Istio is keep sending requests on dead IPs(Pods)

Hey, Can anyone help me with that issue we are keeping facing in production. I’m using a multicluster scenario ServiceA(Cluster A) ---> ServiceB(Cluster B). When A service sends concurrent requests to B service and somewhere b/t Service B replica down due to ungraceful node termination and spin new pod on the other node. At that time our services start failing because Service A is keeping sending requests on 118.16.70.134 but new service is now running on 118.16.70.137 any solution for that?

[2023-03-13T15:27:01.514Z] "GET / HTTP/1.1" 503 UF,URX upstream_reset_before_response_started{connection_failure,delayed_connect_error:_113} - "-" 0 145 9217 - "-" "curl/7.88.1-DEV" "daea83f0-cd24-49f7-9f86-474714e7dd73" "svc-service.namespace:3000" "118.16.70.134:3000" outbound|3000||svc-service.namespace - 240.240.0.27:3000 116.16.69.40:48462 - default