Add HPA example and fix autoscaling support (#13)

api/v1/deployment_types.go

@@ -20,7 +20,6 @@ type ApplicationSpec struct {
 	// Ignored when HorizontalPodAutoscaler is configured.
 	// +optional
 	// +kubebuilder:validation:Minimum=0
-	// +kubebuilder:default=1
 	Replicas *int32 `json:"replicas,omitempty"`
 
 	// Resources specifies compute resource requirements.

api/v1/logicflowruntime_webhook.go

@@ -6,6 +6,25 @@ import (
 	"sigs.k8s.io/controller-runtime/pkg/webhook/admission"
 )
 
+// +kubebuilder:webhook:path=/mutate-logic-kubesmarts-org-v1-logicflowruntime,mutating=true,failurePolicy=fail,sideEffects=None,groups=logic.kubesmarts.org,resources=logicflowruntimes,verbs=create;update,versions=v1,name=mlogicflowruntime-v1.kb.io,admissionReviewVersions=v1
+
+// +kubebuilder:object:generate=false
+type LogicFlowRuntimeDefaulter struct{}
+
+var _ admission.Defaulter[*LogicFlowRuntime] = &LogicFlowRuntimeDefaulter{}
+
+// Default sets Replicas to 1 if not specified.
+// We use a webhook instead of +kubebuilder:default=1 because ApplicationSpec is also used by LogicPlatform.DataIndex,
+// which needs to support HPA management without operator enforcement. The webhook allows us to apply the default
+// conditionally only for LogicFlowRuntime, leaving DataIndex.Application.Replicas nil for HPA to manage freely.
+func (d *LogicFlowRuntimeDefaulter) Default(_ context.Context, obj *LogicFlowRuntime) error {
+	if obj.Spec.Replicas == nil {
+		one := int32(1)
+		obj.Spec.Replicas = &one
+	}
+	return nil
+}
+
 // +kubebuilder:webhook:path=/validate-logic-kubesmarts-org-v1-logicflowruntime,mutating=false,failurePolicy=fail,sideEffects=None,groups=logic.kubesmarts.org,resources=logicflowruntimes,verbs=create;update,versions=v1,name=vlogicflowruntime-v1.kb.io,admissionReviewVersions=v1
 
 type LogicFlowRuntimeValidator struct{}

api/v1/webhook_integration_test.go

@@ -81,6 +81,7 @@ func TestWebhookIntegration(t *testing.T) {
 		t.Fatalf("register LFD webhook: %v", err)
 	}
 	if err := builder.WebhookManagedBy(mgr, &LogicFlowRuntime{}).
+		WithDefaulter(&LogicFlowRuntimeDefaulter{}).
 		WithValidator(&LogicFlowRuntimeValidator{}).
 		Complete(); err != nil {
 		t.Fatalf("register LFR webhook: %v", err)

cmd/main.go

@@ -248,6 +248,7 @@ func main() {
 		os.Exit(1)
 	}
 	if err := builder.WebhookManagedBy(mgr, &logicv1.LogicFlowRuntime{}).
+		WithDefaulter(&logicv1.LogicFlowRuntimeDefaulter{}).
 		WithValidator(&logicv1.LogicFlowRuntimeValidator{}).
 		Complete(); err != nil {
 		setupLog.Error(err, "unable to create webhook", "webhook", "LogicFlowRuntime")

config/crd/bases/logic.kubesmarts.org_logicflowruntimes.yaml

@@ -9474,7 +9474,6 @@ spec:
                     type: array
                 type: object
               replicas:
-                default: 1
                 description: |-
                   Replicas is the desired number of pod replicas.
                   Ignored when HorizontalPodAutoscaler is configured.

config/crd/bases/logic.kubesmarts.org_logicplatforms.yaml

@@ -9500,7 +9500,6 @@ spec:
                             type: array
                         type: object
                       replicas:
-                        default: 1
                         description: |-
                           Replicas is the desired number of pod replicas.
                           Ignored when HorizontalPodAutoscaler is configured.
@@ -19276,7 +19275,6 @@ spec:
                                 type: array
                             type: object
                           replicas:
-                            default: 1
                             description: |-
                               Replicas is the desired number of pod replicas.
                               Ignored when HorizontalPodAutoscaler is configured.
@@ -28834,7 +28832,6 @@ spec:
                         type: array
                     type: object
                   replicas:
-                    default: 1
                     description: |-
                       Replicas is the desired number of pod replicas.
                       Ignored when HorizontalPodAutoscaler is configured.

config/webhook/manifests.yaml

@@ -4,6 +4,26 @@ kind: MutatingWebhookConfiguration
 metadata:
   name: mutating-webhook-configuration
 webhooks:
+- admissionReviewVersions:
+  - v1
+  clientConfig:
+    service:
+      name: webhook-service
+      namespace: system
+      path: /mutate-logic-kubesmarts-org-v1-logicflowruntime
+  failurePolicy: Fail
+  name: mlogicflowruntime-v1.kb.io
+  rules:
+  - apiGroups:
+    - logic.kubesmarts.org
+    apiVersions:
+    - v1
+    operations:
+    - CREATE
+    - UPDATE
+    resources:
+    - logicflowruntimes
+  sideEffects: None
 - admissionReviewVersions:
   - v1
   clientConfig:

docs/antora/modules/ROOT/pages/deployment/production.adoc

@@ -90,13 +90,81 @@ spec:
 
 == Horizontal Pod Autoscaling
 
-The operator sets `spec.replicas` on the managed `Deployment`.
-Create an `HorizontalPodAutoscaler` targeting that `Deployment` to scale based on CPU or custom metrics.
+The operator exposes the Kubernetes scale subresource on LogicFlowRuntime and does not enforce default replicas on LogicPlatform.DataIndex, enabling HPA management for both.
+
+=== LogicFlowRuntime Autoscaling
+
+Target LogicFlowRuntime directly (not the Deployment).
+HPA updates `spec.replicas` on the CR, which the operator uses to reconcile both the Deployment AND lease pool.
+If you target the Deployment directly, leases won't track replica changes.
+
+Example HPA for a LogicFlowRuntime named `hello-runtime`:
+
+[source,yaml]
+----
+apiVersion: autoscaling/v2
+kind: HorizontalPodAutoscaler
+metadata:
+  name: hello-runtime-autoscaler
+  namespace: default
+spec:
+  scaleTargetRef:
+    apiVersion: logic.kubesmarts.org/v1
+    kind: LogicFlowRuntime
+    name: hello-runtime
+  minReplicas: 1
+  maxReplicas: 10
+  metrics:
+  - type: Resource
+    resource:
+      name: cpu
+      target:
+        type: Utilization
+        averageUtilization: 70
+----
 
 [NOTE]
 ====
 When PostgreSQL persistence is enabled, scale-up is safe — leases ensure each workflow instance is owned by exactly one pod.
 Without persistence, scale-up creates duplicate in-memory state for in-flight workflows.
+
+The operator manages lease assignment/cleanup automatically when replica count changes.
+====
+
+=== Data Index Autoscaling
+
+Target the generated Data Index Deployment directly (`flow-platform-data-index` or similar).
+Leave `spec.dataIndex.application.replicas` unset in LogicPlatform so the operator doesn't enforce a value.
+
+Example HPA for Data Index:
+
+[source,yaml]
+----
+apiVersion: autoscaling/v2
+kind: HorizontalPodAutoscaler
+metadata:
+  name: data-index-autoscaler
+  namespace: data-index-system
+spec:
+  scaleTargetRef:
+    apiVersion: apps/v1
+    kind: Deployment
+    name: flow-platform-data-index  # Deployment created by operator
+  minReplicas: 1
+  maxReplicas: 5
+  metrics:
+  - type: Resource
+    resource:
+      name: cpu
+      target:
+        type: Utilization
+        averageUtilization: 75
+----
+
+[NOTE]
+====
+Do not set `spec.dataIndex.application.replicas` when using HPA.
+The operator respects the value if set, but will override HPA scaling.
 ====
 
 == cert-manager