NamingEventListener通知时延较长

[007idle]

[version]
nacos-sdk = { version = "0.5.2", features = ["default"] }

[question]

现象：nacos客户端实现了NamingEventListener，在使用过程中发现，注册到nacos-server的服务实例，在下线后，NamingEventListener 至少要1-2分钟才能收到事件。

诉求： 有没有好的方式，能尽快收到服务实例变更的事件。

备注： 从nacos-sdk日志的日志中nacos_sdk::naming::observable::service_info_observable 是很快（大概几秒）就收到变更事件。但是看了源代码发现是外部crate调用不了，只能nacos-sdk内部调用

[log]
2026-03-17 15:02:11.275  INFO nacos_sdk::naming::handler::naming_push_request_handler: 38: receive NotifySubscriberRequest from nacos server: NotifySubscriberRequest { service_info: ServiceInfo { name: "pile-cloud-order", group_name: "DEFAULT_GROUP", clusters: "", cache_millis: 10000, last_ref_time: 1773730931274, checksum: "", all_ips: Some(false), all_ips_3x: None, reach_protection_threshold: false, hosts: Some([]) }, headers: {}, request_id: Some("38"), namespace: Some("pile-cloud"), service_name: Some("pile-cloud-order"), group_name: Some("DEFAULT_GROUP") }
2026-03-17 15:02:11.276  WARN emit: nacos_sdk::naming::observable::service_info_observable: 201: empty or error push: ServiceInfo { name: "pile-cloud-order", group_name: "DEFAULT_GROUP", clusters: "", cache_millis: 10000, last_ref_time: 1773730931274, checksum: "", all_ips: Some(false), all_ips_3x: None, reach_protection_threshold: false, hosts: Some([]) }

rror push: ServiceInfo { name: "pile-cloud-order", group_name: "DEFAULT_GROUP", clusters: "pile-cloud", cache_millis: 10000, last_ref_time: 1773730967350, checksum: "", all_ips: Some(false), all_ips_3x: None, reach_protection_threshold: false, hosts: Some([]) }
2026-03-17 15:03:00.463  INFO nacos_sdk::naming::handler::naming_push_request_handler: 38: receive NotifySubscriberRequest from nacos server: NotifySubscriberRequest { service_info: ServiceInfo { name: "pile-cloud-order", group_na
me: "DEFAULT_GROUP", clusters: "", cache_millis: 10000, last_ref_time: 1773730980461, checksum: "", all_ips: Some(false), all_ips_3x: None, reach_protection_threshold: false, hosts: Some([ServiceInstance { instance_id: Some("192.1
68.91.1#6334"), ip: "192.168.91.1", port: 6334, weight: 1.0, healthy: true, enabled: true, ephemeral: true, cluster_name: Some("pile-cloud"), service_name: Some("DEFAULT_GROUP@@pile-cloud-order"), metadata: {} }]) }, headers: {}, request_id: Some("39"), namespace: Some("pile-cloud"), service_name: Some("pile-cloud-order"), group_name: Some("DEFAULT_GROUP") }
2026-03-17 15:03:00.463  INFO emit: nacos_sdk::naming::observable::service_info_observable: 303: new ips(1) service: DEFAULT_GROUP@@pile-cloud-order -> [{"instanceId":"192.168.91.1#6334","ip":"192.168.91.1","port":6334,"weight":1.0,"healthy":true,"enabled":true,"ephemeral":true,"clusterName":"pile-cloud","serviceName":"DEFAULT_GROUP@@pile-cloud-order","metadata":{}}]
2026-03-17 15:03:00.464  INFO emit: nacos_sdk::naming::observable::service_info_observable: 314: removed ips(1) service: DEFAULT_GROUP@@pile-cloud-order -> [{"instanceId":"192.168.91.1#7705","ip":"192.168.91.1","port":7705,"weight":1.0,"healthy":true,"enabled":true,"ephemeral":true,"clusterName":"pile-cloud","serviceName":"DEFAULT_GROUP@@pile-cloud-order","metadata":{}}]
2026-03-17 15:03:00.464  INFO emit: nacos_sdk::naming::observable::service_info_observable: 215: current ips:(1) service: DEFAULT_GROUP@@pile-cloud-order -> [{"instanceId":"192.168.91.1#6334","ip":"192.168.91.1","port":6334,"weight":1.0,"healthy":true,"enabled":true,"ephemeral":true,"clusterName":"pile-cloud","serviceName":"DEFAULT_GROUP@@pile-cloud-order","metadata":{}}]
2026-03-17 15:03:00.465  WARN nacos_sdk::naming::observable::service_info_observable: 79: the key "DEFAULT_GROUP@@pile-cloud-order" is not subscribed.
 service: DEFAULT_GROUP@@pile-cloud-order@@pile-cloud -> [{"instanceId":"192.168.91.1#6334","ip":"192.168.91.1","port":6334,"weight":1.0,"healthy":true,"enabled":true,"ephemeral":true,"clusterName":"pile-cloud","serviceName":"DEFAULT_GROUP@@pile-cloud-order","metadata":{}}]
2026-03-17 15:03:47.415  INFO subscribe{client_id="naming:localhost:8848:pile-cloud:1"}:schedule_update{group_name="DEFAULT_GROUP" cluster="pile-cloud"}:emit: nacos_sdk::naming::observable::service_info_observable: 314: removed ip
s(1) service: DEFAULT_GROUP@@pile-cloud-order@@pile-cloud -> [{"instanceId":"192.168.91.1#7705","ip":"192.168.91.1","port":7705,"weight":1.0,"healthy":true,"enabled":true,"ephemeral":true,"clusterName":"pile-cloud","serviceName":"DEFAULT_GROUP@@pile-cloud-order","metadata":{}}]
2026-03-17 15:03:47.416  INFO subscribe{client_id="naming:localhost:8848:pile-cloud:1"}:schedule_update{group_name="DEFAULT_GROUP" cluster="pile-cloud"}:emit: nacos_sdk::naming::observable::service_info_observable: 215: current ip
s:(1) service: DEFAULT_GROUP@@pile-cloud-order@@pile-cloud -> [{"instanceId":"192.168.91.1#6334","ip":"192.168.91.1","port":6334,"weight":1.0,"healthy":true,"enabled":true,"ephemeral":true,"clusterName":"pile-cloud","serviceName":"DEFAULT_GROUP@@pile-cloud-order","metadata":{}}]
2026-03-17 15:03:47.416  INFO nacos_sdk::naming::observable::service_info_observable: 100: notify listener: "DEFAULT_GROUP@@pile-cloud-order@@pile-cloud", notify data: NamingChangeEvent { service_name: "pile-cloud-order", group_na
me: "DEFAULT_GROUP", clusters: "pile-cloud", instances: Some([ServiceInstance { instance_id: Some("192.168.91.1#6334"), ip: "192.168.91.1", port: 6334, weight: 1.0, healthy: true, enabled: true, ephemeral: true, cluster_name: Some("pile-cloud"), service_name: Some("DEFAULT_GROUP@@pile-cloud-order"), metadata: {} }]) }


2026-03-17 15:03:47.416  WARN subscribe{client_id="naming:localhost:8848:pile-cloud:1"}:schedule_update{group_name="DEFAULT_GROUP" cluster="pile-cloud"}:emit: common::load_balance: 39: [服务变更]:内容:Some([ServiceInstance { insta
nce_id: Some("192.168.91.1#6334"), ip: "192.168.91.1", port: 6334, weight: 1.0, healthy: true, enabled: true, ephemeral: true, cluster_name: Some("pile-cloud"), service_name: Some("DEFAULT_GROUP@@pile-cloud-order"), metadata: {} }])>>

[CherishCai]

你是说比较快的时间是 15:02:11.276 empty or error push 吗？这里是被防推空拦截了。如果你的服务实例全部下线了，会被拦截住，你可以关闭房推空，则应该就能通知到。

[007idle]

你是说比较快的时间是 15:02:11.276 empty or error push 吗？这里是被防推空拦截了。如果你的服务实例全部下线了，会被拦截住，你可以关闭房推空，则应该就能通知到。

对。你说的关闭房推空后，确实很快能推送通知。不好意思，我的问题写了一半，我真正想表达的是，服务实例下线之后又马上注册到nacos-server（控制台可以看到），然后NamingEventListener要等1-2分钟，才能收新实例的变更通知。这种情况下，如何尽快收到新实例的事件

[CherishCai]

看日志。是不是中间少了订阅器？

2026-03-17 15:03:00.465  WARN nacos_sdk::naming::observable::service_info_observable: 79: the key "DEFAULT_GROUP@@pile-cloud-order" is not subscribed.

变化应该都是会快速推下来的。

[007idle]

看日志。是不是中间少了订阅器？

2026-03-17 15:03:00.465  WARN nacos_sdk::naming::observable::service_info_observable: 79: the key "DEFAULT_GROUP@@pile-cloud-order" is not subscribed.

变化应该都是会快速推下来的。

我也注意到这个日志了。但是我看了源码，发现nacos_sdk::naming::observable::service_info_observable 是crate级别的。外部是使用不了。还是说有其他方式可以使用nacos_sdk::naming::observable::service_info_observable呢？

[CherishCai]

还真的没有非常深刻研究过这问题，不过完全下线再上线，就不该是经常发生的事

[CherishCai]

或者你先用自己的代码改动引入，打印更多日志辅助分析。若有必要的话请提交 PR 将日志信息丰富

[CherishCai]

我也注意到这个日志了。但是我看了源码，发现nacos_sdk::naming::observable::service_info_observable 是crate级别的。外部是使用不了。还是说有其他方式可以使用nacos_sdk::naming::observable::service_info_observable呢？

@007idle 对外只暴露了 api 模块的东西，所以外部没有其它方式的