feat: tune websocket hub backpressure

internal/observability/catalog.go

@@ -66,5 +66,9 @@ func catalog() map[string]descriptor {
 	c["orbit_runtime_tick_interval_seconds"] = descriptor{gauge, "Configured background controller tick interval in seconds", map[string][]string{"component": components}}
 	c["orbit_runtime_batch_limit"] = descriptor{gauge, "Configured maximum candidates per background batch", nil}
 	c["orbit_shutdown_timeout_seconds"] = descriptor{gauge, "Configured graceful drain and shutdown window in seconds", nil}
+	c["orbit_worker_connection_limit"] = descriptor{gauge, "Configured maximum Worker WebSocket connections on this node", nil}
+	c["orbit_event_connection_limit"] = descriptor{gauge, "Configured maximum Event WebSocket connections on this node", nil}
+	c["orbit_dispatch_workers"] = descriptor{gauge, "Configured database claim workers per WebSocket hub", nil}
+	c["orbit_dispatch_capacity"] = descriptor{gauge, "Configured bounded in-memory claim queue per WebSocket hub", nil}
 	return c
 }

internal/observability/metrics_test.go

@@ -96,7 +96,7 @@ func TestInvalidNumbersAndMonotoneCounters(t *testing.T) {
 }
 func TestRuntimeObservationDoesNotExposeErrors(t *testing.T) {
 	r := NewRegistry()
-	r.RecordRuntimeConfig(RuntimeConfig{WorkflowTickInterval: time.Second, ScheduleTickInterval: 1500 * time.Millisecond, RecoveryTickInterval: 2 * time.Second, BatchLimit: 37, ShutdownTimeout: 45 * time.Second})
+	r.RecordRuntimeConfig(RuntimeConfig{WorkflowTickInterval: time.Second, ScheduleTickInterval: 1500 * time.Millisecond, RecoveryTickInterval: 2 * time.Second, BatchLimit: 37, ShutdownTimeout: 45 * time.Second, WorkerConnectionLimit: 512, EventConnectionLimit: 256, DispatchWorkers: 16, DispatchCapacity: 2048})
 	r.RecordDraining(true)
 	r.RecordBatch("workflow", time.Now().Add(-time.Millisecond), nil)
 	r.RecordBatch("workflow", time.Now(), errors.New("private-dsn"))
@@ -108,7 +108,7 @@ func TestRuntimeObservationDoesNotExposeErrors(t *testing.T) {
 			t.Fatal("missing runtime evidence", want)
 		}
 	}
-	for _, want := range []string{`orbit_runtime_tick_interval_seconds{component="workflow"} 1`, `orbit_runtime_tick_interval_seconds{component="schedule"} 1.5`, `orbit_runtime_tick_interval_seconds{component="execution_recovery"} 2`, "orbit_runtime_batch_limit 37\n", "orbit_shutdown_timeout_seconds 45\n", "orbit_process_draining 1\n"} {
+	for _, want := range []string{`orbit_runtime_tick_interval_seconds{component="workflow"} 1`, `orbit_runtime_tick_interval_seconds{component="schedule"} 1.5`, `orbit_runtime_tick_interval_seconds{component="execution_recovery"} 2`, "orbit_runtime_batch_limit 37\n", "orbit_shutdown_timeout_seconds 45\n", "orbit_worker_connection_limit 512\n", "orbit_event_connection_limit 256\n", "orbit_dispatch_workers 16\n", "orbit_dispatch_capacity 2048\n", "orbit_process_draining 1\n"} {
 		if !strings.Contains(out, want) {
 			t.Fatal("missing runtime configuration", want)
 		}

internal/observability/runtime.go

@@ -13,16 +13,20 @@ import (
 // RuntimeConfig 是节点启动后实际生效的后台控制参数。
 // 它只用于固定基数观测，不参与业务授权或持久化状态判断。
 type RuntimeConfig struct {
-	WorkflowTickInterval time.Duration
-	ScheduleTickInterval time.Duration
-	RecoveryTickInterval time.Duration
-	BatchLimit           int
-	ShutdownTimeout      time.Duration
+	WorkflowTickInterval  time.Duration
+	ScheduleTickInterval  time.Duration
+	RecoveryTickInterval  time.Duration
+	BatchLimit            int
+	ShutdownTimeout       time.Duration
+	WorkerConnectionLimit int // Worker WebSocket 单节点连接上限。
+	EventConnectionLimit  int // Event WebSocket 单节点连接上限。
+	DispatchWorkers       int // 每个 Hub 同时执行数据库领取调用的协程上限。
+	DispatchCapacity      int // 每个 Hub 领取调用的有界内存排队槽位。
 }
 
 // RecordRuntimeConfig records the effective bounded controller settings.
 func (r *Registry) RecordRuntimeConfig(config RuntimeConfig) {
-	if r == nil || config.WorkflowTickInterval <= 0 || config.ScheduleTickInterval <= 0 || config.RecoveryTickInterval <= 0 || config.BatchLimit <= 0 || config.ShutdownTimeout <= 0 {
+	if r == nil || config.WorkflowTickInterval <= 0 || config.ScheduleTickInterval <= 0 || config.RecoveryTickInterval <= 0 || config.BatchLimit <= 0 || config.ShutdownTimeout <= 0 || config.WorkerConnectionLimit <= 0 || config.EventConnectionLimit <= 0 || config.DispatchWorkers <= 0 || config.DispatchCapacity <= 0 {
 		return
 	}
 	r.SetGauge("orbit_runtime_tick_interval_seconds", config.WorkflowTickInterval.Seconds(), Labels{"component": "workflow"})
@@ -30,6 +34,10 @@ func (r *Registry) RecordRuntimeConfig(config RuntimeConfig) {
 	r.SetGauge("orbit_runtime_tick_interval_seconds", config.RecoveryTickInterval.Seconds(), Labels{"component": "execution_recovery"})
 	r.SetGauge("orbit_runtime_batch_limit", float64(config.BatchLimit), nil)
 	r.SetGauge("orbit_shutdown_timeout_seconds", config.ShutdownTimeout.Seconds(), nil)
+	r.SetGauge("orbit_worker_connection_limit", float64(config.WorkerConnectionLimit), nil)
+	r.SetGauge("orbit_event_connection_limit", float64(config.EventConnectionLimit), nil)
+	r.SetGauge("orbit_dispatch_workers", float64(config.DispatchWorkers), nil)
+	r.SetGauge("orbit_dispatch_capacity", float64(config.DispatchCapacity), nil)
 }
 
 // RecordDraining records whether this process has entered graceful shutdown.