在项目开发前期，使用jdbc访问hive数据仓库，在配置文件配置active namenode的地址信息，上线的时候才发现如果namenode宕机了，程序将无法运行。后来改成通过zookeeper获取active namenode地址，这样即使namenode宕机了，也能通过zookeeper获取备机地址从而不影响程序运行。记录下代码：

package cp.app.preprocess.service;

import java.io.IOException;
import java.net.URI;
import java.util.ArrayList;
import java.util.List;

import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileStatus;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.hdfs.server.namenode.ha.proto.HAZKInfoProtos;
import org.apache.hadoop.hdfs.server.namenode.ha.proto.HAZKInfoProtos.ActiveNodeInfo;
import org.apache.log4j.Logger;
import org.apache.zookeeper.Watcher;
import org.apache.zookeeper.ZooKeeper;
import org.apache.zookeeper.data.Stat;

import com.google.protobuf.InvalidProtocolBufferException;

import cp.app.batch.utils.ConfigUtil;
/**
 * hdfs公共服务类
 * @author author
 *
 */
public class HDFSService {
	private static Logger log = Logger.getLogger(HDFSService.class.getName());
	static ConfigUtil conf = new ConfigUtil("conf/zookeeper.properties");
	
	private static final String ZOOKEEPER_IP = conf.getString("ZOOKEEPER_IP");
	private static final int ZOOKEEPER_PORT = conf.getInt("ZOOKEEPER_PORT");
	private static final int ZOOKEEPER_TIMEOUT = conf.getInt("ZOOKEEPER_TIMEOUT");
	private static final String DATA_DIR = conf.getString("DATA_DIR");
     
	/**
	 * 判断文件是否存在
	 * @param dirs hdfs路径
	 * @return 存在ture 否则false
	 */
	public static boolean isExist(List<String> dirs,String ddate) {
		if(dirs==null||dirs.size()<1){
			log.info("Please Check your sign directory configure is correct!");
			return false;
		}
		log.info("Check File or Directory, uri is:"+dirs.toString());
		Configuration conf = new Configuration();
		List<String> flags = new ArrayList<String>();
		FileStatus status=null;
		String hostname = getHostname(ZOOKEEPER_IP,ZOOKEEPER_PORT,ZOOKEEPER_TIMEOUT,DATA_DIR);
		log.info("According to Zookeeper get the active namenode domain:"+hostname);
		try {
			for(String dir:dirs){
				String url = dir.replaceAll("url", hostname);
				FileSystem fs = FileSystem.get(URI.create(url+"dt="+ddate), conf);
				status = fs.getFileStatus(new Path(url+"dt="+ddate));
				if(status==null){
					flags.add("false");
				}else{
					flags.add("true");
				}
				fs.close();
			}
			if(flags.contains("false")){
				return false;
			}else{
				return true;
			}
		} catch (IllegalArgumentException e) {
			// TODO Auto-generated catch block
			log.error(e);
			return false;
		} catch (IOException e) {
			// TODO Auto-generated catch block
			log.error(e);
			return false;
		}		
	}
	
	/**
	 * ͨ通过zookeeper获取active namenode地址
	 * @param ZOOKEEPER_IP ip地址
	 * @param ZOOKEEPER_PORT 端口
	 * @param ZOOKEEPER_TIMEOUT 超时时间 
	 * @return 地址
	 */
	public static String getHostname(String ZOOKEEPER_IP, int ZOOKEEPER_PORT,
			int ZOOKEEPER_TIMEOUT,String DATA_DIR) {
		String hostname = null;
		Watcher watcher = new Watcher() {
			public void process(org.apache.zookeeper.WatchedEvent event) {
				log.info("event:"+event.toString());
			}
		};
		ZooKeeper zk = null;
		byte[] data1 = null;
		String[] iparr = ZOOKEEPER_IP.split(";");
		for (String ip : iparr) {
			try {
				zk = new ZooKeeper(ip + ":" + ZOOKEEPER_PORT,
						ZOOKEEPER_TIMEOUT, watcher);
				data1 = zk.getData(DATA_DIR,
						true, new Stat());
			} catch (Exception e) {
				// TODO Auto-generated catch block
				log.info("This ip is not active..."+ip);
				continue;
			}
			if (data1 != null) {
				log.info("This ip is normal..."+ip);
				ActiveNodeInfo activeNodeInfo=null;
				try {
					activeNodeInfo = HAZKInfoProtos.ActiveNodeInfo.parseFrom(data1);
				} catch (InvalidProtocolBufferException e) {
					// TODO Auto-generated catch block
					log.error(e);
				} 
				hostname = activeNodeInfo.getHostname();
				return hostname;
			} 
		}
		return hostname;
	}
}

zookeeper配置文件zookeeper.properties：

ZOOKEEPER_IP=192.168.1.1;192.168.1.2;192.168.1.3
ZOOKEEPER_PORT=2181
ZOOKEEPER_TIMEOUT=30000
DATA_DIR=/hadoop-ha/hdp/ActiveStandbyElectorLock

读取配置文件工具类：

package com.huateng.spdbccc.mkt24.util;

import java.io.IOException;
import java.io.InputStream;
import java.util.Properties;
/**
 * 配置文件读取工具类
 * @author author
 *
 */

public class ConfigUtil {
	
	private Properties props = new Properties();
	
	public ConfigUtil(String file){
		InputStream is = ConfigUtil.class.getClassLoader().getResourceAsStream(file);
		try {
			props.load(is);
		} catch (IOException e) {
			// TODO Auto-generated catch block
			e.printStackTrace();
		}finally{
			try {
				is.close();
			} catch (IOException e) {
				// TODO Auto-generated catch block
				e.printStackTrace();
			}
		}
	}
	
	public int getInt(String key){
		return Integer.parseInt(props.getProperty(key));
	}
	
	public String getString(String key){
		return props.getProperty(key);
	}

}

hiveserer2的配置文件：

driver=org.apache.hive.jdbc.HiveDriver
url=jdbc:hive2://192.168.1.1:10010/default
user=hdfs
password=""