版权声明：本文为博主原创，转载请注明出处！

简介

本文主要解决以下问题：

1、ES启动过程中的Node对象都初始化了那些服务？

构造流程

Step 1、创建一个List暂存初始化失败时需要释放的资源，并使用临时的Logger对象输出开始初始化的日志。

这里首先创建了一个List<Closeable>然后输出日志initializing ...。代码比较简单：

final List<Closeable> resourcesToClose = new ArrayList<>(); // register everything we need to release in the case of an error
boolean success = false;{// use temp logger just to say we are starting. we can't use it later on because the node name might not be setLogger logger = Loggers.getLogger(Node.class, NODE_NAME_SETTING.get(environment.settings()));logger.info("initializing ...");}

Step 2、强制设置settings中client.type的配置为node，设置node.name并检查索引data目录的设置。

这部分首先设置client.type为node，接下来调用TribeService的processSettings方法来处理了“部落”的配置，然后创建NodeEnvironment，检查并设置node.name属性，最后按需检查索引数据的Path的配置并打印一些JVM的信息。代码如下：

 Settings tmpSettings = Settings.builder().put(environment.settings()).put(Client.CLIENT_TYPE_SETTING_S.getKey(), CLIENT_TYPE).build();tmpSettings = TribeService.processSettings(tmpSettings);// create the node environment as soon as possible, to recover the node id and enable loggingtry {nodeEnvironment = new NodeEnvironment(tmpSettings, environment);resourcesToClose.add(nodeEnvironment);} catch (IOException ex) {throw new IllegalStateException("Failed to create node environment", ex);}final boolean hadPredefinedNodeName = NODE_NAME_SETTING.exists(tmpSettings);Logger logger = Loggers.getLogger(Node.class, tmpSettings);final String nodeId = nodeEnvironment.nodeId();tmpSettings = addNodeNameIfNeeded(tmpSettings, nodeId);if (DiscoveryNode.nodeRequiresLocalStorage(tmpSettings)) {checkForIndexDataInDefaultPathData(tmpSettings, nodeEnvironment, logger);}// this must be captured after the node name is possibly added to the settingsfinal String nodeName = NODE_NAME_SETTING.get(tmpSettings);if (hadPredefinedNodeName == false) {logger.info("node name [{}] derived from node ID [{}]; set [{}] to override", nodeName, nodeId, NODE_NAME_SETTING.getKey());} else {logger.info("node name [{}], node ID [{}]", nodeName, nodeId);}

Step 3、创建PluginsService及Environment实例。

在PluginsService的构造方法中会加载plugins和modules目录下的jar包，并创建相应的plugin和module实例。创建完以后，Node的构造方法中会调用pluginsService的updatedSettings方法来获取plugin和module中定义的配置项。接下来Node或使用新的settings和nodeId来创建LocalNodeFactory，并使用最新的settings重新创建Environment对象。代码如下：

 this.pluginsService = new PluginsService(tmpSettings, environment.modulesFile(), environment.pluginsFile(), classpathPlugins);this.settings = pluginsService.updatedSettings();localNodeFactory = new LocalNodeFactory(settings, nodeEnvironment.nodeId());// create the environment based on the finalized (processed) view of the settings// this is just to makes sure that people get the same settings, no matter where they ask them fromthis.environment = new Environment(this.settings);Environment.assertEquivalent(environment, this.environment);

Step 4、创建ThreadPool及ThreadContext实例。

首先，通过pluginsService获取plugin及module中提供的ExecutorBuilder对象列表。接下来基于settings及获取的ExecutorBuilder对象列表创建ThreadPool及ThreadContext实例。代码如下：

 final ThreadPool threadPool = new ThreadPool(settings, executorBuilders.toArray(new ExecutorBuilder[0]));resourcesToClose.add(() -> ThreadPool.terminate(threadPool, 10, TimeUnit.SECONDS));// adds the context to the DeprecationLogger so that it does not need to be injected everywhereDeprecationLogger.setThreadContext(threadPool.getThreadContext());resourcesToClose.add(() -> DeprecationLogger.removeThreadContext(threadPool.getThreadContext()));

Step 5、依次创建NodeClient、ResourceWatcherService、ScriptModule、AnalysisModule、SettingsModule、NetworkService、ClusterService、IngestService及ClusterInfoService等主要模块。

ScriptModule中持有ScriptService通过该服务可以获取到ES中配置的各类脚本引擎的实例。AnalysisModule中持有AnalysisRegistry对象，通过该对象可以获取到ES中配置的各类查询分析器的实例。SettingModule中按类型保存了ES中可以解析的配置对象。NetworkService主要用来解析网络地址，ClusterService用例维护集群的信息。代码如下：

 final List<Setting<?>> additionalSettings = new ArrayList<>(pluginsService.getPluginSettings());final List<String> additionalSettingsFilter = new ArrayList<>(pluginsService.getPluginSettingsFilter());for (final ExecutorBuilder<?> builder : threadPool.builders()) {additionalSettings.addAll(builder.getRegisteredSettings());}client = new NodeClient(settings, threadPool);final ResourceWatcherService resourceWatcherService = new ResourceWatcherService(settings, threadPool);final ScriptModule scriptModule = ScriptModule.create(settings, this.environment, resourceWatcherService,pluginsService.filterPlugins(ScriptPlugin.class));AnalysisModule analysisModule = new AnalysisModule(this.environment, pluginsService.filterPlugins(AnalysisPlugin.class));additionalSettings.addAll(scriptModule.getSettings());// this is as early as we can validate settings at this point. we already pass them to ScriptModule as well as ThreadPool// so we might be late here alreadyfinal SettingsModule settingsModule = new SettingsModule(this.settings, additionalSettings, additionalSettingsFilter);scriptModule.registerClusterSettingsListeners(settingsModule.getClusterSettings());resourcesToClose.add(resourceWatcherService);final NetworkService networkService = new NetworkService(settings,getCustomNameResolvers(pluginsService.filterPlugins(DiscoveryPlugin.class)));final ClusterService clusterService = new ClusterService(settings, settingsModule.getClusterSettings(), threadPool,localNodeFactory::getNode);clusterService.addStateApplier(scriptModule.getScriptService());resourcesToClose.add(clusterService);final IngestService ingestService = new IngestService(clusterService.getClusterSettings(), settings, threadPool, this.environment,scriptModule.getScriptService(), analysisModule.getAnalysisRegistry(), pluginsService.filterPlugins(IngestPlugin.class));final ClusterInfoService clusterInfoService = newClusterInfoService(settings, clusterService, threadPool, client);

Step 6、创建ModulesBuilder并加入各种Module。

ES使用google开源的Guice管理程序中的依赖。加入ModulesBuilder中的Module有：通过PluginsService获取的插件提供的Module；NodeModule内部持有MonitorService；ClusterModule内部持有ClusterService及相关的ClusterPlugin；IndicesModule内部持有MapperPlugin；SearchModule内部持有相关的SearchPlugin；ActionModule内部持有ThreadPool、ActionPlugin、NodeClient及CircuitBreakerService；GatewayModule；RepositoriesModule内部持有RepositoryPlugin；SttingsModule内部ES可用的各类配置对象等；最好调用modules的createInjector方法创建应用的“依赖注入器”。

Step 7、收集各plugin的LifecycleComponent对象，并出初始化NodeClient。

代码如下：

 List<LifecycleComponent> pluginLifecycleComponents = pluginComponents.stream().filter(p -> p instanceof LifecycleComponent).map(p -> (LifecycleComponent) p).collect(Collectors.toList());pluginLifecycleComponents.addAll(pluginsService.getGuiceServiceClasses().stream().map(injector::getInstance).collect(Collectors.toList()));resourcesToClose.addAll(pluginLifecycleComponents);this.pluginLifecycleComponents = Collections.unmodifiableList(pluginLifecycleComponents);client.initialize(injector.getInstance(new Key<Map<GenericAction, TransportAction>>() {}),() -> clusterService.localNode().getId());if (NetworkModule.HTTP_ENABLED.get(settings)) {logger.debug("initializing HTTP handlers ...");actionModule.initRestHandlers(() -> clusterService.state().nodes());}logger.info("initialized");

Step 8、调用Node的Start方法，在该方法内依次调用各重要模块的start方法。

依次启动各个关键服务。代码如下：

 // hack around dependency injection problem (for now...)injector.getInstance(Discovery.class).setAllocationService(injector.getInstance(AllocationService.class));pluginLifecycleComponents.forEach(LifecycleComponent::start);injector.getInstance(MappingUpdatedAction.class).setClient(client);injector.getInstance(IndicesService.class).start();injector.getInstance(IndicesClusterStateService.class).start();injector.getInstance(IndicesTTLService.class).start();injector.getInstance(SnapshotsService.class).start();injector.getInstance(SnapshotShardsService.class).start();injector.getInstance(RoutingService.class).start();injector.getInstance(SearchService.class).start();injector.getInstance(MonitorService.class).start();final ClusterService clusterService = injector.getInstance(ClusterService.class);final NodeConnectionsService nodeConnectionsService = injector.getInstance(NodeConnectionsService.class);nodeConnectionsService.start();clusterService.setNodeConnectionsService(nodeConnectionsService);// TODO hack around circular dependencies problemsinjector.getInstance(GatewayAllocator.class).setReallocation(clusterService, injector.getInstance(RoutingService.class));injector.getInstance(ResourceWatcherService.class).start();injector.getInstance(GatewayService.class).start();Discovery discovery = injector.getInstance(Discovery.class);clusterService.setDiscoverySettings(discovery.getDiscoverySettings());clusterService.addInitialStateBlock(discovery.getDiscoverySettings().getNoMasterBlock());clusterService.setClusterStatePublisher(discovery::publish);// start before the cluster service since it adds/removes initial Cluster state blocksfinal TribeService tribeService = injector.getInstance(TribeService.class);tribeService.start();// Start the transport service now so the publish address will be added to the local disco node in ClusterServiceTransportService transportService = injector.getInstance(TransportService.class);transportService.getTaskManager().setTaskResultsService(injector.getInstance(TaskResultsService.class));transportService.start();validateNodeBeforeAcceptingRequests(settings, transportService.boundAddress(), pluginsService.filterPlugins(Plugin.class).stream().flatMap(p -> p.getBootstrapChecks().stream()).collect(Collectors.toList()));clusterService.addStateApplier(transportService.getTaskManager());clusterService.start();assert localNodeFactory.getNode() != null;assert transportService.getLocalNode().equals(localNodeFactory.getNode()): "transportService has a different local node than the factory provided";assert clusterService.localNode().equals(localNodeFactory.getNode()): "clusterService has a different local node than the factory provided";// start after cluster service so the local disco is knowndiscovery.start();transportService.acceptIncomingRequests();discovery.startInitialJoin();// tribe nodes don't have a master so we shouldn't register an observer  sfinal TimeValue initialStateTimeout = DiscoverySettings.INITIAL_STATE_TIMEOUT_SETTING.get(settings);if (initialStateTimeout.millis() > 0) {final ThreadPool thread = injector.getInstance(ThreadPool.class);ClusterState clusterState = clusterService.state();ClusterStateObserver observer = new ClusterStateObserver(clusterState, clusterService, null, logger, thread.getThreadContext());if (clusterState.nodes().getMasterNodeId() == null) {logger.debug("waiting to join the cluster. timeout [{}]", initialStateTimeout);final CountDownLatch latch = new CountDownLatch(1);observer.waitForNextChange(new ClusterStateObserver.Listener() {@Overridepublic void onNewClusterState(ClusterState state) { latch.countDown(); }@Overridepublic void onClusterServiceClose() {latch.countDown();}@Overridepublic void onTimeout(TimeValue timeout) {logger.warn("timed out while waiting for initial discovery state - timeout: {}",initialStateTimeout);latch.countDown();}}, state -> state.nodes().getMasterNodeId() != null, initialStateTimeout);try {latch.await();} catch (InterruptedException e) {throw new ElasticsearchTimeoutException("Interrupted while waiting for initial discovery state");}}}if (NetworkModule.HTTP_ENABLED.get(settings)) {injector.getInstance(HttpServerTransport.class).start();}if (WRITE_PORTS_FILE_SETTING.get(settings)) {if (NetworkModule.HTTP_ENABLED.get(settings)) {HttpServerTransport http = injector.getInstance(HttpServerTransport.class);writePortsFile("http", http.boundAddress());}TransportService transport = injector.getInstance(TransportService.class);writePortsFile("transport", transport.boundAddress());}// start nodes now, after the http server, because it may take some timetribeService.startNodes();logger.info("started");