Feat v0.6.4 health metrics (#39)

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

server/main.go

@@ -56,6 +56,7 @@ func main() {
 	}
 
 	// ── Server startup ──────────────────────
+	startTime := time.Now()
 	cfg := config.Load()
 
 	// Structured logging — must be first so all subsequent
@@ -200,19 +201,40 @@ func main() {
 	// ── WebSocket Hub ─────────────────────────
 	hub := events.NewHub(bus, middleware.GetAllowedOrigins(cfg))
 
+	// ── Node Identity ────────────────
+	// Used by cluster registry (PG) and metrics endpoint (all deployments).
+	nodeID := cfg.ClusterNodeID
+	if nodeID == "" {
+		hostname, _ := os.Hostname()
+		nodeID = fmt.Sprintf("%s-%d", hostname, os.Getpid())
+	}
+
 	// ── Cluster Registry ────────────
 	// PG-backed node self-registration + heartbeat. No-op on SQLite.
 	var clusterReg *cluster.Registry
 	if database.IsPostgres() && stores.Cluster != nil {
-		nodeID := cfg.ClusterNodeID
-		if nodeID == "" {
-			hostname, _ := os.Hostname()
-			nodeID = fmt.Sprintf("%s-%d", hostname, os.Getpid())
-		}
 		clusterReg = cluster.NewRegistry(nodeID, cfg.ClusterEndpoint, stores.Cluster, hub, cluster.RegistryConfig{
 			HeartbeatInterval: cfg.ClusterHeartbeatInterval,
 			StaleThreshold:    cfg.ClusterStaleThreshold,
 		})
+		clusterReg.SetSandboxStats(sandbox.SandboxStats)
+		clusterReg.SetTriggerFireCount(triggerEngine.FireCount)
+		clusterReg.SetExtensionCount(func() int {
+			if stores.Packages == nil {
+				return 0
+			}
+			pkgs, err := stores.Packages.List(context.Background())
+			if err != nil {
+				return 0
+			}
+			count := 0
+			for _, p := range pkgs {
+				if p.Status == "active" {
+					count++
+				}
+			}
+			return count
+		})
 		if err := clusterReg.Start(); err != nil {
 			log.Printf("⚠ Cluster registry failed to start: %v", err)
 			clusterReg = nil
@@ -255,7 +277,7 @@ func main() {
 	}
 
 	// Health check (k8s probes hit this directly)
-	base.GET("/health", func(c *gin.Context) {
+	buildHealthResponse := func() gin.H {
 		info := gin.H{
 			"status":         "ok",
 			"version":        Version,
@@ -263,8 +285,15 @@ func main() {
 			"database_name":  database.Name(),
 			"schema_version": database.SchemaVersion(),
 		}
+		if database.IsConnected() {
+			info["registration_enabled"] = handlers.IsRegistrationEnabled(stores)
+		}
 		appendClusterHealth(info, clusterReg, stores)
-		c.JSON(200, info)
+		return info
+	}
+
+	base.GET("/health", func(c *gin.Context) {
+		c.JSON(200, buildHealthResponse())
 	})
 
 	// Liveness: process is alive and serving (no dependency checks).
@@ -352,20 +381,9 @@ func main() {
 
 	api := base.Group("/api/v1")
 	{
-		// Health (routable through ingress)
+		// Health (routable through ingress — same shape as /health)
 		api.GET("/health", func(c *gin.Context) {
-			info := gin.H{
-				"status":         "ok",
-				"version":        Version,
-				"schema_version": database.SchemaVersion(),
-				"database":       database.IsConnected(),
-				"database_name":  database.Name(),
-			}
-			if database.IsConnected() {
-				info["registration_enabled"] = handlers.IsRegistrationEnabled(stores)
-			}
-			appendClusterHealth(info, clusterReg, stores)
-			c.JSON(200, info)
+			c.JSON(200, buildHealthResponse())
 		})
 
 		authGroup := api.Group("/auth")
@@ -794,6 +812,16 @@ func main() {
 				admin.GET("/cluster", clusterH.ListNodes)
 			}
 
+			// ── Metrics ─────────────────
+			metricsCollector := metrics.NewCollector(
+				nodeID, database.DB, hub, bus, stores,
+				sandbox.SandboxStats,
+				triggerEngine.FireCount,
+				startTime,
+			)
+			metricsH := handlers.NewMetricsHandler(metricsCollector)
+			admin.GET("/metrics", metricsH.GetMetrics)
+
 			// ── Backup/Restore ─────────
 			backupH := handlers.NewBackupHandler(stores, packagesDir, cfg.StoragePath)
 			admin.POST("/backup", backupH.CreateBackup)