Downtime once scale down for Kong ingress controller

Current i am using Kong ingress controller on GKE
follow for now: network layer 4 → Kong ingress → service ->pod
I have a problem when we want change config or scale up/scale down by HPA on k8s, so we lost some traffic in that time. can everyone helm me for this case? solution the best for issue?
Thank you

are there any help me in this case?