健康检查

Salmon2026/1/15大约 4 分钟Consul健康检查Health Check

健康检查

一、健康检查概述

Consul 提供多种健康检查方式，确保只有健康的服务实例参与负载均衡。

1.1 检查类型

类型	说明	适用场景
HTTP	发送 HTTP 请求检查	Web 服务
TCP	TCP 连接检查	数据库、缓存
gRPC	gRPC 健康检查	gRPC 服务
Script	执行脚本检查	自定义检查
TTL	服务主动上报	特殊场景
Docker	Docker 容器检查	容器化服务

1.2 检查状态

passing：健康
warning：警告
critical：不健康

二、HTTP 检查

2.1 基本配置

{
  "service": {
    "name": "user-service",
    "port": 8080,
    "check": {
      "http": "http://localhost:8080/health",
      "method": "GET",
      "interval": "10s",
      "timeout": "5s"
    }
  }
}

2.2 带请求头的检查

{
  "check": {
    "http": "http://localhost:8080/health",
    "header": {
      "Authorization": ["Bearer token123"],
      "Content-Type": ["application/json"]
    },
    "interval": "10s"
  }
}

2.3 TLS 检查

{
  "check": {
    "http": "https://localhost:8443/health",
    "tls_skip_verify": false,
    "tls_server_name": "user-service.example.com",
    "interval": "10s"
  }
}

2.4 Spring Boot Actuator

# application.yml
management:
  endpoints:
    web:
      exposure:
        include: health
  endpoint:
    health:
      show-details: always

@Component
public class CustomHealthIndicator implements HealthIndicator {
    
    @Override
    public Health health() {
        // 自定义健康检查逻辑
        boolean isHealthy = checkDatabase();
        
        if (isHealthy) {
            return Health.up()
                .withDetail("database", "connected")
                .build();
        }
        return Health.down()
            .withDetail("database", "disconnected")
            .build();
    }
    
    private boolean checkDatabase() {
        // 检查数据库连接
        return true;
    }
}

三、TCP 检查

3.1 基本配置

{
  "service": {
    "name": "mysql",
    "port": 3306,
    "check": {
      "tcp": "localhost:3306",
      "interval": "10s",
      "timeout": "5s"
    }
  }
}

3.2 Redis 检查示例

{
  "service": {
    "name": "redis",
    "port": 6379,
    "check": {
      "tcp": "localhost:6379",
      "interval": "5s",
      "timeout": "2s"
    }
  }
}

四、gRPC 检查

4.1 配置方式

{
  "service": {
    "name": "grpc-service",
    "port": 9090,
    "check": {
      "grpc": "localhost:9090",
      "grpc_use_tls": false,
      "interval": "10s"
    }
  }
}

4.2 gRPC 健康检查实现

// 实现 gRPC 健康检查协议
public class HealthServiceImpl extends HealthGrpc.HealthImplBase {
    
    @Override
    public void check(HealthCheckRequest request,
                      StreamObserver<HealthCheckResponse> responseObserver) {
        HealthCheckResponse response = HealthCheckResponse.newBuilder()
            .setStatus(HealthCheckResponse.ServingStatus.SERVING)
            .build();
        responseObserver.onNext(response);
        responseObserver.onCompleted();
    }
}

五、Script 检查

5.1 基本配置

{
  "check": {
    "args": ["/usr/local/bin/check-service.sh"],
    "interval": "30s",
    "timeout": "10s"
  }
}

5.2 检查脚本示例

#!/bin/bash
# check-service.sh

# 检查进程是否存在
if pgrep -x "java" > /dev/null; then
    echo "Service is running"
    exit 0  # passing
else
    echo "Service is not running"
    exit 2  # critical
fi

5.3 退出码含义

退出码	状态
0	passing
1	warning
其他	critical

六、TTL 检查

6.1 配置方式

{
  "service": {
    "name": "batch-service",
    "check": {
      "ttl": "30s"
    }
  }
}

6.2 服务主动上报

# 上报健康状态
curl -X PUT http://localhost:8500/v1/agent/check/pass/service:batch-service

# 上报警告状态
curl -X PUT http://localhost:8500/v1/agent/check/warn/service:batch-service

# 上报失败状态
curl -X PUT http://localhost:8500/v1/agent/check/fail/service:batch-service

6.3 Java 实现

@Component
public class ConsulTTLHealthCheck {
    
    @Autowired
    private ConsulClient consulClient;
    
    @Scheduled(fixedRate = 10000)  // 每10秒上报一次
    public void reportHealth() {
        try {
            // 执行健康检查逻辑
            boolean healthy = performHealthCheck();
            
            if (healthy) {
                consulClient.agentCheckPass("service:batch-service");
            } else {
                consulClient.agentCheckFail("service:batch-service", "Health check failed");
            }
        } catch (Exception e) {
            consulClient.agentCheckFail("service:batch-service", e.getMessage());
        }
    }
    
    private boolean performHealthCheck() {
        // 自定义检查逻辑
        return true;
    }
}

七、多重检查

7.1 配置多个检查

{
  "service": {
    "name": "user-service",
    "port": 8080,
    "checks": [
      {
        "id": "http-check",
        "name": "HTTP Health Check",
        "http": "http://localhost:8080/health",
        "interval": "10s"
      },
      {
        "id": "tcp-check",
        "name": "TCP Port Check",
        "tcp": "localhost:8080",
        "interval": "5s"
      },
      {
        "id": "memory-check",
        "name": "Memory Check",
        "args": ["/scripts/check-memory.sh"],
        "interval": "30s"
      }
    ]
  }
}

7.2 检查聚合

所有检查都通过时，服务才被认为是健康的。

八、Spring Cloud Consul 健康检查

8.1 配置

spring:
  cloud:
    consul:
      discovery:
        health-check-path: /actuator/health
        health-check-interval: 10s
        health-check-timeout: 5s
        health-check-critical-timeout: 30s

8.2 自定义健康检查路径

spring:
  cloud:
    consul:
      discovery:
        health-check-url: http://${spring.cloud.client.ip-address}:${server.port}/custom-health

8.3 禁用健康检查

spring:
  cloud:
    consul:
      discovery:
        register-health-check: false

九、健康检查 API

9.1 查询服务健康状态

# 查询所有健康检查
curl http://localhost:8500/v1/health/checks/user-service

# 只返回健康的服务
curl "http://localhost:8500/v1/health/service/user-service?passing=true"

# 查询节点健康状态
curl http://localhost:8500/v1/health/node/node1

9.2 响应示例

[
  {
    "Node": "node1",
    "CheckID": "service:user-service",
    "Name": "Service 'user-service' check",
    "Status": "passing",
    "Output": "HTTP GET http://localhost:8080/health: 200 OK"
  }
]

十、最佳实践

10.1 检查间隔设置

10.2 超时设置

超时时间应小于检查间隔
建议：timeout = interval / 2

10.3 故障转移时间

故障转移时间 = 检查间隔 + 超时时间 + 服务发现缓存时间

10.4 检查建议

场景	推荐检查类型	间隔
Web 服务	HTTP	10s
数据库	TCP	10s
消息队列	TCP	5s
批处理	TTL	30s
gRPC 服务	gRPC	10s