feat(phase2e-2): 落地会议生命周期状态机 + Router 幂等双层防御（Task 8）

核心交付：
- 新建 MeetingLifecycleService（6 钩子 + sync.Map 本地 timer + Redis key 双保险 + RescheduleFromRedis）
- 新建 MeetingCleanupTask（启动重建 timer + 每 N 秒扫 host_grace/empty_ttl 兜底 + 4h stale active 回收）
- MediaOrchestrator 新增 ResolveRouterID；HTTPMediaOrchestrator.CreateRouter 入口 sync.Map 幂等防御
- 业务层 JoinRoom 移除 CreateRouter 调用改走 CancelEmptyTTL + ResolveRouterID；LeaveRoom 空房分支改调 OnAllMembersLeft 不再立即销毁
- MeetingSignalService 新增 OnWSDisconnect 实现 ws.MeetingDisconnectHook；OnRoomJoin 追加 host 重连钩子
- ws.handler 定义 MeetingDisconnectHook 接口 + SetMeetingDisconnectHook，解耦 ws→meeting 反向依赖
- config 新增 MeetingConfig{HostGrace=120, EmptyRoomTTL=300, CleanupInterval=30, StaleRoomHours=4}

关键设计决策：
- Redis key TTL = 业务时长 + max(CleanupIntervalSeconds*2, 30s) buffer：避免本地 timer 与
  Redis 自动过期同步到期导致 DEL 返回 0 被误判为"已被其他路径处理"而跳过业务逻辑
- Router 幂等双层防御（决策 q2_router_dedup=a2_both）：业务层不重复调 + HTTP 层 sync.Map 命中直接返回
- 普通成员 WS 断开仅清 media 资源不动 participant 表（决策 q1_nonhost_disconnect=a1_keep_current）

E2E 验证：docs/verify/meeting_t8_verify.mjs PASS=20 FAIL=0，覆盖 5 场景：
- S1 host 宽限期过期自动转让（meeting.host.changed + DB host_id 更新）
- S2 宽限期内重连保留身份
- S3 empty_ttl 期内新成员加入复活房间
- S4 empty_ttl 过期 → 房间 Ended + 新 join 被拒
- S5 CreateRoom +1 Router / JoinRoom 不再创建新 Router（通过 media-server /internal/info stats.routers 断言）

media-server：/internal/info 响应追加 stats.routers + routers[] 供 E2E 断言 Router 幂等

文档同步：
- docs/progress/CURRENT_STATUS.md 头部 + 新增 Task 8 交付条目
- docs/plans/2026-04-21-phase2e-2-implementation.plan.md Task 8 标记完成 + 实际产出/决策/验证
- docs/api/frontend/meeting.md 补充 host.changed.auto_reason / room.ended.reason=system_error / 空房 TTL 复活语义 + Task 8 验证记录
- docs/architecture/system-architecture.md meeting 模块职责补充"会议生命周期状态机"
- .cursor/rules/project-context.mdc 追加 Task 8 条目并更新 Phase 2e-2 进度（Task 0-8 ✅）

Made-with: Cursor

backend/go-service/app/meeting/dao/meeting_room_dao.go

@@ -160,6 +160,27 @@ func (d *MeetingRoomDAO) ListByHost(ctx context.Context, hostID int64, status in
 	return rooms, total, err
 }
 
+// ListStaleActive 返回 status != Ended 且 started_at（或 created_at 若未开始）早于 hoursAgo 小时的房间 ID
+// 用于定时任务兜底清理"创建 / Active 但长期无任何成员活动的僵尸房间"（设计 §11.4 的 4 小时兜底）
+// 排序：最早 started_at 在前，优先清理最老
+func (d *MeetingRoomDAO) ListStaleActive(ctx context.Context, hoursAgo int, limit int) ([]int64, error) {
+	funcName := "dao.meeting_room_dao.ListStaleActive"
+	cutoff := time.Now().Add(-time.Duration(hoursAgo) * time.Hour)
+
+	var ids []int64
+	err := d.db.WithContext(ctx).
+		Model(&model.MeetingRoom{}).
+		Where("status != ? AND COALESCE(started_at, created_at) < ?",
+			constants.MeetingStatusEnded, cutoff).
+		Order("COALESCE(started_at, created_at) ASC").
+		Limit(limit).
+		Pluck("id", &ids).Error
+	if err != nil {
+		logs.Error(ctx, funcName, "查询 stale 活跃房间失败", zap.Error(err))
+	}
+	return ids, err
+}
+
 // ListExpiredForCleanup 返回已结束且超过指定小时数的房间 ID 列表
 // 供定时任务清理 meeting_chats 使用；room 本身保留归档不删
 // 返回 limit 上限用于分批处理，避免一次捞太多

backend/go-service/app/meeting/provider.go

@@ -4,6 +4,7 @@ import (
 	"github.com/echochat/backend/app/meeting/controller"
 	"github.com/echochat/backend/app/meeting/dao"
 	"github.com/echochat/backend/app/meeting/service"
+	"github.com/echochat/backend/app/meeting/task"
 	"github.com/google/wire"
 )
 
@@ -20,10 +21,12 @@ var MeetingSet = wire.NewSet(
 	dao.NewMeetingParticipantDAO,
 	dao.NewMeetingChatDAO,
 	service.NewMeetingBroadcaster,
+	service.NewMeetingLifecycleService, // Task 8 新增：会议生命周期状态机
 	service.NewMeetingService,
 	service.NewMeetingSignalService,
 	controller.NewMeetingController,
 	controller.NewMeetingWSHandler,
+	task.NewMeetingCleanupTask, // Task 8 新增：生命周期兜底定时任务
 
 	// MediaOrchestrator 使用真实 HTTP 实现（Task 7 落地）
 	// 注入 *config.Config，通过 cfg.MediaServer 获取 base_url / internal_token / timeout 等配置

backend/go-service/app/meeting/service/http_media_orchestrator.go

@@ -85,9 +85,21 @@ func NewHTTPMediaOrchestrator(cfg *config.Config) *HTTPMediaOrchestrator {
 
 // CreateRouter 调用 POST /internal/v1/routers
 // 成功后在本地缓存 roomCode → routerID 映射供 CloseRouter 使用
+// Task 8 幂等防御（双层，与业务层 JoinRoom 不再主动调用形成互补）：
+//   - 命中缓存直接返回已缓存 routerID，不发起 HTTP 请求
+//   - 仅在缓存缺失时真正调 Node；避免业务误多次调用产生 Node 侧重复 Router
 func (h *HTTPMediaOrchestrator) CreateRouter(ctx context.Context, roomCode string) (string, error) {
 	funcName := "service.http_media_orchestrator.CreateRouter"
 
+	if cached, ok := h.roomRouterIDs.Load(roomCode); ok {
+		if rid, _ := cached.(string); rid != "" {
+			logs.Debug(ctx, funcName, "命中本地 Router 缓存，跳过 Node 调用",
+				zap.String("room_code", roomCode),
+				zap.String("router_id", rid))
+			return rid, nil
+		}
+	}
+
 	reqBody := map[string]any{"roomCode": roomCode}
 	var resp struct {
 		RouterID        string          `json:"routerId"`
@@ -104,8 +116,17 @@ func (h *HTTPMediaOrchestrator) CreateRouter(ctx context.Context, roomCode strin
 		return "", err
 	}
 
-	// 记忆映射：一个 roomCode 仅对应一个 Router；若此前存在旧 Router ID（极少见），以新值覆盖
-	h.roomRouterIDs.Store(roomCode, resp.RouterID)
+	// LoadOrStore 防止并发首次创建时重复覆盖：若他人已写入先到的值就用已存在的
+	actual, loaded := h.roomRouterIDs.LoadOrStore(roomCode, resp.RouterID)
+	if loaded {
+		if existing, _ := actual.(string); existing != "" && existing != resp.RouterID {
+			logs.Warn(ctx, funcName, "并发创建 Router 出现不一致，保留先到值",
+				zap.String("room_code", roomCode),
+				zap.String("existing", existing),
+				zap.String("new", resp.RouterID))
+			return existing, nil
+		}
+	}
 
 	logs.Info(ctx, funcName, "Node Router 创建成功",
 		zap.String("room_code", roomCode),
@@ -113,6 +134,21 @@ func (h *HTTPMediaOrchestrator) CreateRouter(ctx context.Context, roomCode strin
 	return resp.RouterID, nil
 }
 
+// ResolveRouterID 从本地缓存读取 roomCode 对应的 routerID（不触发 HTTP）
+// 返回 (id, true) 表示命中；返回 (_, false) 表示缓存缺失（通常意味着该房间尚未创建 Router）
+// 供 MeetingService.JoinRoom 等需复用已有 Router 的场景使用，避免重复调 Node
+func (h *HTTPMediaOrchestrator) ResolveRouterID(roomCode string) (string, bool) {
+	v, ok := h.roomRouterIDs.Load(roomCode)
+	if !ok {
+		return "", false
+	}
+	rid, _ := v.(string)
+	if rid == "" {
+		return "", false
+	}
+	return rid, true
+}
+
 // CloseRouter 调用 DELETE /internal/v1/routers/:routerId
 // 入参是 roomCode（符合设计 §6.6 NodeClient 契约），内部反查本地缓存拿到 routerID
 // 若本地缓存不存在映射（可能是 go-service 重启后状态丢失），直接返回 nil：

backend/go-service/app/meeting/service/interfaces.go

@@ -86,9 +86,14 @@ type CreateConsumerReq struct {
 //   - 所有方法均幂等：重复调用不报错，符合 WS 信令重试语义
 type MediaOrchestrator interface {
 	// CreateRouter 为会议房间创建 mediasoup Router
+	// Task 8 起实现需自带 sync.Map 幂等缓存：roomCode 已有 Router 时直接返回缓存值
 	CreateRouter(ctx context.Context, roomCode string) (routerID string, err error)
 	// CloseRouter 关闭房间对应 Router 及其下所有资源（幂等）
 	CloseRouter(ctx context.Context, roomCode string) error
+	// ResolveRouterID 从本地缓存查询 roomCode 对应的 routerID（不触发 HTTP 调用）
+	// Task 8 引入：用于 JoinRoom 等"仅需复用"的场景，避免重复调 CreateRouter
+	// 返回 (id, true) 命中；(_, false) 缺失（通常说明房间未创建 Router 或服务重启后未 reschedule）
+	ResolveRouterID(roomCode string) (string, bool)
 
 	// CreateTransport 为用户创建 send/recv WebRTC Transport
 	CreateTransport(ctx context.Context, req *CreateTransportReq) (*TransportInfo, error)
@@ -127,6 +132,11 @@ func (n *NoopMediaOrchestrator) CloseRouter(_ context.Context, _ string) error {
 	return nil
 }
 
+// ResolveRouterID 占位实现：始终命中，返回 noop-router-{code}
+func (n *NoopMediaOrchestrator) ResolveRouterID(roomCode string) (string, bool) {
+	return "noop-router-" + roomCode, true
+}
+
 // CreateTransport 占位：返回以 "noop-transport-" 为前缀的伪造 ID，带最小合法 JSON 结构
 func (n *NoopMediaOrchestrator) CreateTransport(_ context.Context, req *CreateTransportReq) (*TransportInfo, error) {
 	return &TransportInfo{

backend/go-service/app/meeting/service/meeting_lifecycle_service.go

@@ -0,0 +1,475 @@
+// Package service 提供 meeting 模块的业务逻辑
+package service
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"sync"
+	"time"
+
+	"github.com/echochat/backend/app/constants"
+	"github.com/echochat/backend/app/meeting/dao"
+	"github.com/echochat/backend/config"
+	"github.com/echochat/backend/pkg/logs"
+	"github.com/redis/go-redis/v9"
+	"go.uber.org/zap"
+)
+
+// Redis key 前缀（设计文档 §5.4 生命周期 keys；Task 8 落地）
+const (
+	redisKeyHostGracePrefix = "echo:meeting:host_grace:"
+	redisKeyEmptyTTLPrefix  = "echo:meeting:empty_ttl:"
+)
+
+// hostGracePayload host 宽限期 Redis value 结构
+type hostGracePayload struct {
+	HostID     int64 `json:"host_id"`
+	StartedAt  int64 `json:"started_at"`  // 秒级 unix
+	GraceUntil int64 `json:"grace_until"` // 秒级 unix
+}
+
+// MeetingLifecycleService 会议生命周期状态机
+// 负责设计 §6.5 的 5 类状态跃迁副作用（host 宽限期、自动转让、空房 TTL、TTL 过期销毁）
+// 采用 time.AfterFunc 本地 timer + Redis key TTL 双保险：
+//   - 本地 timer 负责低延迟触发；服务重启由 RescheduleFromRedis 按 Redis 剩余 PTTL 重新装载
+//   - 后台定时任务 MeetingCleanupTask 每 N 秒扫描 Redis 兜底防御（timer 丢失 / 多实例部署重复触发场景）
+//
+// 并发保护：HandleHostGraceExpired / HandleEmptyRoomExpired 入口先 DEL Redis key
+// 通过 DEL 返回值确认唯一处理权（0 = 已被其他 goroutine/节点清走，直接 return 避免重复操作）
+type MeetingLifecycleService struct {
+	roomDAO        *dao.MeetingRoomDAO
+	participantDAO *dao.MeetingParticipantDAO
+	db             dbExecutor
+	redis          *redis.Client
+	broadcaster    *MeetingBroadcaster
+	media          MediaOrchestrator
+	cfg            config.MeetingConfig
+
+	graceTimers    sync.Map // roomCode -> *time.Timer（host 宽限期）
+	emptyTTLTimers sync.Map // roomCode -> *time.Timer（空房 TTL）
+}
+
+// dbExecutor MeetingLifecycleService 对事务 DB 的最小依赖抽象
+// 仅用于 HandleEmptyRoomExpired 的 MarkEnded 调用链路，避免强绑 *gorm.DB
+type dbExecutor interface{}
+
+// NewMeetingLifecycleService 构造生命周期服务
+// media 由上游通过 wire 注入 HTTPMediaOrchestrator
+func NewMeetingLifecycleService(
+	roomDAO *dao.MeetingRoomDAO,
+	participantDAO *dao.MeetingParticipantDAO,
+	redis *redis.Client,
+	broadcaster *MeetingBroadcaster,
+	media MediaOrchestrator,
+	cfg *config.Config,
+) *MeetingLifecycleService {
+	meetingCfg := cfg.Meeting
+	// 零值兜底：防止 yaml 未配置时走 0 秒 TTL
+	if meetingCfg.HostGraceSeconds <= 0 {
+		meetingCfg.HostGraceSeconds = constants.MeetingHostGraceSeconds
+	}
+	if meetingCfg.EmptyRoomTTLSeconds <= 0 {
+		meetingCfg.EmptyRoomTTLSeconds = constants.MeetingEmptyRoomTTLSeconds
+	}
+	if meetingCfg.CleanupIntervalSeconds <= 0 {
+		meetingCfg.CleanupIntervalSeconds = 30
+	}
+	if meetingCfg.StaleRoomHours <= 0 {
+		meetingCfg.StaleRoomHours = 4
+	}
+	return &MeetingLifecycleService{
+		roomDAO:        roomDAO,
+		participantDAO: participantDAO,
+		redis:          redis,
+		broadcaster:    broadcaster,
+		media:          media,
+		cfg:            meetingCfg,
+	}
+}
+
+// Config 返回生效的生命周期配置（供定时任务读取扫描周期等）
+func (s *MeetingLifecycleService) Config() config.MeetingConfig {
+	return s.cfg
+}
+
+// HostGraceKey 返回 host 宽限期 Redis key
+func HostGraceKey(roomCode string) string {
+	return redisKeyHostGracePrefix + roomCode
+}
+
+// EmptyTTLKey 返回空房 TTL Redis key
+func EmptyTTLKey(roomCode string) string {
+	return redisKeyEmptyTTLPrefix + roomCode
+}
+
+// OnHostDisconnect host WS 断线钩子
+// 写入 host_grace key（TTL HostGraceSeconds）并启动本地 AfterFunc
+// 若已有同 roomCode 的 grace 记录（重复触发），幂等刷新 TTL + 替换 timer
+func (s *MeetingLifecycleService) OnHostDisconnect(ctx context.Context, roomCode string, hostID int64) {
+	funcName := "service.meeting_lifecycle_service.OnHostDisconnect"
+
+	now := time.Now()
+	payload := hostGracePayload{
+		HostID:     hostID,
+		StartedAt:  now.Unix(),
+		GraceUntil: now.Add(time.Duration(s.cfg.HostGraceSeconds) * time.Second).Unix(),
+	}
+	raw, _ := json.Marshal(payload)
+
+	graceDur := time.Duration(s.cfg.HostGraceSeconds) * time.Second
+	// Redis TTL 比本地 timer 多留一段 buffer（max(cleanup*2, 30s)）
+	// 保证本地 timer 先触发 → DEL 命中；即便本地 timer 丢失，cleanup 也能在 buffer 内兜底扫到
+	redisTTL := graceDur + s.ttlBuffer()
+	if err := s.redis.Set(ctx, HostGraceKey(roomCode), string(raw), redisTTL).Err(); err != nil {
+		logs.Warn(ctx, funcName, "写入 host_grace key 失败",
+			zap.String("room_code", roomCode), zap.Int64("host_id", hostID), zap.Error(err))
+		return
+	}
+
+	s.scheduleHostGraceTimer(roomCode, graceDur)
+	logs.Info(ctx, funcName, "host 进入宽限期",
+		zap.String("room_code", roomCode),
+		zap.Int64("host_id", hostID),
+		zap.Int("grace_seconds", s.cfg.HostGraceSeconds))
+}
+
+// OnHostReconnect host 重连钩子
+// 清除 host_grace key 并取消本地 timer；用于 host 掉线后 TTL 内重新出现在 WS 层
+// 幂等：key 不存在也不报错
+func (s *MeetingLifecycleService) OnHostReconnect(ctx context.Context, roomCode string, hostID int64) {
+	funcName := "service.meeting_lifecycle_service.OnHostReconnect"
+
+	deleted, err := s.redis.Del(ctx, HostGraceKey(roomCode)).Result()
+	if err != nil {
+		logs.Warn(ctx, funcName, "清除 host_grace key 失败",
+			zap.String("room_code", roomCode), zap.Error(err))
+	}
+	s.cancelTimer(&s.graceTimers, roomCode)
+	if deleted > 0 {
+		logs.Info(ctx, funcName, "host 宽限期内重连，撤销转让计划",
+			zap.String("room_code", roomCode), zap.Int64("host_id", hostID))
+	}
+}
+
+// HandleHostGraceExpired host 宽限期过期处理
+// 触发时机：本地 timer 到期 或 cleanup task 扫到 key TTL <=0
+// 流程：
+//  1. DEL host_grace key → 返回 0 表示已被其他路径处理，直接 return 防止重复转让
+//  2. 加载 room + 活跃成员；若 room 已 Ended / host 已变更，视为幂等完成
+//  3. 活跃成员存在：选最早加入者转 host（事务），广播 host.changed，auto_reason=host_grace_expired
+//  4. 无活跃成员：走 OnAllMembersLeft 路径（置 empty_ttl）
+func (s *MeetingLifecycleService) HandleHostGraceExpired(ctx context.Context, roomCode string) {
+	funcName := "service.meeting_lifecycle_service.HandleHostGraceExpired"
+
+	deleted, err := s.redis.Del(ctx, HostGraceKey(roomCode)).Result()
+	if err != nil {
+		logs.Warn(ctx, funcName, "DEL host_grace key 失败（跳过本次处理）",
+			zap.String("room_code", roomCode), zap.Error(err))
+		return
+	}
+	s.cancelTimer(&s.graceTimers, roomCode)
+	if deleted == 0 {
+		logs.Debug(ctx, funcName, "host_grace key 不存在（已被重连/其他节点清走）",
+			zap.String("room_code", roomCode))
+		return
+	}
+
+	room, err := s.roomDAO.GetByCode(ctx, roomCode)
+	if err != nil || room == nil {
+		logs.Warn(ctx, funcName, "加载会议失败或已不存在",
+			zap.String("room_code", roomCode), zap.Error(err))
+		return
+	}
+	if room.Status == constants.MeetingStatusEnded {
+		logs.Info(ctx, funcName, "会议已结束，跳过 host 宽限期过期处理",
+			zap.String("room_code", roomCode))
+		return
+	}
+
+	actives, err := s.participantDAO.ListActiveByRoom(ctx, room.ID)
+	if err != nil {
+		logs.Warn(ctx, funcName, "拉取活跃成员失败",
+			zap.String("room_code", roomCode), zap.Error(err))
+		return
+	}
+	// 排除原 host（若 left_at IS NULL 但 WS 掉线未落库）
+	candidates := make([]int64, 0, len(actives))
+	for _, p := range actives {
+		if p.UserID == room.HostID {
+			continue
+		}
+		candidates = append(candidates, p.UserID)
+	}
+
+	if len(candidates) == 0 {
+		logs.Info(ctx, funcName, "host 宽限期过期且无其他活跃成员，转空房 TTL",
+			zap.String("room_code", roomCode), zap.Int64("old_host_id", room.HostID))
+		// 清掉 host 自己的记录（若 WS 掉线未落库，由兜底 LeaveRoom 处理）
+		_, _ = s.participantDAO.LeaveRoom(ctx, room.ID, room.HostID, constants.MeetingLeftReasonDisconnect)
+		s.OnAllMembersLeft(ctx, roomCode)
+		return
+	}
+
+	newHostID := candidates[0]
+	if err := s.participantDAO.TransferHost(ctx, room.ID, room.HostID, newHostID); err != nil {
+		logs.Error(ctx, funcName, "host 自动转让 TransferHost 失败",
+			zap.String("room_code", roomCode), zap.Error(err))
+		return
+	}
+	if err := s.roomDAO.UpdateHost(ctx, room.ID, newHostID); err != nil {
+		logs.Warn(ctx, funcName, "UpdateHost 失败（已发生参与者 role 变更）",
+			zap.String("room_code", roomCode), zap.Error(err))
+	}
+	// 老 host 彻底离会（宽限期内未重连视为网络断开）
+	_, _ = s.participantDAO.LeaveRoom(ctx, room.ID, room.HostID, constants.MeetingLeftReasonDisconnect)
+
+	s.broadcaster.BroadcastToMeeting(ctx, room.ID, constants.MeetingWSEventHostChanged, map[string]interface{}{
+		"room_code":   roomCode,
+		"old_host_id": room.HostID,
+		"new_host_id": newHostID,
+		"auto_reason": "host_grace_expired",
+	})
+	logs.Info(ctx, funcName, "host 宽限期过期自动转让完成",
+		zap.String("room_code", roomCode),
+		zap.Int64("old_host_id", room.HostID),
+		zap.Int64("new_host_id", newHostID))
+}
+
+// OnAllMembersLeft 全员退出钩子
+// 设置 empty_ttl key + 启动本地 AfterFunc；期间有人重新加入可由 CancelEmptyTTL 复活房间
+func (s *MeetingLifecycleService) OnAllMembersLeft(ctx context.Context, roomCode string) {
+	funcName := "service.meeting_lifecycle_service.OnAllMembersLeft"
+
+	ttl := time.Duration(s.cfg.EmptyRoomTTLSeconds) * time.Second
+	// Redis TTL 同样加 buffer 避免与本地 timer 同时到期导致 DEL 无法命中
+	redisTTL := ttl + s.ttlBuffer()
+	if err := s.redis.Set(ctx, EmptyTTLKey(roomCode), "1", redisTTL).Err(); err != nil {
+		logs.Warn(ctx, funcName, "写入 empty_ttl key 失败",
+			zap.String("room_code", roomCode), zap.Error(err))
+		return
+	}
+	s.scheduleEmptyTTLTimer(roomCode, ttl)
+	logs.Info(ctx, funcName, "空房 TTL 已启动",
+		zap.String("room_code", roomCode), zap.Int("ttl_seconds", s.cfg.EmptyRoomTTLSeconds))
+}
+
+// CancelEmptyTTL 取消空房 TTL（有新成员加入时调用，房间复活）
+// 幂等：key 不存在也不报错
+func (s *MeetingLifecycleService) CancelEmptyTTL(ctx context.Context, roomCode string) {
+	funcName := "service.meeting_lifecycle_service.CancelEmptyTTL"
+
+	deleted, err := s.redis.Del(ctx, EmptyTTLKey(roomCode)).Result()
+	if err != nil {
+		logs.Warn(ctx, funcName, "清除 empty_ttl key 失败",
+			zap.String("room_code", roomCode), zap.Error(err))
+	}
+	s.cancelTimer(&s.emptyTTLTimers, roomCode)
+	if deleted > 0 {
+		logs.Info(ctx, funcName, "空房 TTL 已取消（有人加入）",
+			zap.String("room_code", roomCode))
+	}
+}
+
+// HandleEmptyRoomExpired 空房 TTL 过期处理
+// 流程：DEL empty_ttl key → 确认没被新成员清走 → MarkEnded(reason=empty_ttl) + CloseRouter 幂等
+func (s *MeetingLifecycleService) HandleEmptyRoomExpired(ctx context.Context, roomCode string) {
+	funcName := "service.meeting_lifecycle_service.HandleEmptyRoomExpired"
+
+	deleted, err := s.redis.Del(ctx, EmptyTTLKey(roomCode)).Result()
+	if err != nil {
+		logs.Warn(ctx, funcName, "DEL empty_ttl key 失败",
+			zap.String("room_code", roomCode), zap.Error(err))
+		return
+	}
+	s.cancelTimer(&s.emptyTTLTimers, roomCode)
+	if deleted == 0 {
+		logs.Debug(ctx, funcName, "empty_ttl key 已消失（有人重入 或 已处理过）",
+			zap.String("room_code", roomCode))
+		return
+	}
+
+	room, err := s.roomDAO.GetByCode(ctx, roomCode)
+	if err != nil || room == nil {
+		logs.Warn(ctx, funcName, "加载会议失败或已不存在",
+			zap.String("room_code", roomCode), zap.Error(err))
+		return
+	}
+	if room.Status == constants.MeetingStatusEnded {
+		return
+	}
+
+	// 二次校验：DEL 成功与 DB 校验之间可能有新成员加入（并发场景）
+	activeCount, _ := s.participantDAO.CountActiveByRoom(ctx, room.ID)
+	if activeCount > 0 {
+		logs.Info(ctx, funcName, "检测到活跃成员，放弃销毁",
+			zap.String("room_code", roomCode), zap.Int64("active_count", activeCount))
+		return
+	}
+
+	if _, mErr := s.roomDAO.MarkEnded(ctx, room.ID, constants.MeetingEndedReasonEmptyTTL, time.Now()); mErr != nil {
+		logs.Error(ctx, funcName, "MarkEnded 失败",
+			zap.String("room_code", roomCode), zap.Error(mErr))
+		return
+	}
+	if err := s.media.CloseRouter(ctx, roomCode); err != nil {
+		logs.Warn(ctx, funcName, "CloseRouter 失败（不影响 DB 状态）",
+			zap.String("room_code", roomCode), zap.Error(err))
+	}
+
+	s.broadcaster.BroadcastToMeeting(ctx, room.ID, constants.MeetingWSEventRoomEnded, map[string]interface{}{
+		"room_code": roomCode,
+		"reason":    constants.MeetingEndedReasonEmptyTTL,
+	})
+	logs.Info(ctx, funcName, "空房 TTL 过期，会议已销毁",
+		zap.String("room_code", roomCode), zap.Int64("room_id", room.ID))
+}
+
+// RescheduleFromRedis 服务启动时扫描 Redis 已存在的 grace / empty_ttl key，按剩余 PTTL 补装本地 timer
+// 服务重启后调用一次，避免重启期间 WS 断线事件的本地 timer 丢失导致宽限期/TTL 永远不触发
+func (s *MeetingLifecycleService) RescheduleFromRedis(ctx context.Context) {
+	funcName := "service.meeting_lifecycle_service.RescheduleFromRedis"
+
+	s.rescheduleByPattern(ctx, funcName, redisKeyHostGracePrefix+"*", func(roomCode string, dur time.Duration) {
+		s.scheduleHostGraceTimer(roomCode, dur)
+	})
+	s.rescheduleByPattern(ctx, funcName, redisKeyEmptyTTLPrefix+"*", func(roomCode string, dur time.Duration) {
+		s.scheduleEmptyTTLTimer(roomCode, dur)
+	})
+}
+
+// ScanExpired 扫描指定 key pattern，返回 TTL <=0 （或即将过期）对应的 roomCode 列表
+// 供定时任务兜底调用：若本地 timer 已丢失，此时 Redis key 可能仍存在但 TTL 为负
+// 本实现采用 SCAN + PTTL，单次扫描 limit 控制内存
+func (s *MeetingLifecycleService) ScanExpired(ctx context.Context, keyPrefix string, limit int64) []string {
+	funcName := "service.meeting_lifecycle_service.ScanExpired"
+
+	expired := make([]string, 0, limit)
+	var cursor uint64
+	for {
+		keys, next, err := s.redis.Scan(ctx, cursor, keyPrefix+"*", limit).Result()
+		if err != nil {
+			logs.Warn(ctx, funcName, "SCAN 失败", zap.String("pattern", keyPrefix), zap.Error(err))
+			return expired
+		}
+		for _, k := range keys {
+			pttl, err := s.redis.PTTL(ctx, k).Result()
+			if err != nil {
+				continue
+			}
+			// PTTL<=0（redis 过期延迟删除场景）；PTTL==-2 已不存在
+			if pttl > 0 {
+				continue
+			}
+			roomCode := keyToRoomCode(k, keyPrefix)
+			if roomCode != "" {
+				expired = append(expired, roomCode)
+			}
+		}
+		if next == 0 {
+			break
+		}
+		cursor = next
+	}
+	return expired
+}
+
+// HostGracePrefix / EmptyTTLPrefix 暴露 key 前缀供定时任务扫描使用
+func (s *MeetingLifecycleService) HostGracePrefix() string { return redisKeyHostGracePrefix }
+func (s *MeetingLifecycleService) EmptyTTLPrefix() string  { return redisKeyEmptyTTLPrefix }
+
+// ========== 内部辅助 ==========
+
+func (s *MeetingLifecycleService) scheduleHostGraceTimer(roomCode string, d time.Duration) {
+	s.cancelTimer(&s.graceTimers, roomCode)
+	timer := time.AfterFunc(d, func() {
+		ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+		defer cancel()
+		s.HandleHostGraceExpired(ctx, roomCode)
+	})
+	s.graceTimers.Store(roomCode, timer)
+}
+
+func (s *MeetingLifecycleService) scheduleEmptyTTLTimer(roomCode string, d time.Duration) {
+	s.cancelTimer(&s.emptyTTLTimers, roomCode)
+	timer := time.AfterFunc(d, func() {
+		ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+		defer cancel()
+		s.HandleEmptyRoomExpired(ctx, roomCode)
+	})
+	s.emptyTTLTimers.Store(roomCode, timer)
+}
+
+// ttlBuffer 返回 Redis key 相对本地 timer 的额外 buffer 时长
+// 用于保证本地 timer 先触发（DEL 命中），避免 Redis 自动过期后 DEL 返回 0 被误判为重复触发
+// 规则：max(CleanupIntervalSeconds * 2, 30s)，确保单机 cleanup 至少能扫到一轮兜底
+func (s *MeetingLifecycleService) ttlBuffer() time.Duration {
+	buf := time.Duration(s.cfg.CleanupIntervalSeconds*2) * time.Second
+	if buf < 30*time.Second {
+		buf = 30 * time.Second
+	}
+	return buf
+}
+
+// cancelTimer 取消本地 timer（幂等，不存在也安全）
+func (s *MeetingLifecycleService) cancelTimer(store *sync.Map, roomCode string) {
+	if raw, ok := store.LoadAndDelete(roomCode); ok {
+		if t, ok := raw.(*time.Timer); ok {
+			t.Stop()
+		}
+	}
+}
+
+// rescheduleByPattern 通用"按 pattern 扫描 Redis + 按 PTTL 重建本地 timer"逻辑
+func (s *MeetingLifecycleService) rescheduleByPattern(
+	ctx context.Context,
+	funcName string,
+	pattern string,
+	schedule func(roomCode string, dur time.Duration),
+) {
+	var cursor uint64
+	total := 0
+	for {
+		keys, next, err := s.redis.Scan(ctx, cursor, pattern, 100).Result()
+		if err != nil {
+			logs.Warn(ctx, funcName, "SCAN 失败", zap.String("pattern", pattern), zap.Error(err))
+			return
+		}
+		for _, k := range keys {
+			pttl, err := s.redis.PTTL(ctx, k).Result()
+			if err != nil || pttl <= 0 {
+				continue
+			}
+			prefix := pattern[:len(pattern)-1] // 去掉末尾 *
+			roomCode := keyToRoomCode(k, prefix)
+			if roomCode == "" {
+				continue
+			}
+			schedule(roomCode, pttl)
+			total++
+		}
+		if next == 0 {
+			break
+		}
+		cursor = next
+	}
+	if total > 0 {
+		logs.Info(ctx, funcName, "服务启动重建 timer 完成",
+			zap.String("pattern", pattern), zap.Int("count", total))
+	}
+}
+
+// keyToRoomCode 从完整 Redis key 中去掉前缀得到 roomCode
+// 返回空字符串表示格式异常
+func keyToRoomCode(fullKey, prefix string) string {
+	if len(fullKey) <= len(prefix) {
+		return ""
+	}
+	return fullKey[len(prefix):]
+}
+
+// formatErr 构造统一带上下文的错误（用于少数需要 return 的场景，避免裸字符串拼接）
+// nolint:unused
+func formatErr(funcName, msg string, err error) error {
+	return fmt.Errorf("%s: %s: %w", funcName, msg, err)
+}

backend/go-service/app/meeting/service/meeting_service.go

@@ -65,11 +65,13 @@ type MeetingService struct {
 	userResolver      UserInfoResolver
 	onlineChecker     OnlineChecker
 	mediaOrchestrator MediaOrchestrator
+	lifecycleSvc      *MeetingLifecycleService
 }
 
 // NewMeetingService 创建 MeetingService 实例
 // 依赖通过构造函数注入；接口依赖由上游 Wire 绑定到具体实现
 // Task 7 (2026-04-21) 起 mediaOrchestrator 默认绑定 HTTPMediaOrchestrator
+// Task 8 (2026-04-21) 起新增 lifecycleSvc 依赖：JoinRoom/LeaveRoom 的空房/复活场景由 lifecycleSvc 托管
 func NewMeetingService(
 	roomDAO *dao.MeetingRoomDAO,
 	participantDAO *dao.MeetingParticipantDAO,
@@ -81,6 +83,7 @@ func NewMeetingService(
 	userResolver UserInfoResolver,
 	onlineChecker OnlineChecker,
 	mediaOrchestrator MediaOrchestrator,
+	lifecycleSvc *MeetingLifecycleService,
 ) *MeetingService {
 	return &MeetingService{
 		roomDAO:           roomDAO,
@@ -93,6 +96,7 @@ func NewMeetingService(
 		userResolver:      userResolver,
 		onlineChecker:     onlineChecker,
 		mediaOrchestrator: mediaOrchestrator,
+		lifecycleSvc:      lifecycleSvc,
 	}
 }
 
@@ -210,9 +214,10 @@ func (s *MeetingService) CreateRoom(ctx context.Context, hostID int64, req *dto.
 		return nil, nil, "", err
 	}
 
+	// Task 7 起 CreateRouter 对接真实 mediasoup；Task 8 起仅在会议首次创建时调一次（房间级资源）
 	routerID, mediaErr := s.mediaOrchestrator.CreateRouter(ctx, code)
 	if mediaErr != nil {
-		logs.Warn(ctx, funcName, "mediasoup Router 创建失败（Task 7 前为占位实现，不影响流程）",
+		logs.Warn(ctx, funcName, "mediasoup Router 创建失败",
 			zap.String("room_code", code), zap.Error(mediaErr))
 		routerID = ""
 	}
@@ -309,10 +314,18 @@ func (s *MeetingService) JoinRoom(ctx context.Context, userID int64, code, passw
 		return nil, nil, "", err
 	}
 
-	routerID, mediaErr := s.mediaOrchestrator.CreateRouter(ctx, code)
-	if mediaErr != nil {
-		logs.Warn(ctx, funcName, "mediasoup Router 创建失败（占位实现）", zap.String("room_code", code), zap.Error(mediaErr))
-		routerID = ""
+	// Task 8：空房 TTL 复活
+	// 若该房间正处于 empty_ttl 阶段（全员离开后的 5 分钟窗口），新成员加入立即取消销毁
+	if s.lifecycleSvc != nil {
+		s.lifecycleSvc.CancelEmptyTTL(ctx, code)
+	}
+
+	// Task 8：JoinRoom 不再主动调 CreateRouter（Router 在 CreateRoom 时创建、由 HTTPMediaOrchestrator 本地缓存）
+	// 从缓存读取 routerID；缺失时（极少见：服务重启后未重建缓存）保持为空，不阻塞加入流程
+	routerID, _ := s.mediaOrchestrator.ResolveRouterID(code)
+	if routerID == "" {
+		logs.Debug(ctx, funcName, "RouterID 缓存缺失（非致命，可能服务重启）",
+			zap.String("room_code", code))
 	}
 
 	go s.broadcastToActiveParticipants(context.Background(), room.ID, constants.MeetingWSEventMemberJoined, map[string]interface{}{
@@ -378,9 +391,10 @@ func (s *MeetingService) LeaveRoom(ctx context.Context, userID int64, code strin
 		}
 	}
 
-	if len(actives) == 0 {
-		_, _ = s.roomDAO.MarkEnded(ctx, room.ID, constants.MeetingEndedReasonEmptyTTL, time.Now())
-		_ = s.mediaOrchestrator.CloseRouter(ctx, code)
+	// Task 8：全员离会改走生命周期状态机（空房 TTL），不再立即销毁房间
+	// 在 TTL 窗口内如有新成员加入，房间会被 CancelEmptyTTL 复活；TTL 到期由 HandleEmptyRoomExpired 兜底销毁
+	if len(actives) == 0 && s.lifecycleSvc != nil {
+		s.lifecycleSvc.OnAllMembersLeft(ctx, code)
 	}
 
 	go s.broadcastToActiveParticipants(context.Background(), room.ID, constants.MeetingWSEventMemberLeft, map[string]interface{}{

backend/go-service/app/meeting/service/meeting_signal_service.go

@@ -31,15 +31,18 @@ type MeetingSignalService struct {
 
 	broadcaster       *MeetingBroadcaster
 	mediaOrchestrator MediaOrchestrator
+	lifecycleSvc      *MeetingLifecycleService
 }
 
 // NewMeetingSignalService 构造 WS 信令服务
+// Task 8 注入 lifecycleSvc：host 掉线 / 重连的生命周期联动
 func NewMeetingSignalService(
 	roomDAO *dao.MeetingRoomDAO,
 	participantDAO *dao.MeetingParticipantDAO,
 	redis *redis.Client,
 	broadcaster *MeetingBroadcaster,
 	mediaOrchestrator MediaOrchestrator,
+	lifecycleSvc *MeetingLifecycleService,
 ) *MeetingSignalService {
 	return &MeetingSignalService{
 		roomDAO:           roomDAO,
@@ -47,6 +50,7 @@ func NewMeetingSignalService(
 		redis:             redis,
 		broadcaster:       broadcaster,
 		mediaOrchestrator: mediaOrchestrator,
+		lifecycleSvc:      lifecycleSvc,
 	}
 }
 
@@ -116,6 +120,12 @@ func (s *MeetingSignalService) OnRoomJoin(ctx context.Context, userID int64, roo
 	key := resourceTrackKey(roomCode, userID)
 	_ = s.redis.Expire(ctx, key, resourceTTL).Err()
 
+	// Task 8：若本次 WS 在线的用户是会议 host，尝试撤销 host 宽限期
+	// 本方法幂等：若此前无 host_grace key（正常场景）DEL 返回 0，不产生副作用
+	if s.lifecycleSvc != nil && room.HostID == userID {
+		s.lifecycleSvc.OnHostReconnect(ctx, roomCode, userID)
+	}
+
 	logs.Info(ctx, "service.meeting_signal_service.OnRoomJoin", "用户宣告 WS 在线",
 		zap.String("room_code", roomCode),
 		zap.Int64("user_id", userID),
@@ -123,6 +133,47 @@ func (s *MeetingSignalService) OnRoomJoin(ctx context.Context, userID int64, roo
 	return nil
 }
 
+// OnWSDisconnect WS 断线钩子，实现 ws.MeetingDisconnectHook 接口（Task 8）
+// 由 ws.handler 的 SetOnDisconnect 回调触发；场景：用户的最后一条 WS 连接被移除
+// 职责：
+//  1. 若该用户当前存在活跃 meeting_participants 记录：清理其媒体资源 + 若是 host 启动 host 宽限期
+//  2. 普通成员：仅清媒体资源（设计 Q1=A：不动 meeting_participants，长期不活跃由 4h 兜底清理）
+//
+// 容错：所有异常仅 Warn 日志，不返回 error；保证 ws.handler 主断连流程不受影响
+func (s *MeetingSignalService) OnWSDisconnect(ctx context.Context, userID int64) {
+	funcName := "service.meeting_signal_service.OnWSDisconnect"
+
+	active, err := s.participantDAO.FindActiveByUser(ctx, userID)
+	if err != nil {
+		logs.Warn(ctx, funcName, "查询活跃会议失败", zap.Int64("user_id", userID), zap.Error(err))
+		return
+	}
+	if active == nil {
+		// 用户当前不在任何会议
+		return
+	}
+	room, err := s.roomDAO.GetByID(ctx, active.RoomID)
+	if err != nil || room == nil {
+		logs.Warn(ctx, funcName, "加载 room 失败或已不存在",
+			zap.Int64("room_id", active.RoomID), zap.Error(err))
+		return
+	}
+	if room.Status == constants.MeetingStatusEnded {
+		return
+	}
+
+	s.cleanupUserResources(ctx, room.RoomCode, userID)
+
+	if s.lifecycleSvc != nil && room.HostID == userID {
+		s.lifecycleSvc.OnHostDisconnect(ctx, room.RoomCode, userID)
+	}
+
+	logs.Info(ctx, funcName, "WS 断线已处理",
+		zap.String("room_code", room.RoomCode),
+		zap.Int64("user_id", userID),
+		zap.Bool("is_host", room.HostID == userID))
+}
+
 // OnRoomLeave 处理 meeting.room.leave 事件
 // 语义：WS 层面的主动离会（等价 REST leave 但不强制要求落库事务；
 // 当前实现：仅清理该用户在本会议的所有媒体资源（batch close producer/consumer）+ 广播 meeting.member.left

backend/go-service/app/meeting/task/meeting_cleanup_task.go

@@ -0,0 +1,191 @@
+// Package task 提供 meeting 模块的后台定时任务
+package task
+
+import (
+	"context"
+	"time"
+
+	"github.com/echochat/backend/app/constants"
+	"github.com/echochat/backend/app/meeting/dao"
+	"github.com/echochat/backend/app/meeting/service"
+	"github.com/echochat/backend/pkg/logs"
+	"go.uber.org/zap"
+)
+
+// MeetingCleanupTask 会议生命周期兜底清理任务（Phase 2e-2 Task 8）
+// 职责（设计 §11.4 兜底策略 + Task 8 落地）：
+//  1. 启动时一次性 RescheduleFromRedis，按 Redis 残留 grace/empty_ttl key 重建本地 timer
+//  2. 每 cleanupInterval 秒扫描 host_grace / empty_ttl，命中"Redis 仍在但本地 timer 可能已丢失"的场景
+//     兜底触发 HandleHostGraceExpired / HandleEmptyRoomExpired；真正过期判断由 Lifecycle 内部 DEL 互斥
+//  3. 每 10 分钟扫一次 stale active rooms（活跃状态但超过 StaleRoomHours 无成员动作）→ 强制 MarkEnded(system_error)
+//  4. 每 24 小时扫 Ended 超过 chat 保留时长的房间 → DeleteByRoomIDs 清理 meeting_chats
+//
+// 所有周期都由独立 ticker 驱动，相互不阻塞；退出走 stopCh 统一关闭
+type MeetingCleanupTask struct {
+	lifecycleSvc *service.MeetingLifecycleService
+	roomDAO      *dao.MeetingRoomDAO
+	chatDAO      *dao.MeetingChatDAO
+
+	cleanupInterval time.Duration // 扫 grace/empty_ttl 的周期
+	staleInterval   time.Duration // 扫 stale active 的周期
+	chatInterval    time.Duration // 扫 chat 清理的周期
+	staleHours      int           // 超过此小时的活跃房间视为 stale
+
+	stopCh chan struct{}
+}
+
+// NewMeetingCleanupTask 构造定时任务
+// 扫描参数从 lifecycleSvc.Config() 读取，避免两处配置分离
+func NewMeetingCleanupTask(
+	lifecycleSvc *service.MeetingLifecycleService,
+	roomDAO *dao.MeetingRoomDAO,
+	chatDAO *dao.MeetingChatDAO,
+) *MeetingCleanupTask {
+	cfg := lifecycleSvc.Config()
+	interval := time.Duration(cfg.CleanupIntervalSeconds) * time.Second
+	if interval <= 0 {
+		interval = 30 * time.Second
+	}
+	staleHours := cfg.StaleRoomHours
+	if staleHours <= 0 {
+		staleHours = 4
+	}
+	return &MeetingCleanupTask{
+		lifecycleSvc:    lifecycleSvc,
+		roomDAO:         roomDAO,
+		chatDAO:         chatDAO,
+		cleanupInterval: interval,
+		staleInterval:   10 * time.Minute,
+		chatInterval:    24 * time.Hour,
+		staleHours:      staleHours,
+		stopCh:          make(chan struct{}),
+	}
+}
+
+// Start 启动任务（非阻塞）
+// 应在 main 进程启动后调用，进程退出时调用 Stop
+func (t *MeetingCleanupTask) Start() {
+	funcName := "task.meeting_cleanup_task.Start"
+	logs.Info(nil, funcName, "启动会议生命周期兜底任务",
+		zap.Duration("cleanup_interval", t.cleanupInterval),
+		zap.Duration("stale_interval", t.staleInterval),
+		zap.Duration("chat_interval", t.chatInterval),
+		zap.Int("stale_hours", t.staleHours))
+
+	go t.runRescheduleOnce()
+	go t.runLoop("grace_empty_ttl", t.cleanupInterval, 5*time.Second, t.scanExpiredKeys)
+	go t.runLoop("stale_active", t.staleInterval, 30*time.Second, t.scanStaleActive)
+	go t.runLoop("chat_cleanup", t.chatInterval, 60*time.Second, t.cleanupExpiredChats)
+}
+
+// Stop 停止所有循环
+func (t *MeetingCleanupTask) Stop() {
+	close(t.stopCh)
+}
+
+// runRescheduleOnce 启动后一次性扫描 Redis 残留 key 并补装本地 timer
+func (t *MeetingCleanupTask) runRescheduleOnce() {
+	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+	defer cancel()
+	t.lifecycleSvc.RescheduleFromRedis(ctx)
+}
+
+// runLoop 通用"固定间隔 + 预热延迟 + stopCh 退出"循环
+// warmup 用于错峰避免启动风暴，单次业务执行由 do 函数自行控制超时
+func (t *MeetingCleanupTask) runLoop(name string, interval, warmup time.Duration, do func()) {
+	funcName := "task.meeting_cleanup_task.runLoop." + name
+
+	timer := time.NewTimer(warmup)
+	select {
+	case <-t.stopCh:
+		timer.Stop()
+		return
+	case <-timer.C:
+	}
+
+	do()
+	ticker := time.NewTicker(interval)
+	defer ticker.Stop()
+	for {
+		select {
+		case <-t.stopCh:
+			logs.Info(nil, funcName, "任务循环退出")
+			return
+		case <-ticker.C:
+			do()
+		}
+	}
+}
+
+// scanExpiredKeys 扫 host_grace 与 empty_ttl 两个 key 集合
+// Redis 原生 TTL 过期时 Go 的 PTTL 会返回 -2（不存在），此时无需兜底
+// 少数场景（AfterFunc 丢失 / 多实例部署）会出现 Redis 仍有 key 但无本地 timer
+// 本扫描对 "TTL≤0 但仍存在" 的 key 主动调 HandleXxxExpired（内部 DEL 互斥）
+func (t *MeetingCleanupTask) scanExpiredKeys() {
+	funcName := "task.meeting_cleanup_task.scanExpiredKeys"
+	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
+	defer cancel()
+
+	hostExpired := t.lifecycleSvc.ScanExpired(ctx, t.lifecycleSvc.HostGracePrefix(), 100)
+	for _, code := range hostExpired {
+		t.lifecycleSvc.HandleHostGraceExpired(ctx, code)
+	}
+	emptyExpired := t.lifecycleSvc.ScanExpired(ctx, t.lifecycleSvc.EmptyTTLPrefix(), 100)
+	for _, code := range emptyExpired {
+		t.lifecycleSvc.HandleEmptyRoomExpired(ctx, code)
+	}
+	if len(hostExpired)+len(emptyExpired) > 0 {
+		logs.Info(ctx, funcName, "兜底处理过期 key 完成",
+			zap.Int("host_grace", len(hostExpired)),
+			zap.Int("empty_ttl", len(emptyExpired)))
+	}
+}
+
+// scanStaleActive 扫超 staleHours 无任何成员动作的活跃会议，强制结束
+// reason=system_error：明确标记非正常终止，便于运维排查 Node worker 掉链等异常
+func (t *MeetingCleanupTask) scanStaleActive() {
+	funcName := "task.meeting_cleanup_task.scanStaleActive"
+	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Minute)
+	defer cancel()
+
+	ids, err := t.roomDAO.ListStaleActive(ctx, t.staleHours, 100)
+	if err != nil {
+		logs.Warn(ctx, funcName, "扫描 stale active 失败", zap.Error(err))
+		return
+	}
+	if len(ids) == 0 {
+		return
+	}
+	now := time.Now()
+	for _, id := range ids {
+		if _, err := t.roomDAO.MarkEnded(ctx, id, constants.MeetingEndedReasonSystemError, now); err != nil {
+			logs.Warn(ctx, funcName, "兜底 MarkEnded 失败",
+				zap.Int64("room_id", id), zap.Error(err))
+		}
+	}
+	logs.Info(ctx, funcName, "stale 会议兜底结束", zap.Int("count", len(ids)))
+}
+
+// cleanupExpiredChats 扫超过 MeetingChatRetentionHours 的 Ended 会议，批量删除聊天记录
+func (t *MeetingCleanupTask) cleanupExpiredChats() {
+	funcName := "task.meeting_cleanup_task.cleanupExpiredChats"
+	ctx, cancel := context.WithTimeout(context.Background(), 10*time.Minute)
+	defer cancel()
+
+	ids, err := t.roomDAO.ListExpiredForCleanup(ctx, constants.MeetingChatRetentionHours, 500)
+	if err != nil {
+		logs.Warn(ctx, funcName, "扫描过期会议失败", zap.Error(err))
+		return
+	}
+	if len(ids) == 0 {
+		return
+	}
+	deleted, err := t.chatDAO.DeleteByRoomIDs(ctx, ids)
+	if err != nil {
+		logs.Warn(ctx, funcName, "批量删除聊天记录失败", zap.Error(err))
+		return
+	}
+	logs.Info(ctx, funcName, "会议聊天清理完成",
+		zap.Int("room_count", len(ids)),
+		zap.Int64("chat_deleted", deleted))
+}

backend/go-service/app/provider/provider.go

@@ -13,6 +13,7 @@ import (
 	imHandler "github.com/echochat/backend/app/im/handler"
 	meetingController "github.com/echochat/backend/app/meeting/controller"
 	meetingService "github.com/echochat/backend/app/meeting/service"
+	meetingTask "github.com/echochat/backend/app/meeting/task"
 	notifyController "github.com/echochat/backend/app/notify/controller"
 	notifyService "github.com/echochat/backend/app/notify/service"
 	notifyTask "github.com/echochat/backend/app/notify/task"
@@ -56,8 +57,10 @@ type App struct {
 	NotifyCleanupTask     *notifyTask.CleanupTask                    // 通知清理定时任务
 	MeetingService        *meetingService.MeetingService             // 会议 REST 业务服务（Task 5 落地）
 	MeetingSignalService  *meetingService.MeetingSignalService       // 会议 WS 信令业务服务（Task 6 落地）
+	MeetingLifecycleSvc   *meetingService.MeetingLifecycleService    // 会议生命周期状态机（Task 8 落地）
 	MeetingController     *meetingController.MeetingController       // 会议 REST 控制器
 	MeetingWSHandler      *meetingController.MeetingWSHandler        // 会议 WS 事件 Handler（构造时自动注册路由到 Hub）
+	MeetingCleanupTask    *meetingTask.MeetingCleanupTask            // 会议生命周期兜底定时任务（Task 8 落地）
 }
 
 // NewApp 创建应用实例
@@ -89,11 +92,14 @@ func NewApp(
 	notifyCleanup *notifyTask.CleanupTask,
 	meetingSvc *meetingService.MeetingService,
 	meetingSignalSvc *meetingService.MeetingSignalService,
+	meetingLifecycleSvc *meetingService.MeetingLifecycleService,
 	meetingCtrl *meetingController.MeetingController,
 	meetingWSHandler *meetingController.MeetingWSHandler,
+	meetingCleanup *meetingTask.MeetingCleanupTask,
 ) *App {
 	wsHandler.SetOfflinePusher(offlinePusher)
 	wsHandler.SetNotifyConnectHook(notifySvc)
+	wsHandler.SetMeetingDisconnectHook(meetingSignalSvc)
 
 	return &App{
 		Config:                  cfg,
@@ -123,8 +129,10 @@ func NewApp(
 		NotifyCleanupTask:       notifyCleanup,
 		MeetingService:          meetingSvc,
 		MeetingSignalService:    meetingSignalSvc,
+		MeetingLifecycleSvc:     meetingLifecycleSvc,
 		MeetingController:       meetingCtrl,
 		MeetingWSHandler:        meetingWSHandler,
+		MeetingCleanupTask:      meetingCleanup,
 	}
 }
 

backend/go-service/app/provider/wire.go

@@ -54,6 +54,8 @@ func InitializeApp(cfg *config.Config) (*App, error) {
 		wire.Bind(new(meetingService.NotifyPusher), new(*notifyService.NotifyService)),
 		wire.Bind(new(meetingService.UserInfoResolver), new(*contactDAO.FriendshipDAO)),
 		wire.Bind(new(meetingService.OnlineChecker), new(*wsApp.OnlineService)),
+		// Task 8：WS 断线钩子 → MeetingSignalService
+		wire.Bind(new(wsApp.MeetingDisconnectHook), new(*meetingService.MeetingSignalService)),
 	)
 	return nil, nil
 }

backend/go-service/app/provider/wire_gen.go

@@ -25,6 +25,7 @@ import (
 	controller7 "github.com/echochat/backend/app/meeting/controller"
 	dao6 "github.com/echochat/backend/app/meeting/dao"
 	service7 "github.com/echochat/backend/app/meeting/service"
+	task2 "github.com/echochat/backend/app/meeting/task"
 	controller6 "github.com/echochat/backend/app/notify/controller"
 	dao5 "github.com/echochat/backend/app/notify/dao"
 	service3 "github.com/echochat/backend/app/notify/service"
@@ -103,10 +104,12 @@ func InitializeApp(cfg *config.Config) (*App, error) {
 	meetingChatDAO := dao6.NewMeetingChatDAO(gormDB)
 	meetingBroadcaster := service7.NewMeetingBroadcaster(meetingParticipantDAO, pubSub)
 	httpMediaOrchestrator := service7.NewHTTPMediaOrchestrator(cfg)
-	meetingService := service7.NewMeetingService(meetingRoomDAO, meetingParticipantDAO, meetingChatDAO, gormDB, client, meetingBroadcaster, notifyService, friendshipDAO, onlineService, httpMediaOrchestrator)
-	meetingSignalService := service7.NewMeetingSignalService(meetingRoomDAO, meetingParticipantDAO, client, meetingBroadcaster, httpMediaOrchestrator)
+	meetingLifecycleService := service7.NewMeetingLifecycleService(meetingRoomDAO, meetingParticipantDAO, client, meetingBroadcaster, httpMediaOrchestrator, cfg)
+	meetingService := service7.NewMeetingService(meetingRoomDAO, meetingParticipantDAO, meetingChatDAO, gormDB, client, meetingBroadcaster, notifyService, friendshipDAO, onlineService, httpMediaOrchestrator, meetingLifecycleService)
+	meetingSignalService := service7.NewMeetingSignalService(meetingRoomDAO, meetingParticipantDAO, client, meetingBroadcaster, httpMediaOrchestrator, meetingLifecycleService)
 	meetingController := controller7.NewMeetingController(meetingService)
 	meetingWSHandler := controller7.NewMeetingWSHandler(meetingSignalService, hub)
-	app := NewApp(cfg, gormDB, client, minioClient, authService, authController, adminAuthController, userManageController, onlineController, contactManageController, groupManageController, messageManageController, handler, hub, pubSub, onlineService, contactController, imController, eventHandler, offlinePusher, fileController, groupController, notifyService, notificationController, cleanupTask, meetingService, meetingSignalService, meetingController, meetingWSHandler)
+	meetingCleanupTask := task2.NewMeetingCleanupTask(meetingLifecycleService, meetingRoomDAO, meetingChatDAO)
+	app := NewApp(cfg, gormDB, client, minioClient, authService, authController, adminAuthController, userManageController, onlineController, contactManageController, groupManageController, messageManageController, handler, hub, pubSub, onlineService, contactController, imController, eventHandler, offlinePusher, fileController, groupController, notifyService, notificationController, cleanupTask, meetingService, meetingSignalService, meetingLifecycleService, meetingController, meetingWSHandler, meetingCleanupTask)
 	return app, nil
 }

backend/go-service/app/ws/handler.go

@@ -35,6 +35,15 @@ type NotifyConnectHook interface {
 	PushUnreadTotalOnConnect(ctx context.Context, userID int64)
 }
 
+// MeetingDisconnectHook 会议 WS 断线钩子接口（Phase 2e-2 Task 8）
+// 由 meeting.service.MeetingSignalService 隐式实现
+// 触发时机：用户最后一条 WS 连接被移除（最终下线）
+// 职责：清理该用户在会议中的媒体资源；若为 host 则启动 host 宽限期
+// 抽象为接口避免 ws 包反向依赖 meeting 包引发循环引用
+type MeetingDisconnectHook interface {
+	OnWSDisconnect(ctx context.Context, userID int64)
+}
+
 var upgrader = websocket.Upgrader{
 	ReadBufferSize:  1024,
 	WriteBufferSize: 1024,
@@ -45,13 +54,14 @@ var upgrader = websocket.Upgrader{
 
 // Handler WebSocket 连接处理器
 type Handler struct {
-	hub              *ws.Hub
-	pubsub           *ws.PubSub
-	jwtCfg           *config.JWTConfig
-	onlineService    *OnlineService
-	tokenValidator   TokenValidator
-	offlinePusher    OfflineMessagePusher
-	notifyConnectHook NotifyConnectHook
+	hub                   *ws.Hub
+	pubsub                *ws.PubSub
+	jwtCfg                *config.JWTConfig
+	onlineService         *OnlineService
+	tokenValidator        TokenValidator
+	offlinePusher         OfflineMessagePusher
+	notifyConnectHook     NotifyConnectHook
+	meetingDisconnectHook MeetingDisconnectHook // Task 8 注入
 }
 
 // NewHandler 创建 WebSocket Handler 实例
@@ -75,6 +85,12 @@ func (h *Handler) SetNotifyConnectHook(hook NotifyConnectHook) {
 	h.notifyConnectHook = hook
 }
 
+// SetMeetingDisconnectHook 设置会议 WS 断线钩子（由 meeting 模块在初始化时注入）
+// Phase 2e-2 Task 8：WS 最终下线时触发 host 宽限期 / 媒体资源清理
+func (h *Handler) SetMeetingDisconnectHook(hook MeetingDisconnectHook) {
+	h.meetingDisconnectHook = hook
+}
+
 // Upgrade 处理 WebSocket 升级请求
 // GET /ws?token=xxx → JWT 认证 → 升级连接 → 注册 Hub → 订阅 Redis 频道
 func (h *Handler) Upgrade(c *gin.Context) {
@@ -120,6 +136,11 @@ func (h *Handler) Upgrade(c *gin.Context) {
 		}
 		h.pubsub.Unsubscribe(userID)
 		h.onlineService.UserOffline(context.Background(), userID)
+		// Task 8：通知会议模块处理 host 宽限期 / 媒体资源清理
+		// 钩子为可选（测试或 meeting 模块未加载场景安全跳过）
+		if h.meetingDisconnectHook != nil {
+			go h.meetingDisconnectHook.OnWSDisconnect(context.Background(), userID)
+		}
 	})
 	h.hub.Register(client)
 	h.pubsub.Subscribe(claims.UserID)

backend/go-service/cmd/server/main.go

@@ -82,6 +82,11 @@ func main() {
 	app.NotifyCleanupTask.Start()
 	defer app.NotifyCleanupTask.Stop()
 
+	// 7.2 启动 meeting 模块生命周期兜底任务（Phase 2e-2 Task 8）
+	// 内部会先 RescheduleFromRedis 恢复服务重启前残留的 host_grace / empty_ttl timer
+	app.MeetingCleanupTask.Start()
+	defer app.MeetingCleanupTask.Stop()
+
 	// 8. 启动 HTTP 服务（优雅关闭）
 	addr := fmt.Sprintf(":%d", cfg.Server.Port)
 	srv := &http.Server{

backend/go-service/config/config.dev.yaml

@@ -63,3 +63,11 @@ media_server:
   timeout_ms: 5000                           # 创建类接口超时（毫秒）
   close_timeout_ms: 2000                     # 关闭类接口超时（毫秒）
   close_retry: 2                             # 关闭类接口失败重试次数（指数退避 200/500ms）
+
+# 会议模块生命周期参数（Phase 2e-2 Task 8 会议状态机）
+# E2E 测试可通过环境变量 ECHOCHAT_MEETING_HOST_GRACE_SECONDS 等覆盖为更短值以加速触发
+meeting:
+  host_grace_seconds: 120           # 主持人掉线宽限期（秒），期间保留 host 身份
+  empty_room_ttl_seconds: 300       # 空房自动销毁 TTL（秒），期间有人加入可复活
+  cleanup_interval_seconds: 30      # 兜底扫描周期（秒），定时任务检查过期 grace/empty_ttl key
+  stale_room_hours: 4               # 活跃超过此小时且无活跃成员视为 stale，强制结束

backend/go-service/config/config.docker.yaml

@@ -55,3 +55,10 @@ media_server:
   timeout_ms: 5000
   close_timeout_ms: 2000
   close_retry: 2
+
+# 会议模块生命周期参数（Phase 2e-2 Task 8）
+meeting:
+  host_grace_seconds: 120
+  empty_room_ttl_seconds: 300
+  cleanup_interval_seconds: 30
+  stale_room_hours: 4

backend/go-service/config/config.go

@@ -18,6 +18,16 @@ type Config struct {
 	Log         LogConfig         `mapstructure:"log"`
 	Minio       MinioConfig       `mapstructure:"minio"`
 	MediaServer MediaServerConfig `mapstructure:"media_server"`
+	Meeting     MeetingConfig     `mapstructure:"meeting"`
+}
+
+// MeetingConfig 会议模块生命周期参数（Phase 2e-2 Task 8）
+// 所有字段均允许通过环境变量 ECHOCHAT_MEETING_* 覆盖，便于 E2E 测试以短时长加速场景触发
+type MeetingConfig struct {
+	HostGraceSeconds       int `mapstructure:"host_grace_seconds"`       // host 掉线宽限期秒数，默认 120
+	EmptyRoomTTLSeconds    int `mapstructure:"empty_room_ttl_seconds"`   // 空房自动销毁 TTL 秒数，默认 300
+	CleanupIntervalSeconds int `mapstructure:"cleanup_interval_seconds"` // 兜底扫描周期秒数，默认 30
+	StaleRoomHours         int `mapstructure:"stale_room_hours"`         // 活跃超过此小时且无成员视为 stale，默认 4
 }
 
 // MediaServerConfig Node media-server 接入配置（Phase 2e-2 Task 7）