本视频适合对Java有一定了解,熟悉java se的Hadoop爱好者,想对Hadoop进行二次开发。下面是以伪分布为例:
想对Hadoop二次开发:
一.首先需要Hadoop和Java之间搭建Eclipse搭建一个桥梁,也就是所谓的插件。
这个插件是由Hadoop来提供的。只要安装Hadoop,就可以在Hadoop安装目录中找到Hadoop-xx-eclipse-pligin.jar。
<ignore_js_op>
![](http://www.aboutyun.com/data/attachment/forum/201401/09/213138owr4cjl9qzlqqqcu.png)
图1
二.把这个插件添加到Eclipse安装目录的插件文件夹
<ignore_js_op>
![](http://www.aboutyun.com/data/attachment/forum/201401/09/213626f0f7jqq1j7j601kk.png)
图2
在复制的过程中注意权限问题
三.复制完成之后,需要重启Eclipse。然后定位Hadoop安装目录。
<ignore_js_op>
![](http://www.aboutyun.com/data/attachment/forum/201401/09/213947flthkr2ngc9a2g4r.png)
图3
(1)选择Hadoop安装目录
<ignore_js_op>
![](http://www.aboutyun.com/data/attachment/forum/201401/09/213244ct55tlss8ltf5trt.png)
图4
四.调出map/reduce location信息窗口
(1)通过如下操作添加map/reduce视图
<ignore_js_op>
![](http://www.aboutyun.com/data/attachment/forum/201401/09/214846z22qso2qs0ywn0ss.png)
图5
<ignore_js_op>
![](http://www.aboutyun.com/data/attachment/forum/201401/09/215014e71q71eb1ampg1q1.png)
图6
(2)通过图7,图8操作显示map/reduce location信息窗口
<ignore_js_op>
![](http://www.aboutyun.com/data/attachment/forum/201401/09/215344q992r69jjdmj3359.png)
图7
<ignore_js_op>
![](http://www.aboutyun.com/data/attachment/forum/201401/09/215341ig5nz114n7kny14k.png)
图8
<ignore_js_op>
![](http://www.aboutyun.com/data/attachment/forum/201401/09/215501zhzada1i1ei9lc77.png)
图9
图9就可以看到map/reduce location信息窗口
五连接Hadoop
这一步主要配置开发那一台机器的Hadoop。通过new hadoop location,弹出信息配置窗口。
这里需要注意的是,对于伪分布这里填写的location,远程的需要填写ip,端口的配置要与map-site一致,否则会出错。
图10
<ignore_js_op>
![](http://www.aboutyun.com/data/attachment/forum/201401/09/220014u3hz1a2zhsokea3a.png)
图11
<ignore_js_op>
![](http://www.aboutyun.com/data/attachment/forum/201401/09/220131l0dd2fu2kwkinokb.png)
图12
<ignore_js_op>
![](http://www.aboutyun.com/data/attachment/forum/201401/09/220023vi2u4ux4pzcz8c1w.png)
图13<ignore_js_op>
![](http://www.aboutyun.com/data/attachment/forum/201401/09/220026py3ecyd7zytdd3t8.png)
图14
六连接成功
连接则会出现图15所示
<ignore_js_op>
![](http://www.aboutyun.com/data/attachment/forum/201401/09/220845zydnsyngytvsdn6s.png)