delay reconciliation of during restore (#438)

pkg/controller/replication/reconciler.go

@@ -40,22 +40,26 @@ func newReconciler(name string, mgr manager.Manager, controllerConfig config.DNS
 func (r *reconciler) Reconcile(ctx context.Context, req reconcile.Request) (reconcile.Result, error) {
 	result := reconcile.Result{}
 
+	ext, err := r.findExtension(ctx, req.Namespace)
+	if err != nil {
+		return result, err
+	}
+	if ext == nil || common.IsMigrating(ext) {
+		return result, nil
+	}
+	if common.IsPreparingRestore(ext) || common.IsRestoring(ext) {
+		return reconcile.Result{Requeue: true, RequeueAfter: 10 * time.Second}, nil
+	}
+
 	// ensure that only one DNSEntry is reconciled per extension (shoot) to avoid parallel conflicting updates
 	if !r.lock.TryLock(req.Namespace) {
 		r.Env.Info("delaying as namespace already locked", "namespace", req.Namespace, "entry", req.Name)
 		result.Requeue = true
-		result.RequeueAfter = wait.Jitter(1*time.Second, 0)
+		result.RequeueAfter = wait.Jitter(2*time.Second, 0.1)
 		return result, nil
 	}
 	defer r.lock.Unlock(req.Namespace)
 
-	ext, err := r.findExtension(ctx, req.Namespace)
-	if err != nil {
-		return result, err
-	}
-	if ext == nil || common.IsMigrating(ext) {
-		return result, nil
-	}
 	statehandler, err := common.NewStateHandler(ctx, r.Env, ext, false)
 	if err != nil {
 		return result, err