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