日期:2014-05-17  浏览次数:20922 次

windows配置hadoop-1.1.0的伪分布式环境

?

1. 安装cygwin

http://cygwin.com/install.html

?

2. cygwin中配置sshd

http://docs.oracle.com/cd/E24628_01/install.121/e22624/preinstall_req_cygwin_ssh.htm#CBHIAFGI

?

3. 伪分布式配置

hadoop-1.1.0/docs/single_node_setup.html

?

bin/hadoop namenode -format

bin/start-all.sh
bin/stop-all.sh

http://localhost:50030
http://localhost:50070

?

Q 在真正运行的时刻会遇到几个问题:

1 设置的路径并非使用cygwin linux的路径。

?hadoop.tmp.dir在/tmp目录下面,理论上应该在C:\cygwin\tmp,但实际的路径确实C:\tmp

2 启动datanode和jobtracker,以及tasktacker时会有路径权限的问题

?

2012-11-25 13:53:05,031 ERROR org.apache.hadoop.security.UserGroupInformation: PriviledgedActionException as:Winseliu cause:java.io.IOException: Failed to set permissions of path: C:\cygwin\home\Winseliu\hadoop-1.1.0\logs\history to 0755
2012-11-25 13:53:05,032 FATAL org.apache.hadoop.mapred.JobTracker: java.io.IOException: Failed to set permissions of path: C:\cygwin\home\Winseliu\hadoop-1.1.0\logs\history to 0755
        at org.apache.hadoop.fs.FileUtil.checkReturnValue(FileUtil.java:689)
        at org.apache.hadoop.fs.FileUtil.setPermission(FileUtil.java:670)
        at org.apache.hadoop.fs.RawLocalFileSystem.setPermission(RawLocalFileSystem.java:509)
        at org.apache.hadoop.fs.RawLocalFileSystem.mkdirs(RawLocalFileSystem.java:344)
        at org.apache.hadoop.fs.FilterFileSystem.mkdirs(FilterFileSystem.java:189)
        at org.apache.hadoop.mapred.JobHistory.init(JobHistory.java:510)

?

2012-11-25 13:53:04,389 INFO org.apache.hadoop.mapred.TaskTracker: Good mapred local directories are: /cygwin/home/Winseliu/cloud/mapred/local
2012-11-25 13:53:04,396 ERROR org.apache.hadoop.mapred.TaskTracker: Can not start task tracker because java.io.IOException: Failed to set permissions of path: \cygwin\home\Winseliu\cloud\mapred\local\taskTracker to 0755
        at org.apache.hadoop.fs.FileUtil.checkReturnValue(FileUtil.java:689)
        at org.apache.hadoop.fs.FileUtil.setPermission(FileUtil.java:670)

?

3 使用jps查不全真正执行的java进程,不知道那个进程启动或未启动

Winseliu@WINSE ~/hadoop-1.1.0
$ jps
6364 NameNode
7168 JobTracker
2692 Jps

Winseliu@WINSE ~/hadoop-1.1.0
$ ps aux | grep java
     7880       1    5544       7028  ?       1001 13:20:40 /cygdrive/c/Java/jdk1.7.0_02/bin/java
     5968       1    7500       4592  ?       1001 13:20:36 /cygdrive/c/Java/jdk1.7.0_02/bin/java
     5784       1     484       6364  pty0    1001 13:20:31 /cygdrive/c/Java/jdk1.7.0_02/bin/java
     6732       1     484       7168  pty0    1001 13:20:38 /cygdrive/c/Java/jdk1.7.0_02/bin/java
     7976       1    5716       5628  ?       1001 13:20:34 /cygdrive/c/Java/jdk1.7.0_02/bin/java
     4492       0       0       4492  pty0    1001   Jan  1 /cygdrive/c/Java/jdk1.7.0_02/bin/java

?

?

R 处理办法:

1 路径不同意,我们就设置自己的目录就可以了

     <property>
         <name>hadoop.tmp.dir</name>
         <value>/cygwin/home/Winseliu/cloud</value>
     </property>

?

2 权限问题,直接修改FileUtils的checkReturnValue()方法,替换hadoop-core-1.1.0.jar中的FileUtils.class文件

  private static void checkReturnValue(boolean rv, File p, 
                                       FsPermission permission
                                       ) throws IOException {
    if (!rv) {
    	// FIXME
			try {
				throw new IOException("Failed to set permissions of path: " + p
						+ " to " + String.format("%04o", permission.toShort()));
			} catch (Exception e) {
				e.printStackTrace();
			}
    }
  }

?

3 查看已经启动的hadoop相关的进程

直接再执行一次start-all.sh,如果会提示让你先stop就说明该进程已经启动了。

Winseliu@WINSE ~
$ cd ha