uint64("order_id", row.ID),
-			zap.String("order_no", row.OrderNo),
-			zap.String("refund_status", row.RefundStatus),
-			zap.String("settlement_status", row.SettlementStatus),
-			zap.Int64("refund_amount_cent", row.RefundAmountCent),
+			zap.Int("count", len(missingOrderIDs)),
+			zap.Uint64s("order_ids", missingOrderIDs),
+		)
+	}
+	if len(rebuildFailOrderIDs) > 0 && j.shouldWarn(ctx, "rebuild_fail", 0) {
+		j.logger.Warn("auto rebuild arbitration refund failed, need manual check",
+			zap.Int("count", len(rebuildFailOrderIDs)),
+			zap.Uint64s("order_ids", rebuildFailOrderIDs),
 		)
 	}
 	return missing, rebuilt, nil
 }
+
+// shouldWarn 用 Redis 做 24h 节流；无 Redis 时每个 job 周期最多打一次同类汇总（由调用方聚合）。
+func (j *Job) shouldWarn(ctx context.Context, kind string, id uint64) bool {
+	if j.redis == nil {
+		return true
+	}
+	key := fmt.Sprintf("hfb:job:refundretry:warn:%s", kind)
+	if id > 0 {
+		key = fmt.Sprintf("%s:%d", key, id)
+	}
+	ok, err := j.redis.SetNX(ctx, key, "1", manualWarnThreshold).Result()
+	if err != nil {
+		return true
+	}
+	return ok
+}
@@ -46,6 +46,50 @@ func TestRetryBackoffCapsAtOneHour(t *testing.T) {
 	}
 }

+func TestWarnMaxRetryRefundsThrottlesByLastRetryAt(t *testing.T) {
+	db := setupRefundRetryTestDB(t)
+	job := New(db, nil, zap.NewNop(), nil)
+	now := time.Date(2026, 7, 11, 12, 0, 0, 0, time.UTC)
+	old := now.Add(-25 * time.Hour)
+	row := model.PaymentOrder{
+		PaymentNo:    "PAY202607110001",
+		OrderID:      9,
+		OrderNo:      "ORD202607110001",
+		UserID:       2,
+		Provider:     "mock",
+		ThirdOrderID: "REF202607110001",
+		BizType:      "admin_refund",
+		Status:       "failed",
+		RetryCount:   maxRetryCount,
+		LastRetryAt:  &old,
+	}
+	if err := db.Create(&row).Error; err != nil {
+		t.Fatalf("创建退款单失败: %v", err)
+	}
+
+	job.warnMaxRetryRefunds(t.Context(), now)
+
+	var latest model.PaymentOrder
+	if err := db.First(&latest, row.ID).Error; err != nil {
+		t.Fatalf("查询退款单失败: %v", err)
+	}
+	if latest.LastRetryAt == nil || !latest.LastRetryAt.Equal(now) {
+		t.Fatalf("LastRetryAt 应更新为 now 以节流，got %v", latest.LastRetryAt)
+	}
+
+	// 立即再扫不应匹配（last_retry_at 未超过 24h）
+	var count int64
+	if err := db.Model(&model.PaymentOrder{}).
+		Where("retry_count >= ? AND (last_retry_at IS NULL OR last_retry_at <= ?)",
+			maxRetryCount, now.Add(-manualWarnThreshold)).
+		Count(&count).Error; err != nil {
+		t.Fatalf("count: %v", err)
+	}
+	if count != 0 {
+		t.Fatalf("throttled rows count = %d, want 0", count)
+	}
+}
+
 func TestMarkRetryFailedStopsAtMaxCount(t *testing.T) {
 	db := setupRefundRetryTestDB(t)
 	job := New(db, nil, zap.NewNop(), nil)