feat(api): add QdrantCluster.spec.onDemandReplication field

api/v1/integration_test.go

@@ -7,21 +7,22 @@ import (
 	. "github.com/onsi/gomega"
 	corev1 "k8s.io/api/core/v1"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"sigs.k8s.io/controller-runtime/pkg/client"
 )
 
 var _ = Describe("QdrantCluster", func() {
 	Context("API integration tests", func() {
 		const namespaceName = "test-namespace"
 		ctx := context.Background()
-		namespace := &corev1.Namespace{
-			ObjectMeta: metav1.ObjectMeta{
-				Name: namespaceName,
-			},
-		}
 		BeforeEach(func() {
 			By("Creating the Namespace to perform the tests")
+			namespace := &corev1.Namespace{
+				ObjectMeta: metav1.ObjectMeta{
+					Name: namespaceName,
+				},
+			}
 			err := k8sClient.Create(ctx, namespace)
-			Expect(err).To(Not(HaveOccurred()))
+			Expect(client.IgnoreAlreadyExists(err)).To(Not(HaveOccurred()))
 		})
 		It("should not flip ServicePerNode value on update", func() {
 			qc := QdrantCluster{
@@ -44,6 +45,26 @@ var _ = Describe("QdrantCluster", func() {
 			Expect(err).To(Not(HaveOccurred()))
 			Expect(DerefPointer(qc.Spec.ServicePerNode)).To(BeFalse())
 		})
+		It("should default OnDemandReplication to Off when omitted", func() {
+			qc := QdrantCluster{
+				ObjectMeta: metav1.ObjectMeta{
+					Namespace: namespaceName,
+					Name:      "test-cluster-on-demand-replication",
+				},
+				Spec: QdrantClusterSpec{
+					Id:      "test-cluster-on-demand-replication",
+					Version: "v1.18.0",
+					Size:    1,
+				},
+			}
+			err := k8sClient.Create(ctx, &qc)
+			Expect(err).To(Not(HaveOccurred()))
+
+			created := &QdrantCluster{}
+			err = k8sClient.Get(ctx, client.ObjectKeyFromObject(&qc), created)
+			Expect(err).To(Not(HaveOccurred()))
+			Expect(created.Spec.OnDemandReplication).To(Equal(OnDemandReplicationOff))
+		})
 	})
 
 })

api/v1/qdrantcluster_types.go

@@ -23,6 +23,11 @@ const (
 	// If the value is updated it will retrigger the restart.
 	// For historical reasons the key doesn't start with `operator.qdrant.com/`
 	RestartedAtAnnotationKey = "restartedAt"
+	// OnDemandReplicationRestartAnnotationKey is the annotation key to trigger an on-demand replication restart.
+	// The annotation should be placed on the QdrantCluster instance.
+	// The value should be a [RFC3339 formatted] date.
+	// If the value is updated it will retrigger the restart.
+	OnDemandReplicationRestartAnnotationKey = "operator.qdrant.com/on-demand-replication-restart"
 	// RecreateNodeAnnotationKey is the annotation key to recreate a certain node.
 	// The annotation should be placed on the pod created by the operator (for the node that need to be recreated).
 	// It is allowed to add this annotation to multiple pods, the operator will handle them all.
@@ -58,6 +63,17 @@ const (
 	ByCountAndSize RebalanceStrategy = "by_count_and_size"
 )
 
+// OnDemandReplicationType specifies the on-demand replication restart mode.
+// +kubebuilder:validation:Enum=Off;Auto;On
+type OnDemandReplicationType string
+
+//goland:noinspection GoUnusedConst
+const (
+	OnDemandReplicationOff  OnDemandReplicationType = "Off"
+	OnDemandReplicationAuto OnDemandReplicationType = "Auto"
+	OnDemandReplicationOn   OnDemandReplicationType = "On"
+)
+
 // QdrantClusterSpec defines the desired state of QdrantCluster
 // +kubebuilder:pruning:PreserveUnknownFields
 type QdrantClusterSpec struct {
@@ -136,6 +152,13 @@ type QdrantClusterSpec struct {
 	// If unset, the operator is going to restart nodes concurrently if none of the collections if replicated.
 	// +optional
 	RestartAllPodsConcurrently *bool `json:"restartAllPodsConcurrently,omitempty"`
+	// OnDemandReplication specifies the on-demand replication restart mode.
+	// Off (default): Normal restart behavior. Pods are restarted directly.
+	// Auto: The operator checks telemetry for non-replicated shards. If found, uses the recreate-node flow.
+	// On: Always uses the recreate-node flow for eligible restart triggers.
+	// +kubebuilder:default=Off
+	// +optional
+	OnDemandReplication OnDemandReplicationType `json:"onDemandReplication,omitempty"`
 	// If StartupDelaySeconds is set (> 0), an additional 'sleep <value>' will be emitted to the pod startup.
 	// The sleep will be added when a pod is restarted, it will not force any pod to restart.
 	// This feature can be used for debugging the core, e.g. if a pod is in crash loop, it provided a way

api/v1/qdrantcluster_types_test.go

@@ -1,13 +1,21 @@
 package v1
 
 import (
+	"encoding/json"
 	"fmt"
 	"testing"
 
 	"github.com/stretchr/testify/assert"
 	"k8s.io/utils/ptr"
 )
 
+func TestQdrantClusterSpecJSONOmitsUnsetOnDemandReplication(t *testing.T) {
+	data, err := json.Marshal(QdrantClusterSpec{})
+
+	assert.NoError(t, err)
+	assert.NotContains(t, string(data), "onDemandReplication")
+}
+
 func TestValidate(t *testing.T) {
 	testCases := []struct {
 		name          string

charts/qdrant-kubernetes-api/templates/region-crds/qdrant.io_qdrantclusters.yaml

@@ -525,6 +525,18 @@ spec:
                 description: NodeSelector specifies the node selector for each Qdrant
                   node.
                 type: object
+              onDemandReplication:
+                default: "Off"
+                description: |-
+                  OnDemandReplication specifies the on-demand replication restart mode.
+                  Off (default): Normal restart behavior. Pods are restarted directly.
+                  Auto: The operator checks telemetry for non-replicated shards. If found, uses the recreate-node flow.
+                  On: Always uses the recreate-node flow for eligible restart triggers.
+                enum:
+                - "Off"
+                - Auto
+                - "On"
+                type: string
               pauses:
                 description: |-
                   Pauses specifies a list of pause request by developer for manual maintenance.

crds/qdrant.io_qdrantclusters.yaml

@@ -524,6 +524,18 @@ spec:
                 description: NodeSelector specifies the node selector for each Qdrant
                   node.
                 type: object
+              onDemandReplication:
+                default: "Off"
+                description: |-
+                  OnDemandReplication specifies the on-demand replication restart mode.
+                  Off (default): Normal restart behavior. Pods are restarted directly.
+                  Auto: The operator checks telemetry for non-replicated shards. If found, uses the recreate-node flow.
+                  On: Always uses the recreate-node flow for eligible restart triggers.
+                enum:
+                - "Off"
+                - Auto
+                - "On"
+                type: string
               pauses:
                 description: |-
                   Pauses specifies a list of pause request by developer for manual maintenance.

docs/api.md

@@ -620,6 +620,25 @@ _Appears in:_
 | `snapshotsPVCStatus` _[NodePVCStatus](#nodepvcstatus)_ | Status of the snapshots storage PVC |  | Optional: \{\} <br /> |
 
 
+#### OnDemandReplicationType
+
+_Underlying type:_ _string_
+
+OnDemandReplicationType specifies the on-demand replication restart mode.
+
+_Validation:_
+- Enum: [Off Auto On]
+
+_Appears in:_
+- [QdrantClusterSpec](#qdrantclusterspec)
+
+| Field | Description |
+| --- | --- |
+| `Off` |  |
+| `Auto` |  |
+| `On` |  |
+
+
 #### Pause
 
 
@@ -976,6 +995,7 @@ _Appears in:_
 | `topologySpreadConstraints` _[TopologySpreadConstraint](https://kubernetes.io/docs/reference/generated/kubernetes-api/v1.28/#topologyspreadconstraint-v1-core)_ | TopologySpreadConstraints specifies the topology spread constraints for the cluster. |  | Optional: \{\} <br /> |
 | `podDisruptionBudget` _[PodDisruptionBudgetSpec](https://kubernetes.io/docs/reference/generated/kubernetes-api/v1.28/#poddisruptionbudgetspec-v1-policy)_ | PodDisruptionBudget specifies the pod disruption budget for the cluster. |  | Optional: \{\} <br /> |
 | `restartAllPodsConcurrently` _boolean_ | RestartAllPodsConcurrently specifies whether to restart all pods concurrently (also called one-shot-restart).<br />If enabled, all the pods in the cluster will be restarted concurrently in situations where multiple pods<br />need to be restarted, like when RestartedAtAnnotationKey is added/updated or the Qdrant version needs to be upgraded.<br />This helps sharded but not replicated clusters to reduce downtime to a possible minimum during restart.<br />If unset, the operator is going to restart nodes concurrently if none of the collections if replicated. |  | Optional: \{\} <br /> |
+| `onDemandReplication` _[OnDemandReplicationType](#ondemandreplicationtype)_ | OnDemandReplication specifies the on-demand replication restart mode.<br />Off (default): Normal restart behavior. Pods are restarted directly.<br />Auto: The operator checks telemetry for non-replicated shards. If found, uses the recreate-node flow.<br />On: Always uses the recreate-node flow for eligible restart triggers. | Off | Enum: [Off Auto On] <br />Optional: \{\} <br /> |
 | `startupDelaySeconds` _integer_ | If StartupDelaySeconds is set (> 0), an additional 'sleep <value>' will be emitted to the pod startup.<br />The sleep will be added when a pod is restarted, it will not force any pod to restart.<br />This feature can be used for debugging the core, e.g. if a pod is in crash loop, it provided a way<br />to inspect the attached storage. |  | Optional: \{\} <br /> |
 | `rebalanceStrategy` _[RebalanceStrategy](#rebalancestrategy)_ | RebalanceStrategy specifies the strategy to use for automaticially rebalancing shards the cluster.<br />Cluster-manager needs to be enabled for this feature to work. |  | Enum: [by_count by_size by_count_and_size] <br />Optional: \{\} <br /> |
 | `readClusters` _[ReadCluster](#readcluster) array_ | ReadClusters specifies the read clusters for this cluster to synchronize.<br />Cluster-manager needs to be enabled for this feature to work. |  | Optional: \{\} <br /> |