Feat v0.6.4 health metrics (#39)

Co-authored-by: Jeffrey Smith <jasafpro@gmail.com>
Co-committed-by: Jeffrey Smith <jasafpro@gmail.com>

server/cluster/registry.go

@@ -34,6 +34,26 @@ type Registry struct {
 	startTime time.Time
 	stopCh    chan struct{}
 	wg        sync.WaitGroup
+
+	// Optional callbacks — set after construction via setters.
+	sandboxStats   func() (exec, errors uint64, avgMs float64)
+	triggerFires   func() int64
+	extensionCount func() int
+}
+
+// SetSandboxStats registers a callback for sandbox execution counters.
+func (r *Registry) SetSandboxStats(fn func() (uint64, uint64, float64)) {
+	r.sandboxStats = fn
+}
+
+// SetTriggerFireCount registers a callback for trigger fire count.
+func (r *Registry) SetTriggerFireCount(fn func() int64) {
+	r.triggerFires = fn
+}
+
+// SetExtensionCount registers a callback for active extension count.
+func (r *Registry) SetExtensionCount(fn func() int) {
+	r.extensionCount = fn
 }
 
 // NewRegistry creates a cluster registry instance.
@@ -129,13 +149,28 @@ func (r *Registry) collectStats() json.RawMessage {
 	runtime.ReadMemStats(&m)
 
 	stats := map[string]any{
-		"goroutines":  runtime.NumGoroutine(),
-		"heap_alloc":  m.HeapAlloc,
-		"heap_sys":    m.HeapSys,
-		"gc_cycles":   m.NumGC,
-		"gc_pause_ns": m.PauseNs[(m.NumGC+255)%256],
-		"uptime_sec":  time.Since(r.startTime).Seconds(),
-		"ws_clients":  r.hub.ConnCount(),
+		"goroutines":    runtime.NumGoroutine(),
+		"heap_alloc":    m.HeapAlloc,
+		"heap_sys":      m.HeapSys,
+		"stack_in_use":  m.StackInuse,
+		"gc_cycles":     m.NumGC,
+		"gc_pause_ns":   m.PauseNs[(m.NumGC+255)%256],
+		"gc_cpu_pct":    m.GCCPUFraction * 100,
+		"uptime_sec":    time.Since(r.startTime).Seconds(),
+		"ws_clients":    r.hub.ConnCount(),
+	}
+
+	if r.extensionCount != nil {
+		stats["extensions_loaded"] = r.extensionCount()
+	}
+	if r.sandboxStats != nil {
+		exec, errors, avgMs := r.sandboxStats()
+		stats["starlark_exec_total"] = exec
+		stats["starlark_errors_total"] = errors
+		stats["starlark_avg_duration_ms"] = avgMs
+	}
+	if r.triggerFires != nil {
+		stats["trigger_fires_total"] = r.triggerFires()
 	}
 
 	data, _ := json.Marshal(stats)

server/cluster/registry_test.go

@@ -75,6 +75,46 @@ func TestCollectStats(t *testing.T) {
 	}
 }
 
+func TestCollectStatsFattened(t *testing.T) {
+	hub := &mockHub{count: 3}
+	reg := NewRegistry("test-node", "", nil, hub, RegistryConfig{
+		HeartbeatInterval: 10 * time.Second,
+		StaleThreshold:    30 * time.Second,
+	})
+	reg.SetSandboxStats(func() (uint64, uint64, float64) { return 100, 5, 12.3 })
+	reg.SetTriggerFireCount(func() int64 { return 42 })
+	reg.SetExtensionCount(func() int { return 8 })
+
+	data := reg.collectStats()
+
+	var stats map[string]any
+	if err := json.Unmarshal(data, &stats); err != nil {
+		t.Fatalf("collectStats returned invalid JSON: %v", err)
+	}
+
+	// New fattened keys
+	fatKeys := []string{
+		"stack_in_use", "gc_cpu_pct",
+		"extensions_loaded", "starlark_exec_total", "starlark_errors_total",
+		"starlark_avg_duration_ms", "trigger_fires_total",
+	}
+	for _, key := range fatKeys {
+		if _, ok := stats[key]; !ok {
+			t.Errorf("missing fattened stats key: %s", key)
+		}
+	}
+
+	if v := stats["extensions_loaded"].(float64); int(v) != 8 {
+		t.Errorf("extensions_loaded = %v, want 8", v)
+	}
+	if v := stats["starlark_exec_total"].(float64); int(v) != 100 {
+		t.Errorf("starlark_exec_total = %v, want 100", v)
+	}
+	if v := stats["trigger_fires_total"].(float64); int(v) != 42 {
+		t.Errorf("trigger_fires_total = %v, want 42", v)
+	}
+}
+
 func TestRegistryStartStop(t *testing.T) {
 	ms := &mockClusterStore{heartbeatRows: 1}
 	hub := &mockHub{count: 0}