关于Linux服务器集群的开机自启动和有序关闭应用的实现

有如下需求：公司一套开发环境多个服务器，比较耗电，决定每天早上开启，下班关机，为了方便，需要实现以下功能：

1、开机时自启动相应服务。

2、关机前有序关闭各应用然后关机。

3、服务器多台，Linux系统，包含oracle服务器、web应用、hbase（Hadoop集群）、svn、中间件、应用支撑等

 

方案如下：

1、各服务器分别配置，Linux开机自启动脚本，修改/etc/rc.d/rc.local加入启动应用的脚本（或者新建Linux服务）。

2、用java语言基于ssh2.jar（远程连接Linux主机并可以通过Linux命令操作远程主机的一个java工具包）、jdom、log4j等jar包，

3、主机列表、关闭应用的命令通过xml配置，通过jdom解析，并有序执行，log4j打印并记录日志。

 

具体实现：

1、ssh调用及运行命令

SSHConnection.java

package com.run.common;

import java.io.BufferedReader;
import java.io.IOException;
import java.io.InputStream;
import java.io.InputStreamReader;
import java.util.ArrayList;
import java.util.List;

import org.apache.log4j.Logger;
import ch.ethz.ssh2.Connection;
import ch.ethz.ssh2.Session;
import ch.ethz.ssh2.StreamGobbler;

public class SSHConnection {

 private static final Logger logger = Logger.getLogger(SSHConnection.class);

 private static Connection getOpenedConnection(String host, String username,
   String password) {
  if (logger.isDebugEnabled()) {
   logger.debug("connecting to " + host + " with user " + username
     + " and pwd " + password);
  }
  Connection conn = new Connection(host);
  boolean isOpen = true;
  try {
   conn.connect(); // make sure the connection is opened
  } catch (IOException e) {
   logger.info("服务器：" + host + "未开机");
   isOpen = false;
  }

  if (isOpen) {
   boolean isAuthenticated = true;
   try {
    isAuthenticated = conn.authenticateWithPassword(username,
      password);
   } catch (IOException e) {
    isAuthenticated = false;
   }
   if (!isAuthenticated) {
    logger.info("用户名或密码错误");
    conn = null;
   }
  } else {
   conn = null;
  }

  return conn;
 }

 public static List<String> runSshCmd(String host, String username,
   String password, String cmd) throws IOException {
  if (logger.isDebugEnabled()) {
   logger.debug("running SSH cmd [" + cmd + "]");
  }
  List<String> list = new ArrayList<String>();
  Connection conn = getOpenedConnection(host, username, password);
  if (conn != null) {
   Session sess = conn.openSession();
   sess.execCommand(cmd);
   InputStream stdout = new StreamGobbler(sess.getStdout());
   BufferedReader br = new BufferedReader(
     new InputStreamReader(stdout));

   while (true) {
    String line = br.readLine();
    if (line == null)
     break;
    list.add(trim(line));
    if (logger.isDebugEnabled()) {
     logger.debug(line);
    }
   }
   sess.close();
   conn.close();
  }

  return list;
 }

 /**
  * @Description: 将字符间空白转换为一个空格
  * @param s
  * @return String
  * @throws
  */
 private static String trim(String s) {
  if (s == null) {
   return null;
  }
  return s.trim().replaceAll("\\s{2,}", " ");
 }

 // 测试
 public static void main(String[] args) throws IOException, Exception {
  List<String> l = SSHConnection.runSshCmd("192.168.1.200", "hadoop",
    "hadoop", "pwd");
  for (String string : l) {
   System.out.println(string);
  }
 }

}

2、CommonUtils.java

package com.run.common;

import java.io.File;
import java.io.InputStream;
import java.net.URL;
import java.util.List;
import java.util.Properties;

import org.dom4j.Document;
import org.dom4j.io.SAXReader;

public class CommonUtils {

 public static List readXml(String url) throws Exception {
  SAXReader saxReader = new SAXReader();
  Document doc = saxReader.read(new File(url));
  return doc.selectNodes("//machine");
 }
 /**
  * 读取xml中某个节点的信息
  * @param url
  * @param node
  * @return
  * @throws Exception
  */
 public static List resolveXml(String url,String node) throws Exception {
  SAXReader saxReader = new SAXReader();
  InputStream in = CommonUtils.class.getResourceAsStream(url);
  Document doc = saxReader.read(in);
  return doc.selectNodes(node);
 }
 /**
  * 读取properties配置文件
  * @param path
  * @param key
  * @return
  * @throws Exception
  */
 public static String getValue(String path,String key) throws Exception {
  InputStream in = CommonUtils.class.getClassLoader().getResourceAsStream(path);
  Properties prop = new Properties();
  prop.load(in);
  return prop.getProperty(key);
 }
 /**
  * 读取commonconfig.properties配置文件
  * @param key
  * @return
  * @throws Exception
  */
 public static String getValue(String key) throws Exception {
  return getValue("commonconfig.properties",key);
 }
 
 public static URL getResource(String path) throws Exception {
  return CommonUtils.class.getClassLoader().getResource(path);
 }
}

3、Main.java程序入口

package com.run.common;

import java.io.IOException;
import java.util.Iterator;
import java.util.List;

import org.apache.log4j.Logger;
import org.dom4j.Element;

public class Main {
 private static final Logger logger = Logger.getLogger(Main.class);

 public static void main(String[] args) {
  try {
   List list = CommonUtils.resolveXml("/shutdown.xml", "//machine");
   Iterator iter = list.iterator();
   while (iter.hasNext()) {
    Element element = (Element) iter.next();
    String host = element.element("ipAddress").getText();
    String userName = element.element("username").getText();
    String password = element.element("password").getText();
    Element cmds = element.element("commands");
    Iterator it = cmds.elementIterator("command");
    while (it.hasNext()) {
     Element cmdEl = (Element) it.next();
     String cmd = cmdEl.getText();

     logger.info(host + " " + userName + " " + password + " "
       + cmd);
     try {
      SSHConnection.runSshCmd(host, userName, password, cmd);
     } catch (IOException ex) {
      logger.debug("服务器：" + host + "关闭异常", ex);
     }

    }
    if (!iter.hasNext()) {
     logger.debug("所有服务器关机完毕！");
    }
   }

   // Thread.sleep(Long.valueOf(CommonUtils.getValue("sleep_time")));
   // System.exit(0);
  } catch (Exception e) {
   logger.debug("解析xml文件异常", e);
  }
 }

}

 

shutdown.xml

<?xml version="1.0" encoding="UTF-8"?>
<machines>
 <!-- 关闭hadoop服务 -->
 <machine>
  <ipAddress>192.168.1.200</ipAddress>
  <username>hadoop</username>
  <password>111111</password>
  <commands>
   <command>sh /home/hadoop/hbase/bin/stop-hbase.sh</command>
   <command>sh /home/hadoop/hadoop/bin/stop-all.sh</command>
  </commands>
 </machine>

 <!-- 关机 -->
 <machine>
  <ipAddress>192.168.1.200</ipAddress>
  <username>root</username>
  <password>1111111</password>
  <commands>
   <command>halt -p</command>
  </commands>
 </machine>

 

 <!-- 关oracle实例 -->
 <machine>
  <ipAddress>192.168.1.207</ipAddress>
  <username>root</username>
  <password>111111</password>
  <commands>
   <command>su - oracle -c "sqlplus / as sysdba @/home/oracle/shutdown.sql"</command>
   <command>halt -p</command>  

  </commands>
 </machine>

……………………
</machines>

 

 commonconfig.properties

sleep_time=5000

由于是公司开发环境需要考虑电费成本，如果是生产环境，这套方案意义不是太大。