Old IPs still available after NSE reconfigured

[Chunosov]

When running local-nse-death example we reconfigure NSE for different CIDR prefix. Example suggests pinging new IP address 172.16.1.102. In the same time, the old address 172.16.1.100 still available and ping passes.

If we shell into NSE pod

NSE=$(kubectl get pods -l app=nse-kernel -n ${NAMESPACE} --template '{{range .items}}{{.metadata.name}}{{"\n"}}{{end}}')
kubectl exec -it ${NSE} -n ${NAMESPACE} -- sh

we can see that its interface has two addresses assigned

ip addr
# ...
4: icmp-respo-9b77: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9000 qdisc mq state UNKNOWN qlen 1000
    link/ether 02:fe:8c:fd:17:34 brd ff:ff:ff:ff:ff:ff
    inet 172.16.1.100/32 scope global icmp-respo-9b77
       valid_lft forever preferred_lft forever
    inet 172.16.1.102/32 scope global icmp-respo-9b77
       valid_lft forever preferred_lft forever

The old address 172.16.1.100/32 still available even though it doesn't satisfy newly configured CIDR prefix 172.16.1.102/31