fix: keep schedulers alive without metrics

cmd/orbit/main.go

@@ -119,7 +119,9 @@ func run() error {
 				call, cancel := context.WithTimeout(workflowCtx, 5*time.Second)
 				_, err := workflows.Tick(call, 100)
 				cancel()
-				metrics.RecordBatch("workflow", started, err)
+				if metrics != nil {
+					metrics.RecordBatch("workflow", started, err)
+				}
 				if err != nil && workflowCtx.Err() == nil {
 					slog.Error("workflow batch failed", "node_id", node)
 				}
@@ -142,7 +144,9 @@ func run() error {
 				call, cancel := context.WithTimeout(scheduleCtx, 5*time.Second)
 				_, err := schedules.Tick(call, 100)
 				cancel()
-				metrics.RecordBatch("schedule", started, err)
+				if metrics != nil {
+					metrics.RecordBatch("schedule", started, err)
+				}
 				if err != nil && scheduleCtx.Err() == nil {
 					slog.Error("schedule batch failed", "node_id", node)
 				}
@@ -170,7 +174,9 @@ func run() error {
 				call, cancel := context.WithTimeout(recoveryCtx, 5*time.Second)
 				_, err := executions.Recover(call, 100)
 				cancel()
-				metrics.RecordBatch("execution_recovery", started, err)
+				if metrics != nil {
+					metrics.RecordBatch("execution_recovery", started, err)
+				}
 				if err != nil && recoveryCtx.Err() == nil {
 					slog.Error("execution recovery batch failed", "node_id", node)
 				}

cmd/orbit/metrics_integration_test.go

@@ -135,6 +135,13 @@ func TestIntegrationMetricsProcess(t *testing.T) {
 			} else if metrics != "" {
 				t.Fatal("disabled metrics listener was started")
 			}
+			if !enabled {
+				select {
+				case <-done:
+					t.Fatal("server exited after metrics was disabled")
+				case <-time.After(2200 * time.Millisecond):
+				}
+			}
 			if err := cmd.Process.Signal(syscall.SIGTERM); err != nil {
 				t.Fatal(err)
 			}