Changing specified IP of stateful workload will cause multiple IPs being reserved

[sjtufl]

Bug Report

Type: bug report

What happened

1. Specify ip for StatefulSet pod by adding annotation networking.alibaba.com/ip-pool: 100.64.99.111 (sts replica is set to 1 for demo)
2. Change annotation value to 100.64.99.112, new pod will be created with ip 100.64.99.112, and the previously specified 100.64.99.111 is also reserved, which is unintended.

~ # kubectl get pods -owide | grep 100.64.99.112
statefulset-e7v2c-0                             1/1     Running   0          97s     100.64.99.112
~ # kubectl get ipinstance | grep statefulset
100-64-99-111   100.64.99.111/16             statefulset-e7v2c-0                                          init     init
100-64-99-112   100.64.99.112/16             statefulset-e7v2c-0                izbp149b3nd59ljk7lbea7z   init     init

3. If we delete pod now, multiple reserved IPs will cause logic disruptions on inference of ip family when doing stateful allocate for newly created pod.

unable to stateful allocate with provider [hybridnet]: unable to assign: must assign two IPs when DualStack family, but [{ 100.64.99.112}]

What you expected to happen

Do not reserve previously specified IPs.

How to reproduce it (as minimally and precisely as possible)

Anything else we need to know?

Environment

• hybridnet version:
• OS (e.g. cat /etc/os-release):
• Kernel (e.g. uname -a):
• Kubernetes version:
• Install tools:
• Others: