版权声明：本文为博主原创文章，未经博主允许不得转载。欢迎访问我的博客 https://blog.csdn.net/smooth00/article/details/73796622

Loadrunner通过ssh连接调用hadoop的测试Jar包进行基准测试，似乎有点讨巧，而且好像实际意义也不是特别大，但是通过这个方法的展示，能够看到Loadrunner的生命力有多大，而且在linux系统测试和开源技术的测试中，也是能够有用武之地，所以本文算是来个抛砖引玉吧。

1、在loadrunner中新建脚本（本文以LoadRunner11为例），要求选择协议类型为Java->Java Vuser
2、在Run-time Settings设置JDK路径，由于LoadRunner11不支持jdk1.8，本次测试是拷贝了一份低版本的JDK1.6，所以路径选择固定路径模式（Use specified JDK），当然也可以将JDK1.6配到环境变量中，LoadRunner就可以直接调了。

3、上网下载个jsch-0.1.41.jar，然后在LoadRunner中加载jsch的jar包

4、在Loadrunner中以Java Vuser协议创建脚本，脚本样例如下：

/** LoadRunner Java script. (Build: _build_number_)* * Script Description: *                     */
import lrapi.lr;import java.io.BufferedReader;
import java.io.InputStreamReader;
import java.util.Vector;  import com.jcraft.jsch.Channel;
import com.jcraft.jsch.ChannelExec;
import com.jcraft.jsch.JSch;
import com.jcraft.jsch.JSchException;
import com.jcraft.jsch.Session;
import java.io.InputStream;public class Actions
{private String ipAddress;  private String username;  private String password;  public static final int DEFAULT_SSH_PORT = 22;     private Vector<String> stdout; private Session session=null;private Channel channel=null;public void SSHCommandExecutor(final String ipAddress, final String username, final String password) {  this.ipAddress = ipAddress;  this.username = username;  this.password = password;  stdout = new Vector<String>();JSch jsch = new JSch(); try{session = jsch.getSession(username, ipAddress, DEFAULT_SSH_PORT);  session.setPassword(password);session.setConfig("StrictHostKeyChecking", "no");//session.setUserInfo(userInfo);  session.connect();  // Create and connect channel.  channel = session.openChannel("exec");} catch (JSchException e) {  // TODO Auto-generated catch block  e.printStackTrace();  } catch (Exception e) {  e.printStackTrace();  }}  public int CommandExecute(final String command) {  int returnCode = 0;           try {                   ((ChannelExec) channel).setCommand(command);  channel.setInputStream(null);  BufferedReader input = new BufferedReader(new InputStreamReader(channel  .getInputStream()));  channel.connect();  System.out.println("The remote command is: " + command);  // Get the output of remote command.  String line;  while ((line = input.readLine()) != null) {  stdout.add(line);  }  input.close();  // Get the return code only after the channel is closed.  if (channel.isClosed()) {  returnCode = channel.getExitStatus();  }  // Disconnect the channel and session.  //channel.disconnect();  //session.disconnect();  } catch (JSchException e) {  // TODO Auto-generated catch block  e.printStackTrace();  } catch (Exception e) {  e.printStackTrace();  }  return returnCode;  }  public int init() throws Throwable {SSHCommandExecutor("172.17.2.12", "root", "123456");return 0;}//end of initpublic int action() throws Throwable {lr.start_transaction("exe_command");//----------------------------------------写入2个1M的文件---------------------------------/*String commandStr="yarn jar /usr/hdp/2.3.2.0-2950/hadoop-mapreduce/hadoop-mapreduce-client-jobclient-2.7.1.2.3.2.0-2950-tests.jar TestDFSIO -write -nrFiles 2 -fileSize 1";String commandLog="cat -b /root/TestDFSIO_results.log |tail -n10";*///--------------------------------------------TestDFSIO -clean---------------------------//---------------------使用12个mapper和6个reducer来创建12个文件----------------------------String commandStr="yarn jar /usr/hdp/2.3.2.0-2950/hadoop-mapreduce/hadoop-mapreduce-client-jobclient-2.7.1.2.3.2.0-2950-tests.jar nnbench \\";commandStr+="\n -operation create_write -maps 12 -reduces 6 -blockSize 1 \\";commandStr+="\n -bytesToWrite 0 -numberOfFiles 12 -replicationFactorPerFile 3 \\";commandStr+="\n -readFileAfterOpen true -baseDir /benchmarks/NNBench-`hostname -s`";String commandLog="cat -b /root/NNBench_results.log |tail -n30";// ---------------------------------------------------------------------------------------// ------------------------------重复运行一个小作业10次,每次使用2个maps和1个reduces---------/*String commandStr="yarn jar /usr/hdp/2.3.2.0-2950/hadoop-mapreduce/hadoop-mapreduce-client-jobclient-2.7.1.2.3.2.0-2950-tests.jar mrbench -numRuns 10 -maps 2 -reduces 1";String commandLog="";//无日志文件，属于动态输出日志*///----------------------------------------------------------------------------------------CommandExecute(commandStr+" \n "+commandLog);//commandStr表示运行hadoop的命令，commandLog表示要输出的日志for (String str : stdout) {  System.out.println(str);  }lr.end_transaction("exe_command", lr.AUTO);return 0;}//end of actionpublic int end() throws Throwable {try{channel.disconnect();session.disconnect();}catch(Exception e){System.out.println(e);}return 0;}//end of end
}

以上脚本的核心部分是CommandExecute，通过ssh连接linux执行shell命令，一般情况下运行成功后会输出运行日志，但是hadoop基础测试包的调用有些日志是写到后台中的，所以可以通过加上cat -b /root/NNBench_results.log |tail -n30这样的命令（具体需要准确知道日志文件的所在路径），来输出当前执行成功后的日志（比如NNBench测试输出的最后30行就是当前日志）。

5、运行脚本，测试通过的输出如下所示：

6、对于Loadrunner通过C的方式远程连接linux，也是有方法的，具体可以参考别人的文章：

http://blog.csdn.net/hualusiyu/article/details/8530319