配置中心 Apollo 源码解析 —— 服务的注册与发现

1,891 阅读9分钟
原文链接: www.iocoder.cn
《Dubbo 实现原理与源码解析 —— 精品合集》 《Netty 实现原理与源码解析 —— 精品合集》
《Spring 实现原理与源码解析 —— 精品合集》 《MyBatis 实现原理与源码解析 —— 精品合集》
《Spring MVC 实现原理与源码解析 —— 精品合集》 《数据库实体设计合集》
《Spring Boot 实现原理与源码解析 —— 精品合集》 《Java 面试题 + Java 学习指南》

摘要: 原创出处 http://www.iocoder.cn/Apollo/service-register-discovery/ 「芋道源码」欢迎转载,保留摘要,谢谢!



1. 概述

老艿艿:本系列假定胖友已经阅读过 《Apollo 官方 wiki 文档》 ,特别是 《架构模块》

本文分享 Apollo 服务的注册与发现。如下图所示:

服务的注册与发现

2. Eureka Server

2.1 启动 Eureka Server

apollo-configservice 项目中,com.ctrip.framework.apollo.configservice.ConfigServiceApplication 中,通过 @EnableEurekaServer 注解启动 Eureka Server 。代码如下:

@EnableEurekaServer // 启动 Eureka Server
@EnableAspectJAutoProxy
@EnableAutoConfiguration // (exclude = EurekaClientConfigBean.class)
@Configuration
@EnableTransactionManagement
@PropertySource(value = {"classpath:configservice.properties"})
@ComponentScan(basePackageClasses = {ApolloCommonConfig.class,
        ApolloBizConfig.class,
        ConfigServiceApplication.class,
        ApolloMetaServiceConfig.class})
public class ConfigServiceApplication {

    public static void main(String[] args) throws Exception {
        ConfigurableApplicationContext context = new SpringApplicationBuilder(ConfigServiceApplication.class).run(args);
        context.addApplicationListener(new ApplicationPidFileWriter());
        context.addApplicationListener(new EmbeddedServerPortFileWriter());
    }

}

  • 第一行的 @EnableEurekaServer 注解,启动 Eureka Server 。基于 Spring Cloud Eureka ,需要在 Maven 的 pom.xml 中申明如下依赖:

    <!-- eureka -->
    <dependency>
    	<groupId>org.springframework.cloud</groupId>
    	<artifactId>spring-cloud-starter-eureka-server</artifactId>
    	<exclusions>
    		<exclusion>
    			<artifactId>
                    spring-cloud-starter-archaius
                </artifactId>
    			<groupId>org.springframework.cloud</groupId>
    		</exclusion>
    		<exclusion>
    			<artifactId>spring-cloud-starter-ribbon</artifactId>
    			<groupId>org.springframework.cloud</groupId>
    		</exclusion>
    		<exclusion>
    			<artifactId>ribbon-eureka</artifactId>
    			<groupId>com.netflix.ribbon</groupId>
    		</exclusion>
    		<exclusion>
    			<artifactId>aws-java-sdk-core</artifactId>
    			<groupId>com.amazonaws</groupId>
    		</exclusion>
    		<exclusion>
    			<artifactId>aws-java-sdk-ec2</artifactId>
    			<groupId>com.amazonaws</groupId>
    		</exclusion>
    		<exclusion>
    			<artifactId>aws-java-sdk-autoscaling</artifactId>
    			<groupId>com.amazonaws</groupId>
    		</exclusion>
    		<exclusion>
    			<artifactId>aws-java-sdk-sts</artifactId>
    			<groupId>com.amazonaws</groupId>
    		</exclusion>
    		<exclusion>
    			<artifactId>aws-java-sdk-route53</artifactId>
    			<groupId>com.amazonaws</groupId>
    		</exclusion>
    		<!-- duplicated with spring-security-core -->
    		<exclusion>
    			<groupId>org.springframework.security</groupId>
    			<artifactId>spring-security-crypto</artifactId>
    		</exclusion>
    	</exclusions>
    </dependency>
    <!-- end of eureka -->
    

那么 Eureka Server 怎么构建成集群呢?答案在 「2.2 注册到 Eureka Client」 中。

2.2 注册到 Eureka Client

apollo-biz 项目中,com.ctrip.framework.apollo.biz.eureka.ApolloEurekaClientConfig 中,声明 Eureka 的配置。代码如下:

@Component
@Primary
public class ApolloEurekaClientConfig extends EurekaClientConfigBean {

    @Autowired
    private BizConfig bizConfig;

    /**
     * Assert only one zone: defaultZone, but multiple environments.
     */
    @Override
    public List<String> getEurekaServerServiceUrls(String myZone) {
        List<String> urls = bizConfig.eurekaServiceUrls();
        return CollectionUtils.isEmpty(urls) ? super.getEurekaServerServiceUrls(myZone) : urls;
    }

    @Override
    public boolean equals(Object o) {
        return super.equals(o);
    }

}

  • @Primary 注解,保证优先级

  • #getEurekaServerServiceUrls(myZone) 方法,调用 BizConfig#eurekaServiceUrls() 方法,从 ServerConfig 的 "eureka.service.url" 配置项,获得 Eureka Server 地址。代码如下:

    // 获得 Eureka 服务器地址的数组
    public List<String> eurekaServiceUrls() {
        // 获得配置值
        String configuration = getValue("eureka.service.url", "");
        // 分隔成 List
        if (Strings.isNullOrEmpty(configuration)) {
            return Collections.emptyList();
        }
        return splitter.splitToList(configuration);
    }
    

    • Eureka Server 共享该配置,从而形成 Eureka Server 集群
  • 基于 Spring Cloud Eureka ,需要在 Maven 的 pom.xml 中申明如下依赖:

    <!-- eureka -->
    <dependency>
    	<groupId>org.springframework.cloud</groupId>
    	<artifactId>spring-cloud-starter-eureka</artifactId>
    </dependency>
    <!-- end of eureka -->
    

apollo-adminserviceapollo-configservice 项目,引入 apollo-biz 项目,启动 Eureka Client ,向 Eureka Server 注册自己为实例。通过 .properties 配置实例名

// FROM adminservice.properties
spring.application.name= apollo-adminservice

// FROM configservice.properties
spring.application.name= apollo-configservice

3. Meta Service

apollo-configservice 项目中,metaservice 下,看到所有 Meta Service 的类,如下图:Meta Service

3.1 ApolloMetaServiceConfig

@EnableAutoConfiguration
@Configuration
@ComponentScan(basePackageClasses = ApolloMetaServiceConfig.class)
public class ApolloMetaServiceConfig {
}

3.2 ServiceController

@RestController
@RequestMapping("/services")
public class ServiceController {

    @Autowired
    private DiscoveryService discoveryService;

    @RequestMapping("/meta")
    public List<ServiceDTO> getMetaService() {
        List<InstanceInfo> instances = discoveryService.getMetaServiceInstances();
        List<ServiceDTO> result = instances.stream().map(new Function<InstanceInfo, ServiceDTO>() {

            @Override
            public ServiceDTO apply(InstanceInfo instance) {
                ServiceDTO service = new ServiceDTO();
                service.setAppName(instance.getAppName());
                service.setInstanceId(instance.getInstanceId());
                service.setHomepageUrl(instance.getHomePageUrl());
                return service;
            }

        }).collect(Collectors.toList());
        return result;
    }

    @RequestMapping("/config")
    public List<ServiceDTO> getConfigService(
            @RequestParam(value = "appId", defaultValue = "") String appId,
            @RequestParam(value = "ip", required = false) String clientIp) {
        List<InstanceInfo> instances = discoveryService.getConfigServiceInstances();
        List<ServiceDTO> result = instances.stream().map(new Function<InstanceInfo, ServiceDTO>() {

            @Override
            public ServiceDTO apply(InstanceInfo instance) {
                ServiceDTO service = new ServiceDTO();
                service.setAppName(instance.getAppName());
                service.setInstanceId(instance.getInstanceId());
                service.setHomepageUrl(instance.getHomePageUrl());
                return service;
            }

        }).collect(Collectors.toList());
        return result;
    }

    @RequestMapping("/admin")
    public List<ServiceDTO> getAdminService() {
        List<InstanceInfo> instances = discoveryService.getAdminServiceInstances();
        List<ServiceDTO> result = instances.stream().map(new Function<InstanceInfo, ServiceDTO>() {

            @Override
            public ServiceDTO apply(InstanceInfo instance) {
                ServiceDTO service = new ServiceDTO();
                service.setAppName(instance.getAppName());
                service.setInstanceId(instance.getInstanceId());
                service.setHomepageUrl(instance.getHomePageUrl());
                return service;
            }

        }).collect(Collectors.toList());
        return result;
    }

}

  • 提供了三个 API ,services/metaservices/configservices/admin 获得 Meta Service、Config Service、Admin Service 集群地址。😈 实际上,services/meta 暂时是不可用的,获取不到实例,因为 Meta Service 目前内嵌在 Config Service 中。

  • 每个 API 中,调用 DiscoveryService 调用对应的方法,获取服务集群。

  • com.ctrip.framework.apollo.core.dto.ServiceDTO ,服务 DTO 。代码如下:

    public class ServiceDTO {
    
        /**
         * 应用名
         */
        private String appName;
        /**
         * 实例编号
         */
        private String instanceId;
        /**
         * Home URL
         */
        private String homepageUrl;
    }
    

3.3 DiscoveryService

@Service
public class DiscoveryService {

    @Autowired
    private EurekaClient eurekaClient;

    public List<InstanceInfo> getConfigServiceInstances() {
        Application application = eurekaClient.getApplication(ServiceNameConsts.APOLLO_CONFIGSERVICE);
        if (application == null) {
            Tracer.logEvent("Apollo.EurekaDiscovery.NotFound", ServiceNameConsts.APOLLO_CONFIGSERVICE);
        }
        return application != null ? application.getInstances() : Collections.emptyList();
    }

    public List<InstanceInfo> getMetaServiceInstances() {
        Application application = eurekaClient.getApplication(ServiceNameConsts.APOLLO_METASERVICE);
        if (application == null) {
            Tracer.logEvent("Apollo.EurekaDiscovery.NotFound", ServiceNameConsts.APOLLO_METASERVICE);
        }
        return application != null ? application.getInstances() : Collections.emptyList();
    }

    public List<InstanceInfo> getAdminServiceInstances() {
        Application application = eurekaClient.getApplication(ServiceNameConsts.APOLLO_ADMINSERVICE);
        if (application == null) {
            Tracer.logEvent("Apollo.EurekaDiscovery.NotFound", ServiceNameConsts.APOLLO_ADMINSERVICE);
        }
        return application != null ? application.getInstances() : Collections.emptyList();
    }

}

  • 每个方法,调用 EurekaClient#getApplication(appName) 方法,获得服务集群。

  • com.ctrip.framework.apollo.core.ServiceNameConsts ,枚举了所有服务的名字。代码如下:

    public interface ServiceNameConsts {
    
        String APOLLO_METASERVICE = "apollo-metaservice";
    
        String APOLLO_CONFIGSERVICE = "apollo-configservice";
    
        String APOLLO_ADMINSERVICE = "apollo-adminservice";
    
        String APOLLO_PORTAL = "apollo-portal";
    
    }
    

3.4 集群

考虑到高可用,Meta Service 必须集群。因为 Meta Service 自身扮演了目录服务的角色,所以此时不得不引入 Proxy Server 。从选择上,笔者想到的是:

  1. Nginx ,目前互联网上最常用的 Proxy Server 。
  2. Zuul ,可以和 Eureka 打通,实现注册与发现。

因为 Meta Service 目前并未注册到 Zuul 上,所以相比来说,Nginx 会是更合适的选择。当然,😈 Nginx 自身也是要做高可用的,哈哈哈,这块胖友自己 Google 下解决方案。

在高性能之前,一切服务节点必须高可用。任何服务节点的松懈,势必在未来的某个时刻,给我们来一波暴击!!!

4. ConfigServiceLocator

apollo-client 项目中,com.ctrip.framework.apollo.internals.ConfigServiceLocator ,Config Service 定位器。

  • 初始时,从 Meta Service 获取 Config Service 集群地址进行缓存
  • 定时任务,每 5 分钟,从 Meta Service 获取 Config Service 集群地址刷新缓存

🙂 代码比较简单,胖友自己查看。代码如下:

public class ConfigServiceLocator {

    private static final Logger logger = LoggerFactory.getLogger(ConfigServiceLocator.class);

    private static final Joiner.MapJoiner MAP_JOINER = Joiner.on("&").withKeyValueSeparator("=");
    private static final Escaper queryParamEscaper = UrlEscapers.urlFormParameterEscaper();

    private HttpUtil m_httpUtil;
    private ConfigUtil m_configUtil;
    /**
     * ServiceDTO 数组的缓存
     */
    private AtomicReference<List<ServiceDTO>> m_configServices;
    private Type m_responseType;
    /**
     * 定时任务 ExecutorService
     */
    private ScheduledExecutorService m_executorService;

    /**
     * Create a config service locator.
     */
    public ConfigServiceLocator() {
        List<ServiceDTO> initial = Lists.newArrayList();
        m_configServices = new AtomicReference<>(initial);
        m_responseType = new TypeToken<List<ServiceDTO>>() {}.getType();
        m_httpUtil = ApolloInjector.getInstance(HttpUtil.class);
        m_configUtil = ApolloInjector.getInstance(ConfigUtil.class);
        this.m_executorService = Executors.newScheduledThreadPool(1, ApolloThreadFactory.create("ConfigServiceLocator", true));
        // 初始拉取 Config Service 地址
        this.tryUpdateConfigServices();
        // 创建定时任务,定时拉取 Config Service 地址
        this.schedulePeriodicRefresh();
    }

    /**
     * Get the config service info from remote meta server.
     *
     * @return the services dto
     */
    public List<ServiceDTO> getConfigServices() {
        // 缓存为空,强制拉取
        if (m_configServices.get().isEmpty()) {
            updateConfigServices();
        }
        // 返回 ServiceDTO 数组
        return m_configServices.get();
    }

    private boolean tryUpdateConfigServices() {
        try {
            updateConfigServices();
            return true;
        } catch (Throwable ex) {
            //ignore
        }
        return false;
    }

    private void schedulePeriodicRefresh() {
        this.m_executorService.scheduleAtFixedRate(new Runnable() {
            @Override
            public void run() {
                logger.debug("refresh config services");
                Tracer.logEvent("Apollo.MetaService", "periodicRefresh");
                // 拉取 Config Service 地址
                tryUpdateConfigServices();
            }
        }, m_configUtil.getRefreshInterval(), m_configUtil.getRefreshInterval(), m_configUtil.getRefreshIntervalTimeUnit());
    }

    private synchronized void updateConfigServices() {
        // 拼接请求 Meta Service URL
        String url = assembleMetaServiceUrl();

        HttpRequest request = new HttpRequest(url);
        int maxRetries = 2; // 重试两次
        Throwable exception = null;

        // 循环请求 Meta Service ,获取 Config Service 地址
        for (int i = 0; i < maxRetries; i++) {
            Transaction transaction = Tracer.newTransaction("Apollo.MetaService", "getConfigService");
            transaction.addData("Url", url);
            try {
                // 请求
                HttpResponse<List<ServiceDTO>> response = m_httpUtil.doGet(request, m_responseType);
                transaction.setStatus(Transaction.SUCCESS);
                // 获得结果 ServiceDTO 数组
                List<ServiceDTO> services = response.getBody();
                // 获得结果为空,重新请求
                if (services == null || services.isEmpty()) {
                    logConfigService("Empty response!");
                    continue;
                }
                // 更新缓存
                m_configServices.set(services);
                // 打印结果 ServiceDTO 数组
                logConfigServices(services);
                return;
            } catch (Throwable ex) {
                Tracer.logEvent("ApolloConfigException", ExceptionUtil.getDetailMessage(ex));
                transaction.setStatus(ex);
                exception = ex;
            } finally {
                transaction.complete();
            }
            // 请求失败,sleep 等待下次重试
            try {
                m_configUtil.getOnErrorRetryIntervalTimeUnit().sleep(m_configUtil.getOnErrorRetryInterval());
            } catch (InterruptedException ex) {
                // ignore
            }
        }
        // 请求全部失败,抛出 ApolloConfigException 异常
        throw new ApolloConfigException(String.format("Get config services failed from %s", url), exception);
    }

    private String assembleMetaServiceUrl() {
        String domainName = m_configUtil.getMetaServerDomainName();
        String appId = m_configUtil.getAppId();
        String localIp = m_configUtil.getLocalIp();

        // 参数集合
        Map<String, String> queryParams = Maps.newHashMap();
        queryParams.put("appId", queryParamEscaper.escape(appId));
        if (!Strings.isNullOrEmpty(localIp)) {
            queryParams.put("ip", queryParamEscaper.escape(localIp));
        }

        return domainName + "/services/config?" + MAP_JOINER.join(queryParams);
    }

    private void logConfigServices(List<ServiceDTO> serviceDtos) {
        for (ServiceDTO serviceDto : serviceDtos) {
            logConfigService(serviceDto.getHomepageUrl());
        }
    }

    private void logConfigService(String serviceUrl) {
        Tracer.logEvent("Apollo.Config.Services", serviceUrl);
    }
    
}

5. AdminServiceAddressLocator

apollo-portal 项目中,com.ctrip.framework.apollo.portal.component.AdminServiceAddressLocator ,Admin Service 定位器。

  • 初始时,创建延迟 1 秒的任务,从 Meta Service 获取 Config Service 集群地址进行缓存
  • 获取成功时,创建延迟 5 分钟的任务,从 Meta Service 获取 Config Service 集群地址刷新缓存
  • 获取失败时,创建延迟 10 秒的任务,从 Meta Service 获取 Config Service 集群地址刷新缓存

🙂 代码比较简单,胖友自己查看。代码如下:

@Component
public class AdminServiceAddressLocator {

    private static final Logger logger = LoggerFactory.getLogger(AdminServiceAddressLocator.class);

    private static final long NORMAL_REFRESH_INTERVAL = 5 * 60 * 1000;
    private static final long OFFLINE_REFRESH_INTERVAL = 10 * 1000;
    private static final int RETRY_TIMES = 3;
    private static final String ADMIN_SERVICE_URL_PATH = "/services/admin";

    /**
     * 定时任务 ExecutorService
     */
    private ScheduledExecutorService refreshServiceAddressService;
    private RestTemplate restTemplate;
    /**
     * Env 数组
     */
    private List<Env> allEnvs;
    /**
     * List<ServiceDTO 缓存 Map
     *
     * KEY:ENV
     */
    private Map<Env, List<ServiceDTO>> cache = new ConcurrentHashMap<>();
    @Autowired
    private HttpMessageConverters httpMessageConverters; // 暂未使用
    @Autowired
    private PortalSettings portalSettings;
    @Autowired
    private RestTemplateFactory restTemplateFactory;

    @PostConstruct
    public void init() {
        // 获得 Env 数组
        allEnvs = portalSettings.getAllEnvs();
        // init restTemplate
        restTemplate = restTemplateFactory.getObject();
        // 创建 ScheduledExecutorService
        refreshServiceAddressService = Executors.newScheduledThreadPool(1, ApolloThreadFactory.create("ServiceLocator", true));
        // 创建延迟任务,1 秒后拉取 Admin Service 地址
        refreshServiceAddressService.schedule(new RefreshAdminServerAddressTask(), 1, TimeUnit.MILLISECONDS);
    }

    public List<ServiceDTO> getServiceList(Env env) {
        // 从缓存中获得 ServiceDTO 数组
        List<ServiceDTO> services = cache.get(env);
        // 若不存在,直接返回空数组。这点和 ConfigServiceLocator 不同。
        if (CollectionUtils.isEmpty(services)) {
            return Collections.emptyList();
        }
        // 打乱 ServiceDTO 数组,返回。实现 Client 级的负载均衡
        List<ServiceDTO> randomConfigServices = Lists.newArrayList(services);
        Collections.shuffle(randomConfigServices);
        return randomConfigServices;
    }

    // maintain admin server address
    private class RefreshAdminServerAddressTask implements Runnable {

        @Override
        public void run() {
            boolean refreshSuccess = true;
            // 循环多个 Env ,请求对应的 Meta Service ,获得 Admin Service 集群地址
            // refresh fail if get any env address fail
            for (Env env : allEnvs) {
                boolean currentEnvRefreshResult = refreshServerAddressCache(env);
                refreshSuccess = refreshSuccess && currentEnvRefreshResult;
            }
            // 若刷新成功,则创建定时任务,5 分钟后执行
            if (refreshSuccess) {
                refreshServiceAddressService.schedule(new RefreshAdminServerAddressTask(), NORMAL_REFRESH_INTERVAL, TimeUnit.MILLISECONDS);
            // 若刷新失败,则创建定时任务,10 秒后执行
            } else {
                refreshServiceAddressService.schedule(new RefreshAdminServerAddressTask(), OFFLINE_REFRESH_INTERVAL, TimeUnit.MILLISECONDS);
            }
        }
    }

    private boolean refreshServerAddressCache(Env env) {
        for (int i = 0; i < RETRY_TIMES; i++) {
            try {
                // 请求 Meta Service ,获得 Admin Service 集群地址
                ServiceDTO[] services = getAdminServerAddress(env);
                // 获得结果为空,continue ,继续执行下一次请求
                if (services == null || services.length == 0) {
                    continue;
                }
                // 更新缓存
                cache.put(env, Arrays.asList(services));
                // 返回获取成功
                return true;
            } catch (Throwable e) {
                logger.error(String.format("Get admin server address from meta server failed. env: %s, meta server address:%s", env, MetaDomainConsts.getDomain(env)), e);
                Tracer.logError(String.format("Get admin server address from meta server failed. env: %s, meta server address:%s", env, MetaDomainConsts.getDomain(env)), e);
            }
        }
        // 返回获取失败
        return false;
    }

    private ServiceDTO[] getAdminServerAddress(Env env) {
        String domainName = MetaDomainConsts.getDomain(env); // MetaDomainConsts
        String url = domainName + ADMIN_SERVICE_URL_PATH;
        return restTemplate.getForObject(url, ServiceDTO[].class);
    }

}

5.1 MetaDomainConsts

com.ctrip.framework.apollo.core.MetaDomainConsts ,Meta Service 多环境的地址枚举类。代码如下:

/**
 * The meta domain will load the meta server from System environment first, if not exist, will load
 * from apollo-env.properties. If neither exists, will load the default meta url.
 * <p>
 * Currently, apollo supports local/dev/fat/uat/lpt/pro environments.
 */
public class MetaDomainConsts {

    private static Map<Env, Object> domains = new HashMap<>();

    public static final String DEFAULT_META_URL = "http://config.local";

    static {
        // 读取配置文件到 Properties 中
        Properties prop = new Properties();
        prop = ResourceUtils.readConfigFile("apollo-env.properties", prop);
        // 获得系统 Properties
        Properties env = System.getProperties();
        // 添加到 domains 中
        // 优先级,env > prop
        domains.put(Env.LOCAL, env.getProperty("local_meta", prop.getProperty("local.meta", DEFAULT_META_URL)));
        domains.put(Env.DEV, env.getProperty("dev_meta", prop.getProperty("dev.meta", DEFAULT_META_URL)));
        domains.put(Env.FAT, env.getProperty("fat_meta", prop.getProperty("fat.meta", DEFAULT_META_URL)));
        domains.put(Env.UAT, env.getProperty("uat_meta", prop.getProperty("uat.meta", DEFAULT_META_URL)));
        domains.put(Env.LPT, env.getProperty("lpt_meta", prop.getProperty("lpt.meta", DEFAULT_META_URL)));
        domains.put(Env.PRO, env.getProperty("pro_meta", prop.getProperty("pro.meta", DEFAULT_META_URL)));
    }

    public static String getDomain(Env env) {
        return String.valueOf(domains.get(env));
    }

}

  • 具体的读取顺序和说明,见英文注释说明。🙂 英语和我一样有非常大的进步空的同学,可以使用有道词典翻译。

5.2 Env

com.ctrip.framework.apollo.core.enums.Env ,环境枚举。代码如下:

/**
 * Here is the brief description for all the predefined environments:
 * <ul>
 * <li>LOCAL: Local Development environment, assume you are working at the beach with no network access</li>
 * <li>DEV: Development environment</li>
 * <li>FWS: Feature Web Service Test environment</li>
 * <li>FAT: Feature Acceptance Test environment</li>
 * <li>UAT: User Acceptance Test environment</li>
 * <li>LPT: Load and Performance Test environment</li>
 * <li>PRO: Production environment</li>
 * <li>TOOLS: Tooling environment, a special area in production environment which allows
 * access to test environment, e.g. Apollo Portal should be deployed in tools environment</li>
 * </ul>
 *
 * @author Jason Song(song_s@ctrip.com)
 */
public enum Env {

    LOCAL, DEV, FWS, FAT, UAT, LPT, PRO, TOOLS;

    public static Env fromString(String env) {
        Env environment = EnvUtils.transformEnv(env);
        Preconditions.checkArgument(environment != null, String.format("Env %s is invalid", env));
        return environment;
    }

}

😎 前面一直忘记对 Env 介绍,所以有些奇怪的放在这个位置。主要目的是,我们可以参考携程对服务环境的命名和定义。

666. 彩蛋

😝😝😝 第4、5 小节写的比较简略,如果有不理解的胖友,可以给我的公众号留言。啦啦啦,我去刷《复仇者联盟3》啦,美滋滋。

知识星球