Version: Nacos Server 2.0.2 Nacos Client 1.4.2
Problem: NacosClient一台节点在30s收到了3条推送通知 2023-02-16 21:14:45.602 第一条：收到order-pro服务的实例变更(服务下3台节点ip:1、2、3) 2023-02-16 21:15:05.782 第二条：收到order-pro服务的实例变更(服务下2台节点ip:1、2) 发现3节点下线，客户端做出相应处理 2023-02-16 21:15:15.682 第三条：收到order-pro服务的实例变更(服务下3台节点ip:1、2、3) 注：IP为3的节点在这15s内一直在正常运行，并未发生故障或重启
Nacos Client ip=3 naming.log日志(3节点实际IP是10.0.0.88，之所以用1,2,3代替是因为这个服务下面48台机器，所以用1,2,3代替)
2023-02-16 21:15:12.100 INFO [com.alibaba.nacos.naming.beat.sender:client.naming] [REGISTER-SERVICE] public registering service DEFAULT_GROUP@@order-pro with instance: Instance{instanceId='null', ip='10.0.0.88', port=7001, weight=1.0, healthy=true, enabled=true, ephemeral=true, clusterName='DEFAULT', serviceName='DEFAULT_GROUP@@order-pro', metadata={all=512, app_registry_tag=a6ea380c-b38f-4a35-88f6-4bfac2d17cd0, role=1, management.port=7002, priority=12, userName=admin, counterClusterId=1, okcoin.nacos.client=1.4.2, mode=1, management.endpoints.web.base-path=/, shards=[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31], port=10000, commitPositionCounterId=37, preserved.register.source=SPRING_CLOUD, status=NORMAL}} 2023-02-16 21:15:12.170 INFO [com.alibaba.nacos.naming.beat.sender:client.naming] [buildEnableParam] public service DEFAULT-DEFAULT_GROUP-DEFAULT_GROUP@@order-pro instance 10.0.0.88:7001 with enable: true 可以发现，3节点在2023-02-16 21:15:12.100向服务端注册实例。naming.log中在这之前未有任何异常。查看客户端代码，该日志只在服务启动、或者向服务端发送心跳时，发现自身IP实例不存在时，才会走注册流程
Nacos Server naming-server.log日志 2023-02-16 21:14:46,586 INFO Client connection 10.2.9.88:7001#true connect 2023-02-16 21:15:01,620 INFO Client connection 10.2.9.88:7001#true connect 2023-02-16 21:15:01,620 INFO Client remove for service Service{namespace='public', group='DEFAULT_GROUP', name='order-pro', ephemeral=true, revision=3666}, 10.2.9.88:7001#true 2023-02-16 21:15:04,627 INFO Client connection 10.2.9.88:7001#true disconnect, remove instances and subscribers 2023-02-16 21:15:14,049 INFO Client connection 10.2.9.888:7001#true connect 可以发现，Nacos Server在15:01s时，Client remove for service Service，将3节点IP移除的，并在15:04,627 remove instances and subscribers。
服务端日志和客户端日志相互印证，可知，服务端15:01 Client remove for service Service 15:14remove instances and subscribers。 我查看了服务端的代码，发现Client remove for service Service这个日志打印的方法是事件驱动的。所以我想知道，服务端在什么情况下会执行(Client remove for service Service) and (remove instances and subscribers) 流程。 并且一个运行正常的客户端为什么会被服务端认为 2023-02-16 21:15:01,620 INFO Client remove for service Service{namespace='public', group='DEFAULT_GROUP', name='order-pro', ephemeral=true, revision=3666}, 10.2.9.88:7001#true 2023-02-16 21:15:04,627 INFO Client connection 10.2.9.88:7001#true disconnect, remove instances and subscribers。