Hadoop源码解读-Http服务器Jetty的使用

青野

　　Hadoop内嵌了Http服务器Jetty，主要有以下两方面的作用
　　1、Web访问接口，用于展示Hadoop的内部状态
　　2、参与Hadoop集群的运行和管理
　　以Namenode为例
　　Namenode通过

startHttpServer(conf);
　　来启动HttpServer（Jetty），具体代码如下

          httpServer = new HttpServer("hdfs", infoHost, infoPort,
infoPort == 0, conf,
SecurityUtil.getAdminAcls(conf, DFSConfigKeys.DFS_ADMIN));

　　深入HttpServer可以看到，上面的代码是将HADOOP_HOME\webapp下面的hdfs目录作为了jetty的默认Context（datanode就用datanode目录，jobtracker为job目录，tasktracker为task目录，secondarynamenode为secondary目录）

    webAppContext = new WebAppContext();
webAppContext.setDisplayName("WepAppsContext");
webAppContext.setContextPath("/");
webAppContext.setWar(appDir + "/" + name);
webAppContext.getServletContext().setAttribute(CONF_CONTEXT_ATTRIBUTE, conf);
webAppContext.getServletContext().setAttribute(ADMINS_ACL, adminsAcl);
webServer.addHandler(webAppContext);

　　同时添加了对Log和webapps下面static资源（css、js、pic）的访问

  protected void addDefaultApps(ContextHandlerCollection parent,
final String appDir) throws IOException {
// set up the context for "/logs/" if "hadoop.log.dir" property is defined.
String logDir = System.getProperty("hadoop.log.dir");
if (logDir != null) {
Context logContext = new Context(parent, "/logs");
logContext.setResourceBase(logDir);
logContext.addServlet(AdminAuthorizedServlet.class, "/");
logContext.setDisplayName("logs");
setContextAttributes(logContext);
defaultContexts.put(logContext, true);
}
// set up the context for "/static/*"
Context staticContext = new Context(parent, "/static");
staticContext.setResourceBase(appDir + "/static");
staticContext.addServlet(DefaultServlet.class, "/*");
staticContext.setDisplayName("static");
setContextAttributes(staticContext);
defaultContexts.put(staticContext, true);
}
　　以及一些状态信息的访问

  /**
* Add default servlets.
*/
protected void addDefaultServlets() {
// set up default servlets
addServlet("stacks", "/stacks", StackServlet.class);
addServlet("logLevel", "/logLevel", LogLevel.Servlet.class);
addServlet("metrics", "/metrics", MetricsServlet.class);
addServlet("conf", "/conf", ConfServlet.class);
addServlet("jmx", "/jmx", JMXJsonServlet.class);
}
 　　最后，返回namenode，又添加了一些namenode特有的访问接口，例如
　　/fsck用于文件系统的检查
　　/getimage是SecondaryNamenode获取image的入口

          httpServer.addInternalServlet("getDelegationToken",
GetDelegationTokenServlet.PATH_SPEC,
GetDelegationTokenServlet.class, true);
httpServer.addInternalServlet("renewDelegationToken",
RenewDelegationTokenServlet.PATH_SPEC,
RenewDelegationTokenServlet.class, true);
httpServer.addInternalServlet("cancelDelegationToken",
CancelDelegationTokenServlet.PATH_SPEC,
CancelDelegationTokenServlet.class,
true);
httpServer.addInternalServlet("fsck", "/fsck", FsckServlet.class, true);
httpServer.addInternalServlet("getimage", "/getimage",
GetImageServlet.class, true);
httpServer.addInternalServlet("listPaths", "/listPaths/*",
ListPathsServlet.class, false);
httpServer.addInternalServlet("data", "/data/*",
FileDataServlet.class, false);
httpServer.addInternalServlet("checksum", "/fileChecksum/*",
FileChecksumServlets.RedirectServlet.class, false);
httpServer.addInternalServlet("contentSummary", "/contentSummary/*",
ContentSummaryServlet.class, false);
httpServer.start();
// The web-server port can be ephemeral... ensure we have the correct info
infoPort = httpServer.getPort();
httpAddress = new InetSocketAddress(infoHost, infoPort);
conf.set("dfs.http.address", infoHost + ":" + infoPort);
LOG.info("Web-server up at: " + infoHost + ":" + infoPort);
return httpServer;
　　再打开hdfs目录，会发现index页面会直接转跳到dfshealth.jsp，查看web.xml

    <servlet-mapping>
<servlet-name>org.apache.hadoop.hdfs.server.namenode.dfshealth_jsp</servlet-name>
<url-pattern>/dfshealth.jsp</url-pattern>
</servlet-mapping>
　　dfshealth_jsp.class可以从hadoop-core-xxx.jar里面找到