Instances that do not go offline and stop in time within the heartbeat detection period

[xiaolianghu]

心跳检测周期内，未及时下线停止的实例

Nacos currently supports temporary instances to maintain activity by heartbeat reporting. The default interval for sending a heartbeat is 5 seconds. Nacos server will set the instance as unhealthy after 15 seconds without receiving a heartbeat, and remove the temporary instance after 30 seconds without receiving a heartbeat.
If the current instance has been stopped but is within the heartbeat detection cycle, nacos returns the stopped instance, causing an error.Is there any way to avoid it?

Nacos 目前支持临时实例使用心跳上报方式维持活性，发送心跳的周期默认是 5 秒，Nacos 服务端会在 15 秒没收到心跳后将实例设置为不健康，在 30 秒没收到心跳时将这个临时实例摘除。
如果当前实例已经停止，但是刚好处于心跳检测周期内，那么nacos会返回这个已停止的实例，导致错误。有什么办法可以避免吗？

[shengbinxu]

这个问题，主要涉及到jvm shutdown hook、spring 优雅停机等相关知识。
我写了个demo（参考项目中readme），来演示“如何在基于nacos做服务发现的微服务项目中实现优雅停止”

https://github.com/shengbinxu/GracefulShutdown