Spring Boot 2.7(K8s集群)选主机制失效求助：定时任务多实例均执行

在Kubernetes集群部署的Spring Boot 2.7多实例应用中，需实现仅由leader实例执行定时任务。采用Spring Cloud Kubernetes基于ConfigMap的选主机制，但未达预期：所有实例均执行定时任务，ConfigMap未更新leader元数据，且Pod日志无任何输出（含错误信息）。

已完成的配置步骤 #

• 依赖配置：在pom.xml中添加以下依赖

<dependency>
  <groupId>org.springframework.boot</groupId>
  <artifactId>spring-boot-starter-actuator</artifactId>
</dependency>
<dependency>
  <groupId>org.springframework.cloud</groupId>
  <artifactId>spring-cloud-kubernetes-fabric8-leader</artifactId>
  <version>3.0.2</version>
</dependency>

• 定时任务组件：创建ScheduledTasks类

@Component
public class ScheduledTasks {
    private static final Logger log = LoggerFactory.getLogger(ScheduledTasks.class);

    private static final SimpleDateFormat dateFormat = new SimpleDateFormat("HH:mm:ss");

    private ManagerContext managerContext;
    public ScheduledTasks(ManagerContext managerContext){
        this.managerContext = managerContext;
    }
    @Scheduled(fixedRate = 5000)
    public void reportCurrentTime() {

        if (isLeader()) {
            // Perform the task
            log.info("The time is now {}", dateFormat.format(new Date()));
        } else {
            // Not the leader instance, wait until becoming the leader
            log.info("I am not a leader");
        }

    }

    private boolean isLeader() {
        // Logic to check if this instance is the leader
        return managerContext.getContext() != null; // Modify the logic to check real leadership
    }
}

• 应用配置：在application.properties中添加

spring.cloud.kubernetes.leader.role=world
# Configmap to which leader election metadata will be saved
spring.cloud.kubernetes.leader.config-map-name=my-config-map

• 事件监听组件：实现OnGrantedEvent和OnRevokedEvent监听

/**
  * Handle a notification that this instance has become a leader.
  * @param event on granted event
  */
 @EventListener
 public void handleEvent(OnGrantedEvent event) {
  System.out.println(String.format("'%s' leadership granted", event.getRole()));
  this.context = event.getContext();
  managerContext.setContext(this.context);
 }

 /**
  * Handle a notification that this instance's leadership has been revoked.
  * @param event on revoked event
  */
 @EventListener
 public void handleEvent(OnRevokedEvent event) {
  System.out.println(String.format("'%s' leadership revoked", event.getRole()));
  this.context = null;
  managerContext.setContext(null);
 }

• 手动创建了应用所需的my-config-map ConfigMap
• 创建了Role和RoleBinding，确保应用使用的ServiceAccount具备ConfigMap的读写权限

排查与修复方案 #

1. 补全核心依赖 #

Spring Cloud Kubernetes Leader模块需要核心组件支撑，需在pom.xml中添加依赖管理与核心starter：

<dependencyManagement>
    <dependencies>
        <dependency>
            <groupId>org.springframework.cloud</groupId>
            <artifactId>spring-cloud-kubernetes-dependencies</artifactId>
            <version>3.0.2</version>
            <type>pom</type>
            <scope>import</scope>
        </dependency>
    </dependencies>
</dependencyManagement>

<dependency>
    <groupId>org.springframework.cloud</groupId>
    <artifactId>spring-cloud-starter-kubernetes-fabric8</artifactId>
</dependency>

2. 验证ConfigMap权限与存在性 #

• 确认my-config-map已在应用所在Namespace创建（无需提前填充数据，但必须存在）
• 检查Role权限是否包含ConfigMap的get/update/patch操作：

apiVersion: rbac.authorization.k8s.io/v1
kind: Role
metadata:
  name: leader-election-role
rules:
- apiGroups: [""]
  resources: ["configmaps"]
  resourceNames: ["my-config-map"]
  verbs: ["get", "update", "patch"]

• 确认RoleBinding已绑定到应用使用的ServiceAccount，且Namespace完全匹配

3. 修复事件监听类 #

• 给事件监听类添加@Component注解，确保被Spring容器扫描加载
• 替换System.out.println为日志输出，并修正方法签名避免重载冲突：

@Component
public class LeaderEventListener {
    private static final Logger log = LoggerFactory.getLogger(LeaderEventListener.class);
    private Context context;
    private final ManagerContext managerContext;

    public LeaderEventListener(ManagerContext managerContext) {
        this.managerContext = managerContext;
    }

    @EventListener
    public void handleGrantedEvent(OnGrantedEvent event) {
        log.info("'{}' leadership granted", event.getRole());
        this.context = event.getContext();
        managerContext.setContext(this.context);
    }

    @EventListener
    public void handleRevokedEvent(OnRevokedEvent event) {
        log.info("'{}' leadership revoked", event.getRole());
        this.context = null;
        managerContext.setContext(null);
    }
}

4. 确保状态容器为单例 #

ManagerContext需声明为Spring单例Bean，避免多实例状态不一致：

@Component
public class ManagerContext {
    private Context context;

    public Context getContext() {
        return context;
    }

    public void setContext(Context context) {
        this.context = context;
    }
}

同时修正isLeader()判断逻辑：

private boolean isLeader() {
    return managerContext.getContext() != null 
           && managerContext.getContext().isLeader();
}

5. 开启调试日志 #

在application.properties中添加日志配置，排查选主流程：

logging.level.org.springframework.cloud.kubernetes=INFO
logging.level.org.springframework.cloud.kubernetes.fabric8.leader=DEBUG

6. 验证Pod的K8s API访问权限 #

• 执行kubectl exec <pod-name> -- cat /var/run/secrets/kubernetes.io/serviceaccount/token，确认ServiceAccount token存在
• 在Pod内测试ConfigMap访问：

curl -k https://kubernetes.default.svc/api/v1/namespaces/<你的命名空间>/configmaps/my-config-map \
-H "Authorization: Bearer $(cat /var/run/secrets/kubernetes.io/serviceaccount/token)"

确保能正常返回ConfigMap数据，否则需检查RBAC配置。

内容的提问来源于stack exchange，提问作者Montassar Hamadi