Spring Boot负载均衡：实现故障服务器自动忽略与切换

一、添加服务实例健康校验逻辑 #

你目前的实现是静态返回所有ServiceInstance，没有判断实例是否存活，导致故障节点仍会被选中调用。需要在返回实例前完成健康检查，过滤不可用节点：

@Autowired
private WebClient webClient;

@Override
public Flux<List<ServiceInstance>> get() {
    List<ServiceInstance> allInstances = Arrays.asList(
        new DefaultServiceInstance(serviceId + "1", serviceId, "localhost", 8080, false),
        new DefaultServiceInstance(serviceId + "2", serviceId, "localhost", 9092, false),
        new DefaultServiceInstance(serviceId + "3", serviceId, "localhost", 9999, false)
    );
    
    // 异步校验每个实例健康状态，过滤故障节点
    return Flux.fromIterable(allInstances)
        .flatMap(instance -> {
            String healthCheckUrl = String.format("http://%s:%s/actuator/health", instance.getHost(), instance.getPort());
            return webClient.get()
                .uri(healthCheckUrl)
                .retrieve()
                .toBodilessEntity()
                .map(res -> instance)
                .onErrorResume(ex -> Mono.empty()); // 校验失败则丢弃实例
        })
        .collectList();
}

注意：目标服务需开启Spring Boot Actuator的/actuator/health端点；若无Actuator，可替换为服务的核心业务接口作为健康校验地址。

二、配置负载均衡重试机制 #

仅过滤健康节点还不够，需在调用失败时自动重试其他可用实例。基于Spring Cloud LoadBalancer的配置如下：

1. 引入依赖：

<dependency>
    <groupId>org.springframework.cloud</groupId>
    <artifactId>spring-cloud-starter-loadbalancer</artifactId>
</dependency>
<dependency>
    <groupId>org.springframework.retry</groupId>
    <artifactId>spring-retry</artifactId>
</dependency>

2. 配置文件（application.yml）中开启重试规则：

spring:
  cloud:
    loadbalancer:
      retry:
        enabled: true
      client:
        config:
          default:
            retry:
              max-retries-on-next-server: 2 # 切换至其他实例的重试次数
              max-retries-on-current-server: 0 # 当前实例不重试
              retryable-status-codes: 500,502,503,504 # 触发重试的HTTP状态码

若使用OpenFeign调用服务，可额外配置Feign重试：

feign:
  client:
    config:
      default:
        retryer:
          max-attempts: 3
          period: 100
          max-period: 500

三、优化：缓存健康实例+定时更新 #

每次请求都做健康校验会影响性能，可通过定时任务维护健康实例缓存：

private List<ServiceInstance> healthyInstances = new ArrayList<>();
private final ScheduledExecutorService scheduler = Executors.newSingleThreadScheduledExecutor();

@PostConstruct
public void initHealthCheckTask() {
    // 每30秒刷新一次健康实例列表
    scheduler.scheduleAtFixedRate(this::refreshHealthyInstances, 0, 30, TimeUnit.SECONDS);
}

private void refreshHealthyInstances() {
    List<ServiceInstance> allInstances = Arrays.asList(
        new DefaultServiceInstance(serviceId + "1", serviceId, "localhost", 8080, false),
        new DefaultServiceInstance(serviceId + "2", serviceId, "localhost", 9092, false),
        new DefaultServiceInstance(serviceId + "3", serviceId, "localhost", 9999, false)
    );
    
    List<ServiceInstance> tempHealthy = new ArrayList<>();
    for (ServiceInstance instance : allInstances) {
        try {
            String healthUrl = String.format("http://%s:%s/actuator/health", instance.getHost(), instance.getPort());
            webClient.get().uri(healthUrl).retrieve().toBodilessEntity().block();
            tempHealthy.add(instance);
        } catch (Exception ignored) {
            // 实例不可用，跳过
        }
    }
    this.healthyInstances = tempHealthy;
}

@Override
public Flux<List<ServiceInstance>> get() {
    return Flux.just(healthyInstances);
}

四、兜底处理：避免全节点故障导致的500 #

当所有实例都故障时，可添加降级逻辑，返回兜底实例或友好提示：

@Override
public Flux<List<ServiceInstance>> get() {
    if (healthyInstances.isEmpty()) {
        // 返回降级实例（需提前部署降级服务）
        return Flux.just(Collections.singletonList(
            new DefaultServiceInstance(serviceId + "-fallback", serviceId, "localhost", 8081, false)
        ));
    }
    return Flux.just(healthyInstances);
}

内容的提问来源于stack exchange，提问作者Mario Alejandro Ortiz Vargas