diff --git a/charts/openstack-cluster/templates/cluster-openstack.yaml b/charts/openstack-cluster/templates/cluster-openstack.yaml
index 2ac48c1..d0e7560 100644
--- a/charts/openstack-cluster/templates/cluster-openstack.yaml
+++ b/charts/openstack-cluster/templates/cluster-openstack.yaml
@@ -8,8 +8,7 @@ metadata:
     # We let Cluster API clean up this resource
     # Deleting it ourselves, which CAPI is not expecting, can cause some nasty race conditions
     helm.sh/resource-policy: keep
-    # Same, but for Argo
-    argocd.argoproj.io/sync-options: Prune=false
+    # NOTE: Argo won't delete this object itself as it has an owner reference to the cluster
 spec:
   identityRef:
     kind: Secret
diff --git a/charts/openstack-cluster/templates/control-plane/kubeadm-control-plane.yaml b/charts/openstack-cluster/templates/control-plane/kubeadm-control-plane.yaml
index 975b257..57ee9a2 100644
--- a/charts/openstack-cluster/templates/control-plane/kubeadm-control-plane.yaml
+++ b/charts/openstack-cluster/templates/control-plane/kubeadm-control-plane.yaml
@@ -74,8 +74,7 @@ metadata:
     # We let Cluster API clean up the control plane resource
     # Deleting it ourselves, which CAPI is not expecting, can cause some nasty race conditions
     helm.sh/resource-policy: keep
-    # Same, but for Argo
-    argocd.argoproj.io/sync-options: Prune=false
+    # NOTE: Argo won't delete this object itself as it has an owner reference to the cluster
 spec:
   version: v{{ .Values.kubernetesVersion | required ".Values.kubernetesVersion is required" | trimPrefix "v" }}
   replicas: {{ .Values.controlPlane.machineCount }}
diff --git a/charts/openstack-cluster/templates/control-plane/machine-health-check.yaml b/charts/openstack-cluster/templates/control-plane/machine-health-check.yaml
index aa1138a..c21d769 100644
--- a/charts/openstack-cluster/templates/control-plane/machine-health-check.yaml
+++ b/charts/openstack-cluster/templates/control-plane/machine-health-check.yaml
@@ -5,6 +5,9 @@ kind: MachineHealthCheck
 metadata:
   name: {{ include "openstack-cluster.componentName" (list . "control-plane") }}
   labels: {{ include "openstack-cluster.componentLabels" (list . "control-plane") | nindent 4 }}
+  annotations:
+    # Argo should manage this resource even though it has an owner ref for the cluster
+    argocd.argoproj.io/managed-resource: "true"
 spec:
   clusterName: {{ include "openstack-cluster.clusterName" . }}
   selector:
diff --git a/charts/openstack-cluster/templates/control-plane/openstack-machine-template.yaml b/charts/openstack-cluster/templates/control-plane/openstack-machine-template.yaml
index 2093159..aa5dcac 100644
--- a/charts/openstack-cluster/templates/control-plane/openstack-machine-template.yaml
+++ b/charts/openstack-cluster/templates/control-plane/openstack-machine-template.yaml
@@ -50,7 +50,6 @@ metadata:
     # Cluster API complains when old templates disappear before it has rolled all the machines over
     # When deploying with Helm, leave the resource behind and let Cluster API clean it up
     helm.sh/resource-policy: keep
-    # For Argo, we can do better and make sure it is removed only once the cluster becomes healthy again
-    argocd.argoproj.io/sync-options: PruneLast=true
+    # NOTE: Argo won't delete this object itself as it has an owner reference to the cluster
 spec:
   {{- include "openstack-cluster.controlplane.mt.spec" . | nindent 2 }}
diff --git a/charts/openstack-cluster/templates/node-group/kubeadm-config-template.yaml b/charts/openstack-cluster/templates/node-group/kubeadm-config-template.yaml
index 34045fc..165b19d 100644
--- a/charts/openstack-cluster/templates/node-group/kubeadm-config-template.yaml
+++ b/charts/openstack-cluster/templates/node-group/kubeadm-config-template.yaml
@@ -43,8 +43,7 @@ metadata:
     # Cluster API complains when old templates disappear before it has rolled all the machines over
     # When deploying with Helm, leave the resource behind and let Cluster API clean it up
     helm.sh/resource-policy: keep
-    # For Argo, we can do better and make sure it is removed only once the cluster becomes healthy again
-    argocd.argoproj.io/sync-options: PruneLast=true
+    # NOTE: Argo won't delete this object itself as it has an owner reference to the cluster
 spec:
   template:
     spec: {{ include "openstack-cluster.nodegroup.kct.spec" (list $ $nodeGroup) | nindent 6 }}
diff --git a/charts/openstack-cluster/templates/node-group/machine-deployment.yaml b/charts/openstack-cluster/templates/node-group/machine-deployment.yaml
index dfa2b08..0370ec9 100644
--- a/charts/openstack-cluster/templates/node-group/machine-deployment.yaml
+++ b/charts/openstack-cluster/templates/node-group/machine-deployment.yaml
@@ -24,6 +24,8 @@ metadata:
   {{- fail (printf "maximum machine count < minimum machine count for node group '%s'" $nodeGroup.name) }}
   {{- end }}
   annotations:
+    # Argo should manage this resource even though it has an owner ref for the cluster
+    argocd.argoproj.io/managed-resource: "true"
     cluster.x-k8s.io/cluster-api-autoscaler-node-group-min-size: {{ quote $machineCountMin }}
     cluster.x-k8s.io/cluster-api-autoscaler-node-group-max-size: {{ quote $machineCountMax }}
   {{- end }}
diff --git a/charts/openstack-cluster/templates/node-group/machine-health-check.yaml b/charts/openstack-cluster/templates/node-group/machine-health-check.yaml
index 1e23a64..5bee2d2 100644
--- a/charts/openstack-cluster/templates/node-group/machine-health-check.yaml
+++ b/charts/openstack-cluster/templates/node-group/machine-health-check.yaml
@@ -9,6 +9,9 @@ metadata:
   labels:
     {{- include "openstack-cluster.componentLabels" (list $ "worker") | nindent 4 }}
     {{ $.Values.projectPrefix }}/node-group: {{ $nodeGroup.name }}
+  annotations:
+    # Argo should manage this resource even though it has an owner ref for the cluster
+    argocd.argoproj.io/managed-resource: "true"
 spec:
   clusterName: {{ include "openstack-cluster.clusterName" $ }}
   selector:
diff --git a/charts/openstack-cluster/templates/node-group/openstack-machine-template.yaml b/charts/openstack-cluster/templates/node-group/openstack-machine-template.yaml
index 6e4029f..aebdfa0 100644
--- a/charts/openstack-cluster/templates/node-group/openstack-machine-template.yaml
+++ b/charts/openstack-cluster/templates/node-group/openstack-machine-template.yaml
@@ -59,7 +59,6 @@ metadata:
     # Cluster API complains when old templates disappear before it has rolled all the machines over
     # When deploying with Helm, leave the resource behind and let Cluster API clean it up
     helm.sh/resource-policy: keep
-    # For Argo, we can do better and make sure it is removed only once the cluster becomes healthy again
-    argocd.argoproj.io/sync-options: PruneLast=true
+    # NOTE: Argo won't delete this object itself as it has an owner reference to the cluster
 spec: {{ include "openstack-cluster.nodegroup.mt.spec" (list $ $nodeGroup) | nindent 2 }}
 {{- end }}
diff --git a/charts/openstack-cluster/templates/secret-cloud-config.yaml b/charts/openstack-cluster/templates/secret-cloud-config.yaml
index 811641d..aeb6783 100644
--- a/charts/openstack-cluster/templates/secret-cloud-config.yaml
+++ b/charts/openstack-cluster/templates/secret-cloud-config.yaml
@@ -13,8 +13,8 @@ metadata:
     # If the cloud credentials are deleted before the cluster has finished deleting, then the cluster
     # deletion cannot proceed any further. So prevent Helm from deleting it.
     helm.sh/resource-policy: keep
-    # Same, but for Argo
-    argocd.argoproj.io/sync-options: Prune=false
+    # For Argo, this resource should be pruned last out of everything
+    argocd.argoproj.io/sync-options: PruneLast=true
 stringData:
   # Just include the data for the cloud we will be using
   clouds.yaml: |