chore: rename activation context migration to 02151

Co-authored-by: multica-agent <github@multica.ai>

initialize/migrate/database/02151_order_activation_context.down.sql

@@ -0,0 +1,2 @@
+-- Remove activation_context column from order table
+ALTER TABLE `order` DROP COLUMN IF EXISTS `activation_context`;

initialize/migrate/database/02151_order_activation_context.up.sql

@@ -0,0 +1,6 @@
+-- Add activation_context column to order table for Redis fallback persistence (idempotent)
+SET @col_exists = (SELECT COUNT(*) FROM INFORMATION_SCHEMA.COLUMNS WHERE TABLE_SCHEMA = DATABASE() AND TABLE_NAME = 'order' AND COLUMN_NAME = 'activation_context');
+SET @sql = IF(@col_exists = 0, 'ALTER TABLE `order` ADD COLUMN `activation_context` TEXT DEFAULT NULL COMMENT ''Activation context JSON (guest/redemption info for DB fallback)'' AFTER `app_account_token`', 'SELECT 1');
+PREPARE stmt FROM @sql;
+EXECUTE stmt;
+DEALLOCATE PREPARE stmt;

internal/logic/notify/appleIAPNotifyLogic.go

@@ -3,17 +3,20 @@ package notify
 import (
 	"context"
 	"encoding/json"
+	"fmt"
 	"strconv"
 	"strings"
 
 	commonLogic "github.com/perfect-panel/server/internal/logic/common"
 	iapmodel "github.com/perfect-panel/server/internal/model/iap/apple"
+	"github.com/perfect-panel/server/internal/model/order"
 	"github.com/perfect-panel/server/internal/model/subscribe"
 	"github.com/perfect-panel/server/internal/model/user"
 	"github.com/perfect-panel/server/internal/svc"
 	"github.com/perfect-panel/server/internal/types"
 	iapapple "github.com/perfect-panel/server/pkg/iap/apple"
 	"github.com/perfect-panel/server/pkg/logger"
+	"github.com/perfect-panel/server/pkg/tool"
 	"gorm.io/gorm"
 )
 
@@ -84,6 +87,7 @@ func (l *AppleIAPNotifyLogic) Handle(signedPayload string) error {
 				l.Errorw("iap notify insert transaction error", logger.Field("error", e.Error()), logger.Field("productId", txPayload.ProductId), logger.Field("originalTransactionId", txPayload.OriginalTransactionId))
 				return e
 			}
+			existing = rec
 		} else {
 			if txPayload.RevocationDate != nil {
 				// 撤销场景：更新 revocation_at
@@ -96,6 +100,32 @@ func (l *AppleIAPNotifyLogic) Handle(signedPayload string) error {
 				}
 			}
 		}
+
+		// Fix Bug 7: resolve userId when the IAP transaction record has no associated user.
+		// This can happen if the first notification for a subscription arrived before the
+		// in-app purchase flow created the order (race) or the order was made in a previous
+		// build that did not write user_id to the IAP transaction table.
+		if existing != nil && existing.UserId == 0 {
+			var origOrder order.Order
+			if lookupErr := db.Model(&order.Order{}).
+				Where("trade_no = ? AND method = ?", txPayload.OriginalTransactionId, "apple_iap").
+				Order("id ASC").
+				First(&origOrder).Error; lookupErr == nil && origOrder.UserId > 0 {
+				existing.UserId = origOrder.UserId
+				_ = db.Model(&iapmodel.Transaction{}).
+					Where("id = ?", existing.Id).
+					Update("user_id", origOrder.UserId).Error
+				l.Infow("iap notify resolved zero userId from original purchase order",
+					logger.Field("userId", origOrder.UserId),
+					logger.Field("originalTransactionId", txPayload.OriginalTransactionId),
+				)
+			} else {
+				l.Errorw("CRITICAL: iap notify UserId=0 and cannot resolve from order, notification dropped",
+					logger.Field("originalTransactionId", txPayload.OriginalTransactionId))
+				return fmt.Errorf("iap notify: UserId=0 and cannot resolve from order for original_transaction_id=%s", txPayload.OriginalTransactionId)
+			}
+		}
+
 		var days int64
 		{
 			pid := strings.ToLower(txPayload.ProductId)
@@ -169,7 +199,12 @@ func (l *AppleIAPNotifyLogic) Handle(signedPayload string) error {
 			}
 		}
 		if days == 0 {
-			l.Errorw("iap notify product mapping missing", logger.Field("productId", txPayload.ProductId))
+			// Both string-parse and DB fallback failed to map the product to days.
+			// Return an error so Apple retries the notification; silently ignoring
+			// this would cause the subscriber's renewal to be lost.
+			l.Errorw("CRITICAL: iap notify product mapping missing, returning error to trigger retry",
+				logger.Field("productId", txPayload.ProductId))
+			return fmt.Errorf("iap product id %s could not be mapped to subscription days", txPayload.ProductId)
 		}
 		token := "iap:" + txPayload.OriginalTransactionId
 		sub, e := l.svcCtx.UserModel.FindOneSubscribeByToken(l.ctx, token)
@@ -216,6 +251,11 @@ func (l *AppleIAPNotifyLogic) Handle(signedPayload string) error {
 							logger.Field("product_id", txPayload.ProductId),
 						)...,
 					)
+					// Create audit order record for renewal notifications (Bug 7)
+					if err := l.createIAPRenewalAuditOrder(db, ntype, txPayload.TransactionId, candidate.UserId, candidate.SubscribeId, candidate.Token); err != nil {
+						l.Errorw("iap notify fallback create renewal order error", logger.Field("error", err.Error()))
+						// Non-fatal: subscription already updated; order creation failure is logged only
+					}
 					break
 				}
 			}
@@ -248,7 +288,52 @@ func (l *AppleIAPNotifyLogic) Handle(signedPayload string) error {
 					logger.Field("product_id", txPayload.ProductId),
 				)...,
 			)
+			// Create audit order record for renewal notifications (Bug 7)
+			if err := l.createIAPRenewalAuditOrder(db, ntype, txPayload.TransactionId, sub.UserId, sub.SubscribeId, sub.Token); err != nil {
+				l.Errorw("iap notify create renewal order error", logger.Field("error", err.Error()))
+				// Non-fatal: subscription already updated; order creation failure is logged only
+			}
 		}
 		return nil
 	})
 }
+
+// createIAPRenewalAuditOrder creates a finished renewal order record for DID_RENEW and SUBSCRIBED
+// Apple SSNS notifications so that the order table has a complete financial audit trail.
+// The operation is idempotent: if an order with the same trade_no already exists it is skipped.
+func (l *AppleIAPNotifyLogic) createIAPRenewalAuditOrder(db *gorm.DB, ntype, transactionId string, userId, subscribeId int64, subscribeToken string) error {
+	if ntype != "DID_RENEW" && ntype != "SUBSCRIBED" {
+		return nil
+	}
+	// Idempotency check
+	var count int64
+	if err := db.Model(&order.Order{}).
+		Where("trade_no = ? AND method = ?", transactionId, "apple_iap").
+		Count(&count).Error; err != nil {
+		return fmt.Errorf("check existing iap renewal order: %w", err)
+	}
+	if count > 0 {
+		return nil
+	}
+	rec := &order.Order{
+		UserId:         userId,
+		OrderNo:        tool.GenerateTradeNo(),
+		Type:           2, // OrderTypeRenewal
+		Status:         5, // OrderStatusFinished
+		Method:         "apple_iap",
+		TradeNo:        transactionId,
+		SubscribeId:    subscribeId,
+		SubscribeToken: subscribeToken,
+		Quantity:       1,
+		IsNew:          false,
+	}
+	if err := db.Model(&order.Order{}).Create(rec).Error; err != nil {
+		return fmt.Errorf("create iap renewal audit order: %w", err)
+	}
+	l.Infow("iap notify created renewal audit order",
+		logger.Field("orderNo", rec.OrderNo),
+		logger.Field("transactionId", transactionId),
+		logger.Field("userId", userId),
+	)
+	return nil
+}

internal/logic/public/portal/purchaseLogic.go

@@ -154,9 +154,12 @@ func (l *PurchaseLogic) Purchase(req *types.PortalPurchaseRequest) (resp *types.
 		}
 		content, _ := tempOrder.Marshal()
 
-		if _, err = l.svcCtx.Redis.Set(l.ctx, fmt.Sprintf(constant.TempOrderCacheKey, orderInfo.OrderNo), string(content), CloseOrderTimeMinutes*time.Minute).Result(); err != nil {
-			l.Errorw("[Purchase] Redis set error", logger.Field("error", err.Error()), logger.Field("order_no", orderInfo.OrderNo))
-			return err
+		// Persist activation context to DB so the worker can recover if Redis TTL expires.
+		orderInfo.ActivationContext = string(content)
+
+		// Write to Redis as a hot cache (best-effort; non-fatal on failure).
+		if _, redisErr := l.svcCtx.Redis.Set(l.ctx, fmt.Sprintf(constant.TempOrderCacheKey, orderInfo.OrderNo), string(content), CloseOrderTimeMinutes*time.Minute).Result(); redisErr != nil {
+			l.Infow("[Purchase] Redis set error (non-fatal, DB fallback available)", logger.Field("error", redisErr.Error()), logger.Field("order_no", orderInfo.OrderNo))
 		}
 		l.Infow("[Purchase] Guest order", logger.Field("order_no", orderInfo.OrderNo), logger.Field("identifier", req.Identifier))
 
@@ -169,7 +172,7 @@ func (l *PurchaseLogic) Purchase(req *types.PortalPurchaseRequest) (resp *types.
 			}
 		}
 
-		// save guest order
+		// save guest order (activation_context is included)
 		if err = l.svcCtx.OrderModel.Insert(l.ctx, orderInfo, tx); err != nil {
 			return err
 		}

internal/logic/public/redemption/redeemCodeLogic.go

@@ -151,34 +151,48 @@ func (l *RedeemCodeLogic) RedeemCode(req *types.RedeemCodeRequest) (resp *types.
 	}
 
 	// 创建Order记录
+	redemptionContext := struct {
+		Type             string `json:"type"`
+		RedemptionCodeId int64  `json:"redemption_code_id"`
+		UnitTime         string `json:"unit_time"`
+		Quantity         int64  `json:"quantity"`
+	}{
+		Type:             "redemption",
+		RedemptionCodeId: redemptionCode.Id,
+		UnitTime:         redemptionCode.UnitTime,
+		Quantity:         redemptionCode.Quantity,
+	}
+	activationContextJSON, _ := json.Marshal(redemptionContext)
+
 	orderInfo := &order.Order{
-		UserId:         u.Id,
-		OrderNo:        tool.GenerateTradeNo(),
-		Type:           5, // 兑换类型
-		Quantity:       redemptionCode.Quantity,
-		Price:          0, // 兑换无价格
-		Amount:         0, // 兑换无金额
-		Discount:       0,
-		GiftAmount:     0,
-		Coupon:         "",
-		CouponDiscount: 0,
-		PaymentId:      0,
-		Method:         "redemption",
-		FeeAmount:      0,
-		Commission:     0,
-		Status:         2, // 直接设置为已支付
-		SubscribeId:    redemptionCode.SubscribePlan,
-		IsNew:          isNew,
+		UserId:            u.Id,
+		OrderNo:           tool.GenerateTradeNo(),
+		Type:              5, // 兑换类型
+		Quantity:          redemptionCode.Quantity,
+		Price:             0, // 兑换无价格
+		Amount:            0, // 兑换无金额
+		Discount:          0,
+		GiftAmount:        0,
+		Coupon:            "",
+		CouponDiscount:    0,
+		PaymentId:         0,
+		Method:            "redemption",
+		FeeAmount:         0,
+		Commission:        0,
+		Status:            2, // 直接设置为已支付
+		SubscribeId:       redemptionCode.SubscribePlan,
+		IsNew:             isNew,
+		ActivationContext: string(activationContextJSON),
 	}
 
-	// 保存Order到数据库
+	// 保存Order到数据库（activation_context 同步写入，作为 Redis 的持久化兜底）
 	err = l.svcCtx.OrderModel.Insert(l.ctx, orderInfo)
 	if err != nil {
 		l.Errorw("[RedeemCode] Create order failed", logger.Field("error", err.Error()))
 		return nil, errors.Wrapf(xerr.NewErrCode(xerr.DatabaseInsertError), "create order failed")
 	}
 
-	// 缓存兑换码信息到Redis（供队列任务使用）
+	// 缓存兑换码信息到Redis（热缓存，供队列任务快速读取，非关键路径）
 	cacheKey := fmt.Sprintf("redemption_order:%s", orderInfo.OrderNo)
 	cacheData := map[string]interface{}{
 		"redemption_code_id": redemptionCode.Id,
@@ -186,16 +200,8 @@ func (l *RedeemCodeLogic) RedeemCode(req *types.RedeemCodeRequest) (resp *types.
 		"quantity":           redemptionCode.Quantity,
 	}
 	jsonData, _ := json.Marshal(cacheData)
-	err = l.svcCtx.Redis.Set(l.ctx, cacheKey, jsonData, 2*time.Hour).Err()
-	if err != nil {
-		l.Errorw("[RedeemCode] Cache redemption data failed", logger.Field("error", err.Error()))
-		// 缓存失败，删除已创建的Order避免孤儿记录
-		if delErr := l.svcCtx.OrderModel.Delete(l.ctx, orderInfo.Id); delErr != nil {
-			l.Errorw("[RedeemCode] Delete order failed after cache error",
-				logger.Field("order_id", orderInfo.Id),
-				logger.Field("error", delErr.Error()))
-		}
-		return nil, errors.Wrapf(xerr.NewErrCode(xerr.ERROR), "cache redemption data failed")
+	if redisErr := l.svcCtx.Redis.Set(l.ctx, cacheKey, jsonData, 2*time.Hour).Err(); redisErr != nil {
+		l.Infow("[RedeemCode] Cache redemption data failed (non-fatal, DB fallback available)", logger.Field("error", redisErr.Error()))
 	}
 
 	// 触发队列任务

internal/model/order/model.go

@@ -110,6 +110,10 @@ func (m *customOrderModel) UpdateOrderStatus(ctx context.Context, orderNo string
 	if err != nil {
 		return err
 	}
+	keys := m.getCacheKeys(orderInfo)
+	// Pre-delete: evict cache before the DB write so concurrent reads during the update
+	// window go to DB instead of getting a stale cached status (double-delete pattern).
+	_ = m.DelCacheCtx(ctx, keys...)
 	return m.ExecCtx(ctx, func(conn *gorm.DB) error {
 		if len(tx) > 0 {
 			conn = tx[0]
@@ -123,7 +127,7 @@ func (m *customOrderModel) UpdateOrderStatus(ctx context.Context, orderNo string
 			return nil
 		}
 		return nil
-	}, m.getCacheKeys(orderInfo)...)
+	}, keys...)
 }
 
 // FindOneDetailsByOrderNo Find order details by order number

internal/model/order/order.go

@@ -24,8 +24,9 @@ type Order struct {
 	Status         uint8     `gorm:"type:tinyint(1);not null;default:1;comment:Order Status: 1: Pending, 2: Paid, 3:Close, 4: Failed, 5:Finished;"`
 	SubscribeId    int64     `gorm:"type:bigint;not null;default:0;comment:Subscribe Id"`
 	SubscribeToken  string    `gorm:"type:varchar(255);default:null;comment:Renewal Subscribe Token"`
-	AppAccountToken string    `gorm:"type:varchar(36);default:null;comment:Apple IAP App Account Token (UUID)"`
-	IsNew           bool      `gorm:"type:tinyint(1);not null;default:0;comment:Is New Order"`
+	AppAccountToken   string    `gorm:"type:varchar(36);default:null;comment:Apple IAP App Account Token (UUID)"`
+	ActivationContext string    `gorm:"type:text;default:null;comment:Activation context JSON (guest/redemption info for DB fallback)"`
+	IsNew             bool      `gorm:"type:tinyint(1);not null;default:0;comment:Is New Order"`
 	CreatedAt      time.Time `gorm:"<-:create;comment:Create Time"`
 	UpdatedAt      time.Time `gorm:"comment:Update Time"`
 }

queue/handler/routes.go

@@ -48,4 +48,7 @@ func RegisterHandlers(mux *asynq.ServeMux, serverCtx *svc.ServiceContext) {
 	// Apple IAP 对账（第二层：5min 扫描 + 第三层：日终全量）
 	mux.Handle(types.SchedulerIAPReconcile, iapLogic.NewReconcileLogic(serverCtx))
 	mux.Handle(types.SchedulerIAPDailyReconcile, iapLogic.NewDailyReconcileLogic(serverCtx))
+
+	// Stuck order recovery
+	mux.Handle(types.SchedulerStuckOrderRecovery, orderLogic.NewStuckOrderRecoveryLogic(serverCtx))
 }

queue/logic/order/activateOrderLogic.go

@@ -7,6 +7,7 @@ import (
 	"encoding/json"
 	"errors"
 	"fmt"
+	"strings"
 	"time"
 
 	"github.com/perfect-panel/server/internal/logic/admin/group"
@@ -46,7 +47,7 @@ const (
 	OrderStatusPaid     = 2 // Order paid and ready for processing
 	OrderStatusClose    = 3 // Order closed/cancelled
 	OrderStatusFailed   = 4 // Order processing failed
-	OrderStatusClaimed  = 4 // Internal transient claim while a worker processes the order
+	OrderStatusClaimed  = 6 // Internal transient claim while a worker processes the order
 	OrderStatusFinished = 5 // Order successfully completed
 )
 
@@ -92,8 +93,12 @@ func (l *ActivateOrderLogic) ProcessTask(ctx context.Context, task *asynq.Task)
 
 	orderInfo, err := l.claimAndGetOrder(ctx, payload.OrderNo)
 	if err != nil {
-		// 如果订单不存在或状态不对，不重试
 		if errors.Is(err, ErrInvalidOrderStatus) {
+			if strings.Contains(err.Error(), "stuck in claimed") {
+				logger.WithContext(ctx).Error("[ActivateOrderLogic] 订单卡在 claimed，将重试",
+					logger.Field("order_no", payload.OrderNo))
+				return err // 返回错误触发 asynq 重试
+			}
 			logger.WithContext(ctx).Info("[ActivateOrderLogic] 订单状态不是已支付，跳过",
 				logger.Field("order_no", payload.OrderNo))
 			return nil
@@ -116,7 +121,12 @@ func (l *ActivateOrderLogic) ProcessTask(ctx context.Context, task *asynq.Task)
 	)
 
 	if err = l.processOrderByType(ctx, orderInfo, payload.IAPExpireAt); err != nil {
-		l.releaseClaim(ctx, orderInfo.OrderNo)
+		if releaseErr := l.releaseClaim(ctx, orderInfo.OrderNo); releaseErr != nil {
+			logger.WithContext(ctx).Error("[ActivateOrderLogic] releaseClaim also failed, stuck recovery will handle",
+				logger.Field("order_no", orderInfo.OrderNo),
+				logger.Field("release_error", releaseErr.Error()),
+			)
+		}
 		logger.WithContext(ctx).Error("[ActivateOrderLogic] 处理订单失败，将重试",
 			logger.Field("order_no", orderInfo.OrderNo),
 			logger.Field("order_type", orderInfo.Type),
@@ -125,7 +135,12 @@ func (l *ActivateOrderLogic) ProcessTask(ctx context.Context, task *asynq.Task)
 	}
 
 	if err = l.reconcilePostOrderSubscriptions(ctx, orderInfo); err != nil {
-		l.releaseClaim(ctx, orderInfo.OrderNo)
+		if releaseErr := l.releaseClaim(ctx, orderInfo.OrderNo); releaseErr != nil {
+			logger.WithContext(ctx).Error("[ActivateOrderLogic] releaseClaim also failed, stuck recovery will handle",
+				logger.Field("order_no", orderInfo.OrderNo),
+				logger.Field("release_error", releaseErr.Error()),
+			)
+		}
 		logger.WithContext(ctx).Error("[ActivateOrderLogic] 订单订阅兜底合并失败，将重试",
 			logger.Field("order_no", orderInfo.OrderNo),
 			logger.Field("order_type", orderInfo.Type),
@@ -176,6 +191,15 @@ func (l *ActivateOrderLogic) claimAndGetOrder(ctx context.Context, orderNo strin
 		return nil, nil
 	}
 
+	// Detect stuck claimed order — return retryable error so asynq re-tries
+	if orderInfo.Status == OrderStatusClaimed {
+		logger.WithContext(ctx).Error("Order stuck in claimed status",
+			logger.Field("order_no", orderInfo.OrderNo),
+			logger.Field("status", orderInfo.Status),
+		)
+		return nil, fmt.Errorf("order %s stuck in claimed status: %w", orderNo, ErrInvalidOrderStatus)
+	}
+
 	if orderInfo.Status != OrderStatusPaid {
 		logger.WithContext(ctx).Error("Order status error",
 			logger.Field("order_no", orderInfo.OrderNo),
@@ -201,7 +225,7 @@ func (l *ActivateOrderLogic) claimAndGetOrder(ctx context.Context, orderNo strin
 	return &orderInfo, nil
 }
 
-func (l *ActivateOrderLogic) releaseClaim(ctx context.Context, orderNo string) {
+func (l *ActivateOrderLogic) releaseClaim(ctx context.Context, orderNo string) error {
 	if err := l.svc.DB.WithContext(ctx).
 		Model(&order.Order{}).
 		Where("order_no = ? AND status = ?", orderNo, OrderStatusClaimed).
@@ -210,7 +234,9 @@ func (l *ActivateOrderLogic) releaseClaim(ctx context.Context, orderNo string) {
 			logger.Field("error", err.Error()),
 			logger.Field("order_no", orderNo),
 		)
+		return fmt.Errorf("release claim failed for order %s: %w", orderNo, err)
 	}
+	return nil
 }
 
 // processOrderByType routes order processing based on the order type
@@ -563,14 +589,27 @@ func (l *ActivateOrderLogic) finalizeCouponAndOrder(ctx context.Context, orderIn
 		}
 	}
 
-	// Update order status using state-guarded UpdateOrderStatus to prevent double finalization
-	if err := l.svc.OrderModel.UpdateOrderStatus(ctx, orderInfo.OrderNo, OrderStatusFinished); err != nil {
-		logger.WithContext(ctx).Error("Update order status failed",
+	// UpdateOrderStatus uses WHERE status < target, which blocks claimed(6)→finished(5).
+	// Use a direct update matching the exact claimed status, then update the full record
+	// via the model layer to properly invalidate the cache.
+	result := l.svc.DB.WithContext(ctx).
+		Model(&order.Order{}).
+		Where("order_no = ? AND status = ?", orderInfo.OrderNo, OrderStatusClaimed).
+		Update("status", OrderStatusFinished)
+	if result.Error != nil {
+		logger.WithContext(ctx).Error("Update order status from claimed to finished failed",
+			logger.Field("error", result.Error.Error()),
+			logger.Field("order_no", orderInfo.OrderNo),
+		)
+	}
+	// Invalidate order cache regardless of whether the DB update succeeded
+	orderInfo.Status = OrderStatusFinished
+	if err := l.svc.OrderModel.Update(ctx, orderInfo); err != nil {
+		logger.WithContext(ctx).Error("Update order cache after finalization failed",
 			logger.Field("error", err.Error()),
 			logger.Field("order_no", orderInfo.OrderNo),
 		)
 	}
-	orderInfo.Status = OrderStatusFinished
 	commonLogic.SubscriptionTraceInfo(logger.WithContext(ctx), commonLogic.SubscriptionTraceFlowOrder, "order_status_finished",
 		"[SubscriptionFlow] order status updated to finished",
 		commonLogic.OrderTraceFields(orderInfo)...,
@@ -597,7 +636,7 @@ func (l *ActivateOrderLogic) NewPurchase(ctx context.Context, orderInfo *order.O
 		return err
 	}
 
-	if err = validateNewUserOnlyEligibilityAtActivation(ctx, l.svc.DB, orderInfo, sub); err != nil {
+	if err = validateNewUserOnlyEligibilityAtActivation(ctx, l.svc.DB, l.svc.Redis, orderInfo, sub); err != nil {
 		return err
 	}
 
@@ -675,18 +714,23 @@ func (l *ActivateOrderLogic) NewPurchase(ctx context.Context, orderInfo *order.O
 	// 兜底：创建新订阅前，查找用户是否已有同套餐的订阅记录（含过期/赠送），
 	// 有则复用旧记录续期，避免出现重复订阅。
 	// 需要同时检查 UserId 和 SubscriptionUserId，因为家庭组绑定前后 owner 可能不同。
+	// 使用 SELECT ... FOR UPDATE 防止并发下多个 worker 同时选中同一订阅并创建重复记录。
 	if userSub == nil {
 		candidateUserIds := []int64{orderInfo.UserId}
 		if orderInfo.SubscriptionUserId > 0 && orderInfo.SubscriptionUserId != orderInfo.UserId {
 			candidateUserIds = append(candidateUserIds, orderInfo.SubscriptionUserId)
 		}
 		var existingSub user.Subscribe
-		if findErr := l.svc.DB.Model(&user.Subscribe{}).
-			Where("user_id IN ? AND token != ''", candidateUserIds).
-			Order("expire_time DESC").
-			Order("updated_at DESC").
-			Order("id DESC").
-			First(&existingSub).Error; findErr == nil {
+		findErr := l.svc.DB.WithContext(ctx).Transaction(func(tx *gorm.DB) error {
+			return tx.Clauses(clause.Locking{Strength: "UPDATE"}).
+				Model(&user.Subscribe{}).
+				Where("user_id IN ? AND token != ''", candidateUserIds).
+				Order("expire_time DESC").
+				Order("updated_at DESC").
+				Order("id DESC").
+				First(&existingSub).Error
+		})
+		if findErr == nil {
 			// 家庭组场景：订阅 owner 可能变更（如成员注册的试用 → 被家主收归），
 			// 续期前把 user_id 校正为当前订单的 SubscriptionUserId
 			effectiveOwner := orderInfo.UserId
@@ -824,26 +868,52 @@ func (l *ActivateOrderLogic) createGuestUser(ctx context.Context, orderInfo *ord
 	return userInfo, nil
 }
 
-// getTempOrderInfo retrieves temporary order information from Redis cache
+// getTempOrderInfo retrieves temporary order information from Redis cache with DB fallback.
 func (l *ActivateOrderLogic) getTempOrderInfo(ctx context.Context, orderNo string) (*constant.TemporaryOrderInfo, error) {
 	cacheKey := fmt.Sprintf(constant.TempOrderCacheKey, orderNo)
 	data, err := l.svc.Redis.Get(ctx, cacheKey).Result()
-	if err != nil {
-		logger.WithContext(ctx).Error("Get temp order cache failed",
-			logger.Field("error", err.Error()),
-			logger.Field("cache_key", cacheKey),
+	if err == nil {
+		var tempOrder constant.TemporaryOrderInfo
+		if unmarshalErr := tempOrder.Unmarshal([]byte(data)); unmarshalErr != nil {
+			logger.WithContext(ctx).Error("Unmarshal temp order cache failed",
+				logger.Field("error", unmarshalErr.Error()),
+				logger.Field("cache_key", cacheKey),
+				logger.Field("data", data),
+			)
+			return nil, unmarshalErr
+		}
+		return &tempOrder, nil
+	}
+
+	// Redis miss — fall back to DB activation_context field.
+	logger.WithContext(ctx).Infow("Redis cache miss for temp order, falling back to DB",
+		logger.Field("cache_key", cacheKey),
+		logger.Field("error", err.Error()),
+	)
+
+	orderInfo, dbErr := l.svc.OrderModel.FindOneByOrderNo(ctx, orderNo)
+	if dbErr != nil {
+		logger.WithContext(ctx).Error("DB fallback for temp order failed",
+			logger.Field("order_no", orderNo),
+			logger.Field("error", dbErr.Error()),
 		)
-		return nil, err
+		return nil, dbErr
+	}
+
+	if orderInfo.ActivationContext == "" {
+		logger.WithContext(ctx).Error("CRITICAL: activation_context missing in DB and Redis expired; cannot recover guest order",
+			logger.Field("order_no", orderNo),
+		)
+		return nil, errors.New("activation context not found: Redis expired and no DB fallback available")
 	}
 
 	var tempOrder constant.TemporaryOrderInfo
-	if err = tempOrder.Unmarshal([]byte(data)); err != nil {
-		logger.WithContext(ctx).Error("Unmarshal temp order cache failed",
-			logger.Field("error", err.Error()),
-			logger.Field("cache_key", cacheKey),
-			logger.Field("data", data),
+	if unmarshalErr := tempOrder.Unmarshal([]byte(orderInfo.ActivationContext)); unmarshalErr != nil {
+		logger.WithContext(ctx).Error("Unmarshal DB activation_context failed",
+			logger.Field("order_no", orderNo),
+			logger.Field("error", unmarshalErr.Error()),
 		)
-		return nil, err
+		return nil, unmarshalErr
 	}
 
 	return &tempOrder, nil
@@ -1447,7 +1517,40 @@ func (l *ActivateOrderLogic) resolveRenewalActivationSubscription(ctx context.Co
 	}
 	userSub, err := l.getUserSubscription(ctx, orderInfo.SubscribeToken)
 	if err != nil {
-		return nil, err
+		// Fallback: token may have been lost (subscription deleted or owner changed).
+		// Locate the most recent subscription by user_id + subscribe_id with FOR UPDATE
+		// to prevent a concurrent worker from renewing the same record twice.
+		targetUserID := orderInfo.UserId
+		if orderInfo.SubscriptionUserId > 0 {
+			targetUserID = orderInfo.SubscriptionUserId
+		}
+		var fallbackSub user.Subscribe
+		txErr := l.svc.DB.WithContext(ctx).Transaction(func(tx *gorm.DB) error {
+			return tx.Clauses(clause.Locking{Strength: "UPDATE"}).
+				Model(&user.Subscribe{}).
+				Where("user_id = ? AND subscribe_id = ?", targetUserID, orderInfo.SubscribeId).
+				Where("status IN ?", []int64{0, 1, 2, 3}).
+				Order("expire_time DESC").
+				Order("updated_at DESC").
+				Order("id DESC").
+				First(&fallbackSub).Error
+		})
+		if txErr != nil {
+			logger.WithContext(ctx).Error("CRITICAL: Renewal activation token and fallback lookup both failed",
+				logger.Field("token_error", err.Error()),
+				logger.Field("fallback_error", txErr.Error()),
+				logger.Field("order_no", orderInfo.OrderNo),
+				logger.Field("target_user_id", targetUserID),
+				logger.Field("subscribe_id", orderInfo.SubscribeId),
+			)
+			return nil, fmt.Errorf("renewal activation subscription not found by token or user_id+subscribe_id for order %s: %w", orderInfo.OrderNo, err)
+		}
+		logger.WithContext(ctx).Info("Renewal token lookup failed; found subscription via fallback user_id+subscribe_id",
+			logger.Field("order_no", orderInfo.OrderNo),
+			logger.Field("fallback_subscribe_id", fallbackSub.Id),
+			logger.Field("target_user_id", targetUserID),
+		)
+		userSub = &fallbackSub
 	}
 	if orderInfo.UserId <= 0 {
 		return userSub, nil
@@ -1743,25 +1846,50 @@ func (l *ActivateOrderLogic) RedemptionActivate(ctx context.Context, orderInfo *
 		return err
 	}
 
-	// 3. 从Redis获取兑换码信息
-	cacheKey := fmt.Sprintf("redemption_order:%s", orderInfo.OrderNo)
-	data, err := l.svc.Redis.Get(ctx, cacheKey).Result()
-	if err != nil {
-		logger.WithContext(ctx).Error("Get redemption cache failed",
-			logger.Field("error", err.Error()),
-			logger.Field("cache_key", cacheKey),
-		)
-		return err
-	}
-
+	// 3. 从Redis获取兑换码信息，Redis缺失时回查DB
 	var redemptionData struct {
 		RedemptionCodeId int64  `json:"redemption_code_id"`
 		UnitTime         string `json:"unit_time"`
 		Quantity         int64  `json:"quantity"`
 	}
-	if err = json.Unmarshal([]byte(data), &redemptionData); err != nil {
-		logger.WithContext(ctx).Error("Unmarshal redemption cache failed", logger.Field("error", err.Error()))
-		return err
+
+	cacheKey := fmt.Sprintf("redemption_order:%s", orderInfo.OrderNo)
+	data, redisErr := l.svc.Redis.Get(ctx, cacheKey).Result()
+	if redisErr == nil {
+		if err = json.Unmarshal([]byte(data), &redemptionData); err != nil {
+			logger.WithContext(ctx).Error("Unmarshal redemption cache failed", logger.Field("error", err.Error()))
+			return err
+		}
+	} else {
+		// Redis miss — fall back to DB activation_context field.
+		logger.WithContext(ctx).Infow("Redis cache miss for redemption order, falling back to DB",
+			logger.Field("cache_key", cacheKey),
+			logger.Field("error", redisErr.Error()),
+		)
+
+		if orderInfo.ActivationContext == "" {
+			logger.WithContext(ctx).Error("CRITICAL: activation_context missing in DB and Redis expired; cannot recover redemption order",
+				logger.Field("order_no", orderInfo.OrderNo),
+			)
+			return errors.New("activation context not found: Redis expired and no DB fallback available")
+		}
+
+		var fullContext struct {
+			Type             string `json:"type"`
+			RedemptionCodeId int64  `json:"redemption_code_id"`
+			UnitTime         string `json:"unit_time"`
+			Quantity         int64  `json:"quantity"`
+		}
+		if err = json.Unmarshal([]byte(orderInfo.ActivationContext), &fullContext); err != nil {
+			logger.WithContext(ctx).Error("Unmarshal DB activation_context failed",
+				logger.Field("order_no", orderInfo.OrderNo),
+				logger.Field("error", err.Error()),
+			)
+			return err
+		}
+		redemptionData.RedemptionCodeId = fullContext.RedemptionCodeId
+		redemptionData.UnitTime = fullContext.UnitTime
+		redemptionData.Quantity = fullContext.Quantity
 	}
 
 	// 4. 幂等性检查：查询是否已有兑换记录

queue/logic/order/newUserEligibility.go

@@ -10,12 +10,14 @@ import (
 	"github.com/perfect-panel/server/internal/model/order"
 	"github.com/perfect-panel/server/internal/model/subscribe"
 	internaltypes "github.com/perfect-panel/server/internal/types"
+	"github.com/redis/go-redis/v9"
 	"gorm.io/gorm"
 )
 
 func validateNewUserOnlyEligibilityAtActivation(
 	ctx context.Context,
 	db *gorm.DB,
+	rdb *redis.Client,
 	orderInfo *order.Order,
 	sub *subscribe.Subscribe,
 ) error {
@@ -31,6 +33,20 @@ func validateNewUserOnlyEligibilityAtActivation(
 		return nil
 	}
 
+	// Acquire a per-user distributed lock so concurrent new-user-only activations
+	// for the same account are serialised. Without this, two workers can both read
+	// historyCount=0 and both pass the check before either has written the order.
+	lockKey := fmt.Sprintf("new_user_only_activate:%d", orderInfo.UserId)
+	const lockTTL = 30 * time.Second
+	acquired, lockErr := rdb.SetNX(ctx, lockKey, orderInfo.OrderNo, lockTTL).Result()
+	if lockErr != nil {
+		return fmt.Errorf("new user only: acquire lock error: %w", lockErr)
+	}
+	if !acquired {
+		return fmt.Errorf("new user only: another activation is in progress for user %d", orderInfo.UserId)
+	}
+	defer rdb.Del(ctx, lockKey)
+
 	eligibility, err := commonLogic.ResolveNewUserEligibility(ctx, db, orderInfo.UserId)
 	if err != nil {
 		return err

queue/logic/order/stuckOrderRecoveryLogic.go

@@ -0,0 +1,104 @@
+package orderLogic
+
+import (
+	"context"
+	"encoding/json"
+	"time"
+
+	"github.com/hibiken/asynq"
+	"github.com/perfect-panel/server/internal/model/order"
+	"github.com/perfect-panel/server/internal/svc"
+	"github.com/perfect-panel/server/pkg/logger"
+	queueTypes "github.com/perfect-panel/server/queue/types"
+)
+
+// StuckOrderRecoveryLogic scans orders stuck in claimed status and re-queues them for processing.
+type StuckOrderRecoveryLogic struct {
+	svc *svc.ServiceContext
+}
+
+func NewStuckOrderRecoveryLogic(svc *svc.ServiceContext) *StuckOrderRecoveryLogic {
+	return &StuckOrderRecoveryLogic{svc: svc}
+}
+
+// ProcessTask scans for orders stuck in claimed status for over 10 minutes,
+// resets them to paid, and re-enqueues the activate task so they are retried
+// independently of asynq's original retry counter (which may be exhausted).
+func (l *StuckOrderRecoveryLogic) ProcessTask(ctx context.Context, _ *asynq.Task) error {
+	cutoff := time.Now().Add(-10 * time.Minute)
+
+	var stuckOrders []order.Order
+	if err := l.svc.DB.WithContext(ctx).
+		Model(&order.Order{}).
+		Where("status = ? AND updated_at < ?", OrderStatusClaimed, cutoff).
+		Find(&stuckOrders).Error; err != nil {
+		logger.WithContext(ctx).Error("[StuckOrderRecovery] Failed to query stuck orders",
+			logger.Field("error", err.Error()),
+		)
+		return err
+	}
+
+	if len(stuckOrders) == 0 {
+		return nil
+	}
+
+	orderNos := make([]string, 0, len(stuckOrders))
+	for i := range stuckOrders {
+		orderNos = append(orderNos, stuckOrders[i].OrderNo)
+	}
+	logger.WithContext(ctx).Error("[StuckOrderRecovery] Found stuck claimed orders, recovering",
+		logger.Field("count", len(stuckOrders)),
+		logger.Field("order_nos", orderNos),
+	)
+
+	for i := range stuckOrders {
+		o := &stuckOrders[i]
+
+		result := l.svc.DB.WithContext(ctx).
+			Model(&order.Order{}).
+			Where("order_no = ? AND status = ?", o.OrderNo, OrderStatusClaimed).
+			Update("status", OrderStatusPaid)
+		if result.Error != nil {
+			logger.WithContext(ctx).Error("[StuckOrderRecovery] Failed to reset order status",
+				logger.Field("order_no", o.OrderNo),
+				logger.Field("error", result.Error.Error()),
+			)
+			continue
+		}
+		if result.RowsAffected == 0 {
+			// Another process already handled this order
+			continue
+		}
+
+		// Invalidate order cache
+		o.Status = OrderStatusPaid
+		if err := l.svc.OrderModel.Update(ctx, o); err != nil {
+			logger.WithContext(ctx).Error("[StuckOrderRecovery] Failed to update order cache",
+				logger.Field("order_no", o.OrderNo),
+				logger.Field("error", err.Error()),
+			)
+		}
+
+		// Re-enqueue activate task so the order gets processed regardless of asynq retry state
+		payload, err := json.Marshal(queueTypes.ForthwithActivateOrderPayload{OrderNo: o.OrderNo})
+		if err != nil {
+			logger.WithContext(ctx).Error("[StuckOrderRecovery] Failed to marshal task payload",
+				logger.Field("order_no", o.OrderNo),
+				logger.Field("error", err.Error()),
+			)
+			continue
+		}
+		if _, err = l.svc.Queue.EnqueueContext(ctx, asynq.NewTask(queueTypes.ForthwithActivateOrder, payload)); err != nil {
+			logger.WithContext(ctx).Error("[StuckOrderRecovery] Failed to re-enqueue activate task",
+				logger.Field("order_no", o.OrderNo),
+				logger.Field("error", err.Error()),
+			)
+		} else {
+			logger.WithContext(ctx).Info("[StuckOrderRecovery] Re-enqueued activate task",
+				logger.Field("order_no", o.OrderNo),
+			)
+		}
+	}
+
+	return nil
+}

queue/types/scheduler.go

@@ -5,6 +5,7 @@ const (
 	SchedulerTotalServerData    = "scheduler:total:server"
 	SchedulerResetTraffic       = "scheduler:reset:traffic"
 	SchedulerTrafficStat        = "scheduler:traffic:stat"
-	SchedulerIAPReconcile       = "scheduler:iap:reconcile"       // 第二层：每 5 分钟扫描待支付 IAP 订单
-	SchedulerIAPDailyReconcile  = "scheduler:iap:daily:reconcile" // 第三层：日终全量对账
+	SchedulerIAPReconcile       = "scheduler:iap:reconcile"          // 第二层：每 5 分钟扫描待支付 IAP 订单
+	SchedulerIAPDailyReconcile  = "scheduler:iap:daily:reconcile"   // 第三层：日终全量对账
+	SchedulerStuckOrderRecovery = "scheduler:stuck:order:recovery"  // 扫描并恢复超时 claimed 订单
 )

scheduler/scheduler.go

@@ -64,6 +64,12 @@ func (m *Service) Start() {
 		logger.Errorf("register iap daily reconcile task failed: %s", err.Error())
 	}
 
+	// schedule stuck order recovery: every 10 minutes
+	stuckOrderTask := asynq.NewTask(types.SchedulerStuckOrderRecovery, nil)
+	if _, err := m.server.Register("@every 10m", stuckOrderTask, asynq.MaxRetry(1)); err != nil {
+		logger.Errorf("register stuck order recovery task failed: %s", err.Error())
+	}
+
 	if err := m.server.Run(); err != nil {
 		logger.Errorf("run scheduler failed: %s", err.Error())
 	}