Rancher-UI-problems after updating to v2.4.5

Hi. After updating our Rancher-HA-cluster to v2.4.5 (K8s v1.18.6) we observe some strange behaviors using the UI: marked a workload and deleted it has sometimes the deletion of the workload below the intended one as result; the consols often open with status ‘disconnected’; also after scaling down a workload and edit it, the wl scale up automatically (which is often not intended). It’s not sure if these behaviors are connected. Has anyone an idea where can be the cause / where can be the root-problem?