fix terminal addons

Author: rushmash91

apis/v1alpha1/ack-generate-metadata.yaml

@@ -1,8 +1,8 @@
 ack_generate_info:
-  build_date: "2025-07-22T21:59:20Z"
-  build_hash: b2dc0f44e0b08f041de14c3944a5cc005ba97c8f
+  build_date: "2025-08-01T19:06:21Z"
+  build_hash: fc0b91c88a7087d2acd52f8234337378b49d9c93
   go_version: go1.24.5
-  version: v0.50.0
+  version: v0.50.0-2-gfc0b91c
 api_directory_checksum: fecde869000735952a2e71d711fb5e64d9ee638e
 api_version: v1alpha1
 aws_sdk_go_version: v1.36.3

helm/templates/caches-role-binding.yaml

@@ -1,7 +1,7 @@
 apiVersion: rbac.authorization.k8s.io/v1
 kind: ClusterRoleBinding
 metadata:
-  name: {{ include "ack-eks-controller.app.fullname" . }}-namespace-caches
+  name: {{ include "ack-eks-controller.app.fullname" . }}-namespaces-cache
   labels:
     app.kubernetes.io/name: {{ include "ack-eks-controller.app.name" . }}
     app.kubernetes.io/instance: {{ .Release.Name }}
@@ -12,7 +12,7 @@ metadata:
 roleRef:
   kind: ClusterRole
   apiGroup: rbac.authorization.k8s.io
-  name: {{ include "ack-eks-controller.app.fullname" . }}-namespace-caches
+  name: {{ include "ack-eks-controller.app.fullname" . }}-namespaces-cache
 subjects:
 - kind: ServiceAccount
   name: {{ include "ack-eks-controller.service-account.name" . }}

pkg/resource/addon/hooks.go

@@ -52,8 +52,6 @@ var (
 var (
 	// TerminalStatuses defines the list of statuses that are terminal for an addon
 	TerminalStatuses = []string{
-		StatusCreateFailed,
-		StatusUpdateFailed,
 		StatusDeleteFailed,
 		// Still not sure if we should consider DEGRADED as terminal
 		// StatusDegraded,
@@ -101,6 +99,16 @@ func addonHasTerminalStatus(r *resource) bool {
 	return false
 }
 
+// addonInFailedState returns true if the supplied addon is in a failed state
+// that requires retry (CREATE_FAILED or UPDATE_FAILED)
+func addonInFailedState(r *resource) bool {
+	if r.ko.Status.Status == nil {
+		return false
+	}
+	cs := *r.ko.Status.Status
+	return cs == StatusCreateFailed || cs == StatusUpdateFailed
+}
+
 // requeueWaitUntilCanModify returns a `ackrequeue.RequeueNeededAfter` struct
 // explaining the addon cannot be modified until it reaches an active status.
 func requeueWaitUntilCanModify(r *resource) *ackrequeue.RequeueNeededAfter {

pkg/resource/addon/sdk.go

@@ -3,7 +3,11 @@
 		ackcondition.SetSynced(latest, corev1.ConditionFalse, &msg, nil)
 		return latest, requeueWaitWhileDeleting
 	}
-	if !addonActive(latest) {
+
+	// Check if addon is in a failed state that requires retry
+	inFailedState := addonInFailedState(latest)
+
+	if !addonActive(latest) && !inFailedState {
 		msg := "Addon is in '" + *latest.ko.Status.Status + "' status"
 		ackcondition.SetSynced(latest, corev1.ConditionFalse, &msg, nil)
 		if addonHasTerminalStatus(latest) {
@@ -13,16 +17,24 @@
 		return latest, requeueWaitUntilCanModify(latest)
 	}
 
+	// If addon is in failed state, we need to force an update regardless of delta
+	if inFailedState {
+		msg := "Addon is in '" + *latest.ko.Status.Status + "' status, attempting recovery"
+		ackcondition.SetSynced(latest, corev1.ConditionFalse, &msg, nil)
+	}
+
 	if delta.DifferentAt("Spec.Tags") {
 		err := syncTags(
-			ctx, rm.sdkapi, rm.metrics, 
-			string(*desired.ko.Status.ACKResourceMetadata.ARN), 
+			ctx, rm.sdkapi, rm.metrics,
+			string(*desired.ko.Status.ACKResourceMetadata.ARN),
 			aws.ToStringMap(desired.ko.Spec.Tags), aws.ToStringMap(latest.ko.Spec.Tags),
 		)
 		if err != nil {
 			return nil, err
 		}
 	}
-    if !delta.DifferentExcept("Spec.Tags"){
-        return desired, nil
-    }
+	// If addon is in failed state, always attempt update to recover
+	// Otherwise, check if there are differences to update
+	if !inFailedState && !delta.DifferentExcept("Spec.Tags") {
+		return desired, nil
+	}