消息服务优化

src/server/game/message_mgr.go

@@ -37,12 +37,17 @@ type MessageList struct {
 
 // Worker Pool 结构
 type WorkerPool struct {
-	workers   int
-	taskQueue chan *MessageTask
-	wg        sync.WaitGroup
-	ctx       context.Context
-	cancel    context.CancelFunc
-	maxQueue  int
+	workers       int
+	minWorkers    int
+	maxWorkers    int
+	taskQueue     chan *MessageTask
+	wg            sync.WaitGroup
+	ctx           context.Context
+	cancel        context.CancelFunc
+	maxQueue      int
+	mu            sync.Mutex
+	workerCancels []context.CancelFunc
+	monitorTick   *time.Ticker
 }
 
 // 消息任务
@@ -374,35 +379,58 @@ func MessageHandle(m *msg.Msg) error {
 // 创建 Worker Pool
 func NewWorkerPool(workers, maxQueue int) *WorkerPool {
 	ctx, cancel := context.WithCancel(context.Background())
+	if workers <= 0 {
+		workers = 1
+	}
 	pool := &WorkerPool{
-		workers:   workers,
-		taskQueue: make(chan *MessageTask, maxQueue),
-		ctx:       ctx,
-		cancel:    cancel,
-		maxQueue:  maxQueue,
+		workers:    0,
+		minWorkers: workers,
+		maxWorkers: workers * 4,
+		taskQueue:  make(chan *MessageTask, maxQueue),
+		ctx:        ctx,
+		cancel:     cancel,
+		maxQueue:   maxQueue,
 	}
 	pool.start()
+	// 启动监控协程，负责动态扩缩容
+	pool.monitorTick = time.NewTicker(500 * time.Millisecond)
+	go pool.monitor()
 	return pool
 }
 
 // 启动 Worker Pool
 func (p *WorkerPool) start() {
-	for i := 0; i < p.workers; i++ {
-		p.wg.Add(1)
-		go p.worker(i)
+	// 启动最小数量的 worker
+	for i := 0; i < p.minWorkers; i++ {
+		p.spawnWorker()
 	}
 }
 
-// Worker 工作函数
-func (p *WorkerPool) worker(id int) {
+// spawnWorker 启动一个 worker，使用独立的 cancelable context
+func (p *WorkerPool) spawnWorker() {
+	p.mu.Lock()
+	defer p.mu.Unlock()
+	childCtx, childCancel := context.WithCancel(p.ctx)
+	p.workerCancels = append(p.workerCancels, childCancel)
+	id := len(p.workerCancels)
+	p.wg.Add(1)
+	p.workers++
+	go p.worker(childCtx, id)
+}
+
+// Worker 工作函数，监听其子上下文以便单独停止
+func (p *WorkerPool) worker(ctx context.Context, id int) {
 	defer p.wg.Done()
 	log.Debug("Worker %d started", id)
 
 	for {
 		select {
-		case <-p.ctx.Done():
+		case <-ctx.Done():
 			log.Debug("Worker %d stopped", id)
 			return
+		case <-p.ctx.Done():
+			log.Debug("Worker %d pool closed", id)
+			return
 		case task, ok := <-p.taskQueue:
 			if !ok {
 				log.Debug("Worker %d: task queue closed", id)
@@ -426,6 +454,24 @@ func (p *WorkerPool) Submit(task *MessageTask) error {
 	case <-p.ctx.Done():
 		return fmt.Errorf("worker pool is closed")
 	case p.taskQueue <- task:
+		// 如果队列接近满时，尝试扩容少量 worker
+		queued := len(p.taskQueue)
+		capQ := cap(p.taskQueue)
+		if capQ > 0 && queued > capQ*3/4 {
+			go func() {
+				p.mu.Lock()
+				deficit := p.maxWorkers - p.workers
+				p.mu.Unlock()
+				if deficit > 0 {
+					// 最多扩容 2 个以避免剧烈波动
+					add := 2
+					if deficit < add {
+						add = deficit
+					}
+					p.ScaleUp(add)
+				}
+			}()
+		}
 		return nil
 	default:
 		return fmt.Errorf("task queue is full")
@@ -435,12 +481,97 @@ func (p *WorkerPool) Submit(task *MessageTask) error {
 // 关闭 Worker Pool
 func (p *WorkerPool) Shutdown() {
 	log.Debug("Shutting down worker pool...")
+	// 停止监控
+	if p.monitorTick != nil {
+		p.monitorTick.Stop()
+	}
+	// 取消根 context，会让所有子 worker 退出
 	p.cancel()
+	// 关闭任务通道，释放阻塞的接收者
 	close(p.taskQueue)
+	// 等待所有 worker 退出
 	p.wg.Wait()
 	log.Debug("Worker pool shut down complete")
 }
 
+// ScaleUp 按数量增加 worker
+func (p *WorkerPool) ScaleUp(n int) {
+	if n <= 0 {
+		return
+	}
+	p.mu.Lock()
+	can := p.maxWorkers - p.workers
+	if can <= 0 {
+		p.mu.Unlock()
+		return
+	}
+	if n > can {
+		n = can
+	}
+	p.mu.Unlock()
+
+	for i := 0; i < n; i++ {
+		p.spawnWorker()
+	}
+}
+
+// ScaleDown 按数量减少 worker
+func (p *WorkerPool) ScaleDown(n int) {
+	if n <= 0 {
+		return
+	}
+	p.mu.Lock()
+	for i := 0; i < n && len(p.workerCancels) > 0 && p.workers > p.minWorkers; i++ {
+		last := len(p.workerCancels) - 1
+		cancel := p.workerCancels[last]
+		// 从 slice 中移除
+		p.workerCancels = p.workerCancels[:last]
+		p.workers--
+		// 调用 cancel 让对应 worker 退出
+		cancel()
+	}
+	p.mu.Unlock()
+}
+
+// monitor 周期性检查队列长度并做扩缩容
+func (p *WorkerPool) monitor() {
+	for {
+		select {
+		case <-p.ctx.Done():
+			return
+		case <-p.monitorTick.C:
+			queued := len(p.taskQueue)
+			capQ := cap(p.taskQueue)
+			if capQ == 0 {
+				continue
+			}
+			// 扩容条件
+			if queued > capQ*3/4 {
+				p.mu.Lock()
+				deficit := p.maxWorkers - p.workers
+				p.mu.Unlock()
+				if deficit > 0 {
+					add := 1
+					if deficit >= 2 {
+						add = 2
+					}
+					p.ScaleUp(add)
+				}
+			}
+			// 缩容条件
+			if queued < capQ/4 {
+				p.mu.Lock()
+				extra := p.workers - p.minWorkers
+				p.mu.Unlock()
+				if extra > 0 {
+					// 每次缩一个，避免抖动
+					p.ScaleDown(1)
+				}
+			}
+		}
+	}
+}
+
 // ==================== 中间件实现 ====================
 
 // 日志中间件