一、main方式入口
路径:
org.elasticsearch.bootstrap.Elasticsearch
java
/**
* 启动 elasticsearch 的主入口点。
*/
public static void main(final String[] args) {
Bootstrap bootstrap = initPhase1();
assert bootstrap != null;
try {
initPhase2(bootstrap);
initPhase3(bootstrap);
} catch (NodeValidationException e) {
bootstrap.exitWithNodeValidationException(e);
} catch (Throwable t) {
bootstrap.exitWithUnknownException(t);
}
}
这里初始化会有三个初始化阶段。可以直接看initPhase3
二、Elasticsearch初始化第三阶段
java
/**
*初始化的第三阶段
*阶段 3 包含初始化安全管理器后的所有内容。到目前为止,该系统一直是单线程的。此阶段可以生成线程、写入日志,并受安全管理器策略的约束。
* 在第 3 阶段结束时,系统已准备好接受请求,主线程已准备好终止。这意味着:
* 节点组件已构建并启动
* 清理已完成(例如安全设置已关闭)
* 除主线程外,至少有一个线程处于活动状态,并且在主线程终止后将保持活动状态
* 已通知父 CLI 进程系统已准备就绪
*/
private static void initPhase3(Bootstrap bootstrap) throws IOException, NodeValidationException {
//调用checkLucene()函数进行Lucene的检查
checkLucene();
//创建一个Node对象,并重写validateNodeBeforeAcceptingRequests方法,用于在接受请求之前进行节点验证。
Node node = new Node(bootstrap.environment()) {
@Override
protected void validateNodeBeforeAcceptingRequests(
final BootstrapContext context,
final BoundTransportAddress boundTransportAddress,
List<BootstrapCheck> checks
) throws NodeValidationException {
BootstrapChecks.check(context, boundTransportAddress, checks);
}
};
//使用bootstrap.spawner()和之前创建的node对象实例化一个Elasticsearch对象,并将其赋值给INSTANCE变量。
INSTANCE = new Elasticsearch(bootstrap.spawner(), node);
//关闭安全设置
IOUtils.close(bootstrap.secureSettings());
//启动INSTANCE对象,node会启动,并保持一个存活线程
INSTANCE.start();
//如果命令行参数指定了daemonize,则移除控制台输出的日志配置。
if (bootstrap.args().daemonize()) {
LogConfigurator.removeConsoleAppender();
}
//发送CLI标记,表示服务器已经准备好接受请求。
bootstrap.sendCliMarker(BootstrapInfo.SERVER_READY_MARKER);
//如果命令行参数指定了daemonize,则关闭流;否则,启动CLI监视线程。
if (bootstrap.args().daemonize()) {
bootstrap.closeStreams();
} else {
startCliMonitorThread(System.in);
}
}
其中INSTANCE.start();
如下,代表node启动,并且存活线程运行
java
private void start() throws NodeValidationException {
node.start();
keepAliveThread.start();
}
1、构造node节点对象时构造restController
java
public Node(Environment environment) {
this(environment, PluginsService.getPluginsServiceCtor(environment), true);
}
/**
* Constructs a node
* 节点的初始化
*/
protected Node(
final Environment initialEnvironment,
final Function<Settings, PluginsService> pluginServiceCtor,
boolean forbidPrivateIndexSettings
) {
//省略代码。。。。
//里面会初始化restController
ActionModule actionModule = new ActionModule(
settings,
clusterModule.getIndexNameExpressionResolver(),
settingsModule.getIndexScopedSettings(),
settingsModule.getClusterSettings(),
settingsModule.getSettingsFilter(),
threadPool,
pluginsService.filterPlugins(ActionPlugin.class),
client,
circuitBreakerService,
usageService,
systemIndices,
tracer,
clusterService,
reservedStateHandlers
);
modules.add(actionModule);
//restController存入到networkModule,而NetworkModule是用于处理注册和绑定所有网络相关类的模块
//末尾有 actionModule.initRestHandlers初始化hander
final RestController restController = actionModule.getRestController();
final NetworkModule networkModule = new NetworkModule(
settings,
pluginsService.filterPlugins(NetworkPlugin.class),
threadPool,
bigArrays,
pageCacheRecycler,
circuitBreakerService,
namedWriteableRegistry,
xContentRegistry,
networkService,
restController,
actionModule::copyRequestHeadersToThreadContext,
clusterService.getClusterSettings(),
tracer
);
//省略代码。。。。
//初始化Rest的Handler
actionModule.initRestHandlers(() -> clusterService.state().nodesIfRecovered());
}
2、在node构建对象最后执行初始化RestHanders的操作
java
public void initRestHandlers(Supplier<DiscoveryNodes> nodesInCluster) {
//省略代码。。。这里只选几个经常用到的
registerHandler.accept(new RestGetIndicesAction());
registerHandler.accept(new RestIndicesStatsAction());
registerHandler.accept(new RestCreateIndexAction());
registerHandler.accept(new RestDeleteIndexAction());
registerHandler.accept(new RestGetIndexTemplateAction());
registerHandler.accept(new RestPutIndexTemplateAction());
registerHandler.accept(new RestDeleteIndexTemplateAction());
registerHandler.accept(new RestPutMappingAction());
registerHandler.accept(new RestGetMappingAction());
registerHandler.accept(new RestGetFieldMappingAction());
registerHandler.accept(new RestIndexAction());
registerHandler.accept(new RestSearchAction(restController.getSearchUsageHolder()));
//省略代码
}
三、以注册在hander中的RestGetIndicesAction对象为例介绍
java
/**
* The REST handler for get index and head index APIs.
* 用于获取索引和头索引 API 的 REST 处理程序。
*/
@ServerlessScope(Scope.PUBLIC)
public class RestGetIndicesAction extends BaseRestHandler {
//代表路由匹配规则,通过这个规则知道要调用这个实例,每一个实例路由规则都是不一样的
@Override
public List<Route> routes() {
return List.of(new Route(GET, "/{index}"), new Route(HEAD, "/{index}"));
}
@Override
public RestChannelConsumer prepareRequest(final RestRequest request, final NodeClient client) throws IOException {
// starting with 7.0 we don't include types by default in the response to GET requests
if (request.getRestApiVersion() == RestApiVersion.V_7
&& request.hasParam(INCLUDE_TYPE_NAME_PARAMETER)
&& request.method().equals(GET)) {
deprecationLogger.compatibleCritical("get_indices_with_types", TYPES_DEPRECATION_MESSAGE);
}
String[] indices = Strings.splitStringByCommaToArray(request.param("index"));
final GetIndexRequest getIndexRequest = new GetIndexRequest();
getIndexRequest.indices(indices);
getIndexRequest.indicesOptions(IndicesOptions.fromRequest(request, getIndexRequest.indicesOptions()));
getIndexRequest.local(request.paramAsBoolean("local", getIndexRequest.local()));
getIndexRequest.masterNodeTimeout(request.paramAsTime("master_timeout", getIndexRequest.masterNodeTimeout()));
getIndexRequest.humanReadable(request.paramAsBoolean("human", false));
getIndexRequest.includeDefaults(request.paramAsBoolean("include_defaults", false));
getIndexRequest.features(GetIndexRequest.Feature.fromRequest(request));
final var httpChannel = request.getHttpChannel();
return channel -> new RestCancellableNodeClient(client, httpChannel).admin()
.indices()
.getIndex(getIndexRequest, new RestChunkedToXContentListener<>(channel));
}
}
1、继承了BaseRestHandler,routes方法做路由规则,父类调用子类的prepareRequest实现
java
public abstract class BaseRestHandler implements RestHandler {
/**
* {@inheritDoc}
*/
@Override
public abstract List<Route> routes();
@Override
public final void handleRequest(RestRequest request, RestChannel channel, NodeClient client) throws Exception {
//调用prepareRequest方法,准备请求以供执行,并且会对请求参数进行处理。
final RestChannelConsumer action = prepareRequest(request, client);
//过滤未使用的参数,将未使用的参数收集到一个有序集合中。
final SortedSet<String> unconsumedParams = request.unconsumedParams()
.stream()
.filter(p -> responseParams(request.getRestApiVersion()).contains(p) == false)
.collect(Collectors.toCollection(TreeSet::new));
//验证未使用的参数是否有效,如果存在无效参数,则抛出IllegalArgumentException异常。
if (unconsumedParams.isEmpty() == false) {
final Set<String> candidateParams = new HashSet<>();
candidateParams.addAll(request.consumedParams());
candidateParams.addAll(responseParams(request.getRestApiVersion()));
throw new IllegalArgumentException(unrecognized(request, unconsumedParams, candidateParams, "parameter"));
}
//验证请求是否包含请求体,并且请求体是否已被消耗,如果不满足条件,则抛出IllegalArgumentException异常。
if (request.hasContent() && request.isContentConsumed() == false) {
throw new IllegalArgumentException("request [" + request.method() + " " + request.path() + "] does not support having a body");
}
//增加使用计数
usageCount.increment();
//执行action,将结果传递给channel。
action.accept(channel);
}
/**
*准备要执行的请求。
* 实现应在返回可运行对象以进行实际执行之前使用所有请求参数。
* 未使用的参数将立即终止请求的执行。
* 但是,某些参数仅用于处理响应;实现可以覆盖 {@link BaseRestHandlerresponseParams()} 来指示此类参数。
*/
protected abstract RestChannelConsumer prepareRequest(RestRequest request, NodeClient client) throws IOException;
}
2、BaseRestHandler实现的是RestHandler接口
java
/**
* Handler for REST requests
*/
@FunctionalInterface
public interface RestHandler {
/**
* 处理rest请求
*/
void handleRequest(RestRequest request, RestChannel channel, NodeClient client) throws Exception;
/**
* 此 RestHandler 负责处理的 {@link 路由} 的列表。
*/
default List<Route> routes() {
return Collections.emptyList();
}
}
其中调用RestHandler
接口的handerRequest
的上游是
java
@Override
public void handleRequest(RestRequest request, RestChannel channel, NodeClient client) throws Exception {
//使用 OPTIONS 方法的请求应在其他地方处理,而不是通过调用 {@code RestHandlerhandleRequest} 使用 OPTIONS 方法的 HTTP 请求绕过 authn,因此此健全性检查可防止调度未经身份验证的请求
if (request.method() == Method.OPTIONS) {
handleException(
request,
channel,
new ElasticsearchSecurityException("Cannot dispatch OPTIONS request, as they are not authenticated")
);
return;
}
if (enabled == false) {
doHandleRequest(request, channel, client);
return;
}
}
private void doHandleRequest(RestRequest request, RestChannel channel, NodeClient client) throws Exception {
threadContext.sanitizeHeaders();
// operator privileges can short circuit to return a non-successful response
if (operatorPrivilegesService.checkRest(restHandler, request, channel, threadContext)) {
try {
restHandler.handleRequest(request, channel, client);
} catch (Exception e) {
logger.debug(() -> format("Request handling failed for REST request [%s]", request.uri()), e);
throw e;
}
}
}
其他注册在hander
中的API和RestGetIndicesAction
类似