【服务离线】记一次组件引入造成actuator下线问题

4,277 阅读2分钟

SpringBoot Admin可以看作是Spring actuator 端点监控的一个UI界面,可以很方便的查看服务的运行情况。

部分服务不在线

如上图,在项目整合了部分组件后,出现状态显示不在线,但应用还是能够正常的对外提供服务。

  • 我们知道 应用展示的在线状态是读取的 Spring actuator health endpoint数据
curl http://127.0.0.1:5002/actuator/health

{"status":"DOWN"}

  • 此时无法通过 SpringBoot Admin 无法获取更多的细节, 建议开发环境开启端点的详细信息。
management:
endpoints:
web:
exposure:
include: '*'
endpoint:
health:
show-details: always #dev open
  • 再次尝试访问, 输出了全部组件的状态,可以明确的看到由于 sentinel 状态不对影响全局状态为 down
curl http://127.0.0.1:5002/actuator/health

{"status":"DOWN","components":{"discoveryComposite":{"status":"UP","components":{"discoveryClient":{"status":"UP","details":{"services":["pigx-upms-biz","pigx-tx-manager","pigx-daemon-elastic-job","pigx-mp-platform","pigx-daemon-quartz","pigx-gateway","pigx-auth","pigx-pay-platform","pigx-codegen","pigx-oa-platform","pigx-monitor"]}}}},"diskSpace":{"status":"UP","details":{"total":42927636480,"free":21334122496,"threshold":10485760}},"nacosConfig":{"status":"UP"},"nacosDiscovery":{"status":"UP"},"ping":{"status":"UP"},"reactiveDiscoveryClients":{"status":"UP","components":{"Simple Reactive Discovery Client":{"status":"UP","details":{"services":[]}}}},"redis":{"status":"UP","details":{"version":"5.0.7"}},"refreshScope":{"status":"UP"},"sentinel":{"status":"DOWN","details":{"dataSource":{},"enabled":true,"dashboard":{"description":"pigx-sentinel:5020 can't be connected","status":"DOWN"}}}}}

sentinel dashboadr can't be connected

  • SentinelHealthIndicator.doHealthCheck 来看下 sentinel 的健康检查是如何处理的
protected void doHealthCheck(Health.Builder builder) throws Exception {
Map<String, Object> detailMap = new HashMap<>();

// 获取心跳发送器
HeartbeatSender heartbeatSender = HeartbeatSenderProvider
.getHeartbeatSender();
// 发送心跳
boolean result = heartbeatSender.sendHeartbeat();
if (result) {
// 成功标记为UP
detailMap.put("dashboard", Status.UP);
}
else {
// 失败 标记为down
dashboardUp = false;
detailMap.put("dashboard", new Status(Status.DOWN.getCode(),
consoleServer + " can't be connected"));
}
}

// If Dashboard and DataSource are both OK, the health status is UP
if (dashboardUp && dataSourceUp) {
builder.up().withDetails(detailMap);
}
else {
builder.down().withDetails(detailMap);
}
}

  • SimpleHttpHeartbeatSender.sendHeartbeat 心跳发送器的处理逻辑
public boolean sendHeartbeat() throws Exception {

// 1. 判断客户端是和服务端是否交互 ,只有交互时dashboard 才有数据展示
if (TransportConfig.getRuntimePort() <= 0) {
RecordLog.info("[SimpleHttpHeartbeatSender] Runtime port not initialized, won't send heartbeat");
return false;
}
InetSocketAddress addr = getAvailableAddress();
if (addr == null) {
return false;
}

// 2. 访问dashboard 的接口
SimpleHttpRequest request = new SimpleHttpRequest(addr, TransportConfig.getHeartbeatApiPath());
request.setParams(heartBeat.generateCurrentMessage());
SimpleHttpResponse response = httpClient.post(request);
if (response.getStatusCode() == OK_STATUS) {
return true;
}
return false;
}
  • 到底是上文代码 第一步未交互被判断为down、还是第二步访问出错判断为down

查看 sentinel client 的日志

cat ~/csp/sentinel-record.log 

2020-03-23 13:54:04.634 INFO [SimpleHttpHeartbeatSender] Runtime port not initialized, won't send heartbeat

开始是第一步没有初始化交互造成,服务判断为 down

sentinel 如何初始化交互

  • 初始化交互端口 commandCenter.start()
public class CommandCenterInitFunc implements InitFunc {

@Override
public void init() throws Exception {
CommandCenter commandCenter = CommandCenterProvider.getCommandCenter();

if (commandCenter == null) {
RecordLog.warn("[CommandCenterInitFunc] Cannot resolve CommandCenter");
return;
}

commandCenter.beforeStart();
// 开始初始化端口交互
commandCenter.start();
RecordLog.info("[CommandCenterInit] Starting command center: "
+ commandCenter.getClass().getCanonicalName());
}
}
  • 配置启动立即加载初始化,而不是有流量请求再初始化
public class SentinelAutoConfiguration {

@Value("${project.name:${spring.application.name:}}")
private String projectName;

@Autowired
private SentinelProperties properties;

@PostConstruct
private void init() {

// earlier initialize
if (properties.isEager()) {
InitExecutor.doInit();
}

}
spring:
application:
name: @artifactId@
cloud:
sentinel:
eager: true
『★★★★★』 基于Spring Boot 2.2、 Spring Cloud Hoxton & Alibaba、 OAuth2 的RBAC 权限管理系统

image