keep the 2m backoff on config apply failures while pods are waiting or the StatefulSet is not ready

internal/controller/bindcluster_controller.go

@@ -97,7 +97,7 @@ func (r *BindClusterReconciler) Reconcile(ctx context.Context, req ctrl.Request)
 		return ctrl.Result{}, err
 	}
 
-	if !ready {
+	if !ready && requeue == 0 {
 		requeue = requeueShort
 	}
 	if requeue > 0 {
@@ -462,7 +462,8 @@ func sha256Hex(s string) string {
 // reconfigPods runs rndc reconfig on each ready pod whose applied config is not
 // the desired one, once kubelet has propagated it to the pod's mount (eventually
 // consistent, up to ~90s). It records applied hashes in status and sets the
-// ConfigApplied condition, returning a requeue delay while any pod is pending.
+// ConfigApplied condition, returning a requeue delay while any pod is pending
+// (requeueLong if any pod failed, so waiting pods cannot hot-loop the failure).
 func (r *BindClusterReconciler) reconfigPods(ctx context.Context, c *bindv1alpha1.BindCluster, primaryConf, secondaryConf, keysConf string) time.Duration {
 	if r.Exec == nil {
 		return 0
@@ -526,8 +527,7 @@ func (r *BindClusterReconciler) reconfigPods(ctx context.Context, c *bindv1alpha
 	if len(failed)+len(execErrs) > 0 {
 		requeue = requeueLong
 		log.FromContext(ctx).Info("config apply failed", "cluster", c.Name, "reason", cond.Reason, "errors", cond.Message)
-	}
-	if len(waiting) > 0 {
+	} else if len(waiting) > 0 {
 		requeue = requeueShort
 	}
 	meta.SetStatusCondition(&c.Status.Conditions, cond)