after ccm reboot, rollingupdate service delay 2 minutes to watch endpoint change

[monkeydzhao]

What happened: reboot ccm pod, and rollingupdate a service, endpoint change in 10 sec,but ccm watch endpoint change after 2 min and cause traffic fail

What you expected to happen: ccm instantly watch endpoint change

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

1. create a clb on alicloud console
2. create a service on k8s and change to loadbalancer type and set the clb id to service
3. reboot ccm daemonsets
4. rollingupdate service
5. service endpoint changes but clb doesn't update backend server
6. after 2 minutes the ccm watch endpoint change and update clb backend server

Anything else we need to know?:

Environment:

• Kubernetes version (use kubectl version): 1.24
• Cloud provider or hardware configuration: alicloud
• OS (e.g: cat /etc/os-release): Alma 8.8
• Kernel (e.g. uname -a): 4.18.0-477
• Install tools: yaml
• Network plugin and version (if this is a network-related bug): terway
• Others: