chore(operator): bump helm chart version to v1.5.4

.github/workflows/ci.yml

@@ -15,7 +15,7 @@ jobs:
   build-operator:
     runs-on: ubuntu-latest
     env:
-      IMG: ghcr.io/dragonflydb/operator:${{ github.sha }}
+      IMG: ghcr.io/mvasilenko/dragonfly-operator:${{ github.sha }}
 
     steps:
       - uses: actions/checkout@v6
@@ -49,7 +49,7 @@ jobs:
     needs: build-operator
     if: github.event_name == 'push' && startsWith(github.ref, 'refs/tags/')
     env:
-      IMG: ghcr.io/dragonflydb/operator:${{ github.ref_name }}
+      IMG: ghcr.io/mvasilenko/dragonfly-operator:${{ github.ref_name }}
       VERSION: ${{ github.ref_name }}
     steps:
       - uses: actions/checkout@v6
@@ -84,15 +84,15 @@ jobs:
 
           helm package charts/dragonfly-operator
 
-          helm push dragonfly-operator-${{ env.VERSION }}.tgz oci://ghcr.io/dragonflydb/dragonfly-operator/helm
+          helm push dragonfly-operator-${{ env.VERSION }}.tgz oci://ghcr.io/mvasilenko/dragonfly-operator/helm
 
       - name: Build and Publish image into GHCR
         uses: docker/build-push-action@v7
         with:
           context: .
           file: ./Dockerfile
           push: true
-          tags: ghcr.io/dragonflydb/operator:${{ github.ref_name }}
+          tags: ghcr.io/mvasilenko/dragonfly-operator:${{ github.ref_name }}
           platforms: |
             linux/amd64
             linux/arm64
@@ -112,6 +112,6 @@ jobs:
           body: |
             Release ${{ github.ref_name }}
 
-            Docker image: ghcr.io/dragonflydb/operator:${{ github.ref_name }}
-            Helm chart: oci://ghcr.io/dragonflydb/dragonfly-operator/helm
+            Docker image: ghcr.io/mvasilenko/dragonfly-operator:${{ github.ref_name }}
+            Helm chart: oci://ghcr.io/mvasilenko/dragonfly-operator/helm
           draft: true

README.md

@@ -91,6 +91,39 @@ kubectl patch dragonfly dragonfly-sample --type merge -p '{"spec":{"resources":{
 
 To add authentication to the dragonfly pods, you either set the `DFLY_requirepass` environment variable, or add the `--requirepass` argument.
 
+### Customising health-check probe scripts
+
+The operator generates default liveness, readiness, and startup probe scripts and mounts them via ConfigMaps. You can replace any probe with your own script using the `custom*ProbeConfigMap` fields.
+
+Scripts run inside the Dragonfly container and have access to `HEALTHCHECK_PORT` (admin port 9999 — no TLS, no auth).
+
+```yaml
+apiVersion: v1
+kind: ConfigMap
+metadata:
+  name: dragonfly-sample-probes
+  namespace: default
+data:
+  liveness-check.sh: |
+    #!/bin/sh
+    RESPONSE=$(timeout 4 redis-cli -h localhost -p ${HEALTHCHECK_PORT:-9999} PING 2>/dev/null)
+    case "$RESPONSE" in
+      PONG|*LOADING*) exit 0 ;;
+      *)              exit 1 ;;
+    esac
+---
+apiVersion: dragonflydb.io/v1alpha1
+kind: Dragonfly
+metadata:
+  name: dragonfly-sample
+spec:
+  replicas: 1
+  customLivenessProbeConfigMap:
+    name: dragonfly-sample-probes
+```
+
+> **Override precedence:** `spec.additionalVolumes` with a matching volume name (`liveness-probe`, `readiness-probe`, `startup-probe`) takes precedence over `custom*ProbeConfigMap`. Do not use both for the same probe.
+
 ### Deleting a Dragonfly instance
 
 To delete a Dragonfly instance, you can run

api/v1alpha1/dragonfly_types.go

@@ -196,6 +196,21 @@ type DragonflySpec struct {
 	// +optional
 	// +kubebuilder:validation:Optional
 	Pdb *PdbSpec `json:"pdb,omitempty"`
+
+	// (Optional) Custom ConfigMap with key "liveness-check.sh" to override the default liveness probe.
+	// +optional
+	// +kubebuilder:validation:Optional
+	CustomLivenessProbeConfigMap *corev1.LocalObjectReference `json:"customLivenessProbeConfigMap,omitempty"`
+
+	// (Optional) Custom ConfigMap with key "readiness-check.sh" to override the default readiness probe.
+	// +optional
+	// +kubebuilder:validation:Optional
+	CustomReadinessProbeConfigMap *corev1.LocalObjectReference `json:"customReadinessProbeConfigMap,omitempty"`
+
+	// (Optional) Custom ConfigMap with key "startup-check.sh" to override the default startup probe.
+	// +optional
+	// +kubebuilder:validation:Optional
+	CustomStartupProbeConfigMap *corev1.LocalObjectReference `json:"customStartupProbeConfigMap,omitempty"`
 }
 
 // PdbSpec defines the desired state of the PodDisruptionBudget

charts/dragonfly-operator/Chart.yaml

@@ -15,10 +15,10 @@ type: application
 # This is the chart version. This version number should be incremented each time you make changes
 # to the chart and its templates, including the app version.
 # Versions are expected to follow Semantic Versioning (https://semver.org/)
-version: v1.5.0
+version: v1.5.4
 
 # This is the version number of the application being deployed. This version number should be
 # incremented each time you make changes to the application. Versions are not expected to
 # follow Semantic Versioning. They should reflect the version the application is using.
 # It is recommended to use it with quotes.
-appVersion: "v1.5.0"
+appVersion: "v1.5.4"

charts/dragonfly-operator/values.yaml

@@ -85,7 +85,7 @@ rbacProxy:
 
 manager:
   image:
-    repository: docker.dragonflydb.io/dragonflydb/operator
+    repository: ghcr.io/mvasilenko/dragonfly-operator
     pullPolicy: IfNotPresent
     # Overrides the image tag whose default is the chart appVersion.
     tag: ""

config/crd/bases/dragonflydb.io_dragonflies.yaml

@@ -4461,6 +4461,51 @@ spec:
                         type: string
                     type: object
                 type: object
+              customLivenessProbeConfigMap:
+                description: (Optional) Custom ConfigMap with key "liveness-check.sh"
+                  to override the default liveness probe.
+                properties:
+                  name:
+                    default: ""
+                    description: |-
+                      Name of the referent.
+                      This field is effectively required, but due to backwards compatibility is
+                      allowed to be empty. Instances of this type with an empty value here are
+                      almost certainly wrong.
+                      More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
+                    type: string
+                type: object
+                x-kubernetes-map-type: atomic
+              customReadinessProbeConfigMap:
+                description: (Optional) Custom ConfigMap with key "readiness-check.sh"
+                  to override the default readiness probe.
+                properties:
+                  name:
+                    default: ""
+                    description: |-
+                      Name of the referent.
+                      This field is effectively required, but due to backwards compatibility is
+                      allowed to be empty. Instances of this type with an empty value here are
+                      almost certainly wrong.
+                      More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
+                    type: string
+                type: object
+                x-kubernetes-map-type: atomic
+              customStartupProbeConfigMap:
+                description: (Optional) Custom ConfigMap with key "startup-check.sh"
+                  to override the default startup probe.
+                properties:
+                  name:
+                    default: ""
+                    description: |-
+                      Name of the referent.
+                      This field is effectively required, but due to backwards compatibility is
+                      allowed to be empty. Instances of this type with an empty value here are
+                      almost certainly wrong.
+                      More info: https://kubernetes.io/docs/concepts/overview/working-with-objects/names/#names
+                    type: string
+                type: object
+                x-kubernetes-map-type: atomic
               enableReplicationReadinessGate:
                 description: |-
                   (Optional) When enabled, adds a custom readiness gate to pods that prevents

config/manager/kustomization.yaml

@@ -4,5 +4,5 @@ apiVersion: kustomize.config.k8s.io/v1beta1
 kind: Kustomization
 images:
   - name: controller
-    newName: docker.dragonflydb.io/dragonflydb/operator
-    newTag: v1.5.0
+    newName: ghcr.io/mvasilenko/dragonfly-operator
+    newTag: v1.5.3

config/rbac/role.yaml

@@ -7,13 +7,7 @@ rules:
 - apiGroups:
   - ""
   resources:
-  - events
-  verbs:
-  - create
-  - patch
-- apiGroups:
-  - ""
-  resources:
+  - configmaps
   - pods
   - services
   verbs:
@@ -24,6 +18,13 @@ rules:
   - patch
   - update
   - watch
+- apiGroups:
+  - ""
+  resources:
+  - events
+  verbs:
+  - create
+  - patch
 - apiGroups:
   - ""
   resources:

e2e/dragonfly_controller_test.go

@@ -136,10 +136,9 @@ var _ = Describe("Dragonfly Lifecycle tests", Ordered, FlakeAttempts(3), func()
 			err := k8sClient.Create(ctx, &df)
 			Expect(err).To(BeNil())
 
-			// Wait until Dragonfly object is marked initialized
-			waitForDragonflyPhase(ctx, k8sClient, name, namespace, controller.PhaseResourcesCreated, 2*time.Minute)
-			waitForStatefulSetReady(ctx, k8sClient, name, namespace, 2*time.Minute)
-
+			// Wait for master election, then for all replicas to be Ready
+			waitForDragonflyPhase(ctx, k8sClient, name, namespace, controller.PhaseReady, 3*time.Minute)
+			waitForStatefulSetReady(ctx, k8sClient, name, namespace, 3*time.Minute)
 		})
 
 		var ss appsv1.StatefulSet
@@ -750,28 +749,32 @@ var _ = Describe("Dragonfly tiering test with single replica", Ordered, FlakeAtt
 			defer close(stopChan)
 			defer rc.Close()
 
+			// Defensive: FlakeAttempts retries re-run only this It, not the preceding
+			// "Should create successfully" It, so residual data from a prior failed attempt
+			// would leak into the pre-insert assertion below. Flush to start clean.
+			Expect(rc.FlushAll(ctx).Err()).To(BeNil())
+
+			// Poll until tiered counters drain to zero (stash drain is async).
+			Eventually(func() int64 {
+				info, _ := rc.Info(ctx, "tiered").Result()
+				n, _ := parseTieredEntriesFromInfo(info)
+				return n
+			}, 10*time.Second, 200*time.Millisecond).Should(Equal(int64(0)))
+
 			// Insert BIG value (>64B so it is eligible for tiering)
 			const size = 1 << 20 // 1 MiB
 			payload := make([]byte, size)
 			_, _ = rand.Read(payload)
 
-			infoStr, err := rc.Info(ctx, "tiered").Result()
-			Expect(err).To(BeNil())
-
-			entries, err := parseTieredEntriesFromInfo(infoStr)
-			Expect(err).To(BeNil())
-			Expect(entries).To(Equal(int64(0))) // make sure this matches your expectation
-
 			Expect(rc.Set(ctx, "foo", payload, 0).Err()).To(BeNil())
 
-			// Inserted one big key, tiered entries should be 1
-			infoStr, err = rc.Info(ctx, "tiered").Result()
-			Expect(err).To(BeNil())
-
-			fmt.Println("Tiered entried Info: ", infoStr)
-			entries, err = parseTieredEntriesFromInfo(infoStr)
-			Expect(err).To(BeNil())
-			Expect(entries).To(Equal(int64(1))) // make sure this matches your expectation
+			// Tiering stash is asynchronous — poll until the 1MB value has been stashed
+			// rather than reading INFO once immediately after SET (which races the stash).
+			Eventually(func() int64 {
+				info, _ := rc.Info(ctx, "tiered").Result()
+				n, _ := parseTieredEntriesFromInfo(info)
+				return n
+			}, 30*time.Second, 200*time.Millisecond).Should(Equal(int64(1)))
 
 			// Fetch and compare by size
 			data, err := rc.Get(ctx, "foo").Bytes()

internal/controller/dragonfly_controller.go

@@ -45,6 +45,7 @@ type DragonflyReconciler struct {
 //+kubebuilder:rbac:groups="",resources=services,verbs=get;list;watch;create;update;patch;delete
 //+kubebuilder:rbac:groups="",resources=pods,verbs=get;list;watch;create;update;patch;delete
 //+kubebuilder:rbac:groups="",resources=events,verbs=create;patch
+//+kubebuilder:rbac:groups="",resources=configmaps,verbs=get;list;watch;create;update;patch;delete
 //+kubebuilder:rbac:groups=networking.k8s.io,resources=networkpolicies,verbs=get;list;watch;create;update;patch;delete
 
 // Reconcile is part of the main kubernetes reconciliation loop which aims to
@@ -158,6 +159,7 @@ func (r *DragonflyReconciler) SetupWithManager(mgr ctrl.Manager) error {
 		For(&dfv1alpha1.Dragonfly{}, builder.WithPredicates(predicate.GenerationChangedPredicate{})).
 		Owns(&appsv1.StatefulSet{}, builder.MatchEveryOwner).
 		Owns(&corev1.Service{}, builder.MatchEveryOwner).
+		Owns(&corev1.ConfigMap{}, builder.MatchEveryOwner).
 		Owns(&networkingv1.NetworkPolicy{}, builder.MatchEveryOwner).
 		Named("Dragonfly").
 		Complete(r)

internal/controller/dragonfly_instance.go

@@ -745,6 +745,12 @@ func resourceSpecsEqual(desired, existing client.Object) bool {
 	if !reflect.DeepEqual(desired.GetLabels(), existing.GetLabels()) || !reflect.DeepEqual(desired.GetAnnotations(), existing.GetAnnotations()) {
 		return false
 	}
+	// ConfigMaps store content in .Data, not .Spec — compare Data directly.
+	if cmDesired, ok := desired.(*corev1.ConfigMap); ok {
+		if cmExisting, ok := existing.(*corev1.ConfigMap); ok {
+			return reflect.DeepEqual(cmDesired.Data, cmExisting.Data)
+		}
+	}
 	// Compare only the .Spec field using reflection
 	desiredV := reflect.ValueOf(desired).Elem()
 	existingV := reflect.ValueOf(existing).Elem()

internal/controller/util_test.go

@@ -43,9 +43,9 @@ func TestSelectMasterCandidate(t *testing.T) {
 	tests := []struct {
 		name      string
 		pods      []corev1.Pod
-		readyPods map[string]bool
+		readyPods map[string]bool  // pod names that are considered ready
 		offsets   map[string]int64 // per-pod offsets
-		wantName  string
+		wantName  string           // expected winner; "" means nil result
 	}{
 		{
 			name:     "no pods",

internal/resources/const.go

@@ -99,6 +99,24 @@ const (
 
 	OperatorControlPlaneLabelKey   = "control-plane"
 	OperatorControlPlaneLabelValue = "controller-manager"
+
+	// Probe ConfigMap suffixes — appended to df.Name
+	LivenessProbeConfigMapSuffix  = "liveness-probe"
+	ReadinessProbeConfigMapSuffix = "readiness-probe"
+	StartupProbeConfigMapSuffix   = "startup-probe"
+
+	// Script keys — must match the filename in the ConfigMap data
+	LivenessScriptKey  = "liveness-check.sh"
+	ReadinessScriptKey = "readiness-check.sh"
+	StartupScriptKey   = "startup-check.sh"
+
+	// ProbeMountPath is the directory where all probe scripts are mounted
+	ProbeMountPath = "/etc/dragonfly/probes"
+
+	// Volume names for the three probe ConfigMaps
+	LivenessProbeVolumeName  = "liveness-probe"
+	ReadinessProbeVolumeName = "readiness-probe"
+	StartupProbeVolumeName   = "startup-probe"
 )
 
 var DefaultDragonflyArgs = []string{