diff --git a/.github/workflows/deploy.yml b/.github/workflows/deploy.yml
index fa6ff3a..21b7251 100644
--- a/.github/workflows/deploy.yml
+++ b/.github/workflows/deploy.yml
@@ -41,14 +41,20 @@ jobs:
           echo "${{ secrets.KUBECONFIG_DATA }}" > ./kind_kubeconfig
           chmod 600 ./kind_kubeconfig
 
-      # Bereinigt alte Reste und zündet das dezentrale Terraform-Zentral-Binary
       - name: Terraform Deploy
         run: |
-          docker exec local-k3s-platform kubectl delete deployment cloudmon service cloudmon-service ingress-v1 cloudmon-ingress -n default --ignore-not-found=true 2>/dev/null || true
+          # DIE RADIKALE RETTUNG: Zwingt Kubernetes, die Altlasten SOFORT und rückstandslos zu löschen!
+          docker exec local-k3s-platform kubectl delete ingress cloudmon-ingress -n default --ignore-not-found=true --force --grace-period=0 2>/dev/null || true
+          docker exec local-k3s-platform kubectl delete service cloudmon-service -n default --ignore-not-found=true --force --grace-period=0 2>/dev/null || true
+          docker exec local-k3s-platform kubectl delete deployment cloudmon -n default --ignore-not-found=true --force --grace-period=0 2>/dev/null || true
+          
+          # Kurze Atempause für die Cluster-API (2 Sekunden)
+          sleep 2
           
           terraform init
           terraform apply -auto-approve -var="app_version=${{ github.sha }}"
 
+
       - name: Verify Dashboard Health
         run: |
           echo "Warte auf die cloudmon-Pods und das Ingress-Routing..."