fix(agent): reduce reconcile churn with configurable retry interval

Author: syscod3

internal/agent/reconciler.go

@@ -46,6 +46,9 @@ type ImpVMReconciler struct {
 	// StartTimeout is how long a VM may remain in Starting before being
 	// transitioned to Failed. Defaults to 5 minutes when zero.
 	StartTimeout time.Duration
+	// RetryInterval is how long the agent waits before retrying high-churn
+	// reconcile paths (Starting and Terminating stop errors). Defaults to 5s.
+	RetryInterval time.Duration
 	// Recorder emits lifecycle events for VM completion/failure.
 	Recorder record.EventRecorder
 }
@@ -104,6 +107,13 @@ func (r *ImpVMReconciler) startTimeout() time.Duration {
 	return 5 * time.Minute
 }
 
+func (r *ImpVMReconciler) retryInterval() time.Duration {
+	if r.RetryInterval > 0 {
+		return r.RetryInterval
+	}
+	return 5 * time.Second
+}
+
 func (r *ImpVMReconciler) handleStarting(ctx context.Context, vm *impdevv1alpha1.ImpVM) (ctrl.Result, error) {
 	log := logf.FromContext(ctx)
 	if vm.Status.StartedAt != nil {
@@ -112,7 +122,7 @@ func (r *ImpVMReconciler) handleStarting(ctx context.Context, vm *impdevv1alpha1
 			return r.finishFailed(ctx, vm)
 		}
 	}
-	return ctrl.Result{RequeueAfter: 2 * time.Second}, nil
+	return ctrl.Result{RequeueAfter: r.retryInterval()}, nil
 }
 
 func (r *ImpVMReconciler) handleScheduled(ctx context.Context, vm *impdevv1alpha1.ImpVM) (result ctrl.Result, err error) {
@@ -294,7 +304,7 @@ func (r *ImpVMReconciler) handleTerminating(ctx context.Context, vm *impdevv1alp
 
 	if err = r.Driver.Stop(ctx, vm); err != nil {
 		log.Error(err, "Driver Stop failed — will retry")
-		return ctrl.Result{RequeueAfter: 2 * time.Second}, err
+		return ctrl.Result{RequeueAfter: r.retryInterval()}, err
 	}
 
 	if r.Metrics != nil {

internal/agent/reconciler_test.go

@@ -217,10 +217,10 @@ var _ = Describe("ImpVM Agent: Terminating → clears nodeName", func() {
 	})
 })
 
-var _ = Describe("ImpVM Agent: VMPhaseStarting → RequeueAfter 2s", func() {
+var _ = Describe("ImpVM Agent: VMPhaseStarting → RequeueAfter", func() {
 	ctx := context.Background()
 
-	It("returns RequeueAfter=2s and does not mutate status", func() {
+	It("returns RequeueAfter=5s by default and does not mutate status", func() {
 		vm := &impdevv1alpha1.ImpVM{
 			ObjectMeta: metav1.ObjectMeta{
 				Name: "tc5-starting", Namespace: "default",
@@ -239,12 +239,36 @@ var _ = Describe("ImpVM Agent: VMPhaseStarting → RequeueAfter 2s", func() {
 			NamespacedName: types.NamespacedName{Name: "tc5-starting", Namespace: "default"},
 		})
 		Expect(err).NotTo(HaveOccurred())
-		Expect(result.RequeueAfter).To(Equal(2 * time.Second))
+		Expect(result.RequeueAfter).To(Equal(5 * time.Second))
 
 		updated := &impdevv1alpha1.ImpVM{}
 		Expect(k8sClient.Get(ctx, types.NamespacedName{Name: "tc5-starting", Namespace: "default"}, updated)).To(Succeed())
 		Expect(updated.Status.Phase).To(Equal(impdevv1alpha1.VMPhaseStarting))
 	})
+
+	It("uses explicit RetryInterval when configured", func() {
+		vm := &impdevv1alpha1.ImpVM{
+			ObjectMeta: metav1.ObjectMeta{
+				Name: "tc5-starting-override", Namespace: "default",
+				Finalizers: []string{"imp/finalizer"},
+			},
+			Spec: impdevv1alpha1.ImpVMSpec{NodeName: testNode},
+		}
+		Expect(k8sClient.Create(ctx, vm)).To(Succeed())
+		DeferCleanup(func() { k8sClient.Delete(ctx, vm) }) //nolint:errcheck
+
+		base := vm.DeepCopy()
+		vm.Status.Phase = impdevv1alpha1.VMPhaseStarting
+		Expect(k8sClient.Status().Patch(ctx, vm, client.MergeFrom(base))).To(Succeed())
+
+		r := newReconciler(NewStubDriver())
+		r.RetryInterval = 9 * time.Second
+		result, err := r.Reconcile(ctx, reconcile.Request{
+			NamespacedName: types.NamespacedName{Name: "tc5-starting-override", Namespace: "default"},
+		})
+		Expect(err).NotTo(HaveOccurred())
+		Expect(result.RequeueAfter).To(Equal(9 * time.Second))
+	})
 })
 
 var _ = Describe("ImpVM Agent: nodeName filter", func() {