How Nacos Registers Services: A Deep Dive into the Registration Flow and Core Code

1. Introduction

In microservice architecture a service registry is a centralized component that stores service instances and metadata, enabling services to discover each other via HTTP.

2. Nacos Overview

Nacos provides dynamic service discovery, configuration, metadata and traffic management, facilitating agile development of cloud‑native microservices.

Nacos Basic Architecture

3. Nacos Registration Center

The Nacos registration center stores service instances in a database. When a service starts it registers itself, and when it stops it deregisters. The registry can perform health‑check calls to verify instance availability.

Nacos Service Registry Structure

Map<namespace, Map<group::serviceName, Service>>

4. Demo Setup

Build a Nacos client project with Spring Boot 2.1.0 and Spring Cloud Greenwich.

<spring-boot-version>2.1.0.RELEASE</spring-boot-version>
<spring-cloud.version>Greenwich.RELEASE</spring-cloud.version>

Add spring-cloud-starter-alibaba-nacos-discovery dependency and annotate the main class with @EnableDiscoveryClient. Running the application registers the service in the Nacos console.

5. Detailed Principle

After the demo works, the source code is examined. The client registers via spring-cloud-starter-alibaba-nacos-discovery, which triggers NacosDiscoveryAutoConfiguration. The NacosAutoServiceRegistration constructor calls the parent AbstractAutoServiceRegistration, which registers the service when the web server starts.

public NacosAutoServiceRegistration(ServiceRegistry<Registration> serviceRegistry,
      AutoServiceRegistrationProperties autoServiceRegistrationProperties,
      NacosRegistration registration) {
    super(serviceRegistry, autoServiceRegistrationProperties);
    this.registration = registration;
}

The start() method creates a BeatInfo for temporary instances and sends periodic heartbeats.

public void start() {
    if (!isEnabled()) { return; }
    if (!this.running.get()) {
        context.publishEvent(new InstancePreRegisteredEvent(this, getRegistration()));
        register();
        if (shouldRegisterManagement()) { registerManagement(); }
        context.publishEvent(new InstanceRegisteredEvent<>(this, getConfiguration()));
        this.running.compareAndSet(false, true);
    }
}

The client ultimately calls

com.alibaba.cloud.nacos.registry.NacosServiceRegistry#register

, which builds an Instance and invokes NacosNamingService.registerInstance. The server side processes the request in InstanceController.register() and beat(), creating or updating Service and Instance objects, storing them in an in‑memory map serviceMap (Map<namespace, Map<group::serviceName, Service>>).

Instance registration triggers a series of consistency operations: DelegateConsistencyServiceImpl.put → DistroConsistencyServiceImpl.put → onPut → taskDispatcher.addTask. The task dispatcher batches keys and synchronizes them across the cluster via DataSyncer.submit, which serializes the data and sends it to peer nodes using NamingProxy.syncData (HTTP PUT to /v1/ns/distro/datum).

On the server side, Service#onChange updates the instance list, validates weights, and refreshes the IP map. Service#updateIPs and Cluster#updateIPs apply the changes, handling new, updated, and dead instances, and notifying push services.

6. Summary

The analysis shows that Nacos service registration relies on heartbeat and health‑check tasks, asynchronous cluster synchronization, and copy‑on‑write data structures to achieve high concurrency and availability in a microservice environment.