hadoop 2.7.3本地环境运行官方wordcount-基于HDFS

接上篇《hadoop 2.7.3本地环境运行官方wordcount》。继续在本地模式下测试，本次使用hdfs.

2 本地模式使用fs计数wodcount

上面是直接使用的是linux的文件系统。现在使用hadoop fs。在本地模式下，hadoop fs其实也是使用的linux的fs。下面示例说明：

2.1 验证FS

cd /home/jungle/hadoop/hadoop-local
ls -l
total 116
drwxr-xr-x. 2 jungle jungle  4096 Jan  6 15:06 bin
drwxrwxr-x. 4 jungle jungle    31 Jan  6 16:53 dataLocal
drwxr-xr-x. 3 jungle jungle    19 Jan  6 14:56 etc
drwxr-xr-x. 2 jungle jungle   101 Jan  6 14:56 include
drwxr-xr-x. 3 jungle jungle    19 Jan  6 14:56 lib
drwxr-xr-x. 2 jungle jungle  4096 Jan  6 14:56 libexec
-rw-r--r--. 1 jungle jungle 84854 Jan  6 14:56 LICENSE.txt
-rw-r--r--. 1 jungle jungle 14978 Jan  6 14:56 NOTICE.txt
-rw-r--r--. 1 jungle jungle  1366 Jan  6 14:56 README.txt
drwxr-xr-x. 2 jungle jungle  4096 Jan  6 14:56 sbin
drwxr-xr-x. 4 jungle jungle    29 Jan  6 14:56 share

hadoop fs -ls /
Found 20 items
-rw-r--r--   1 root root          0 2016-12-30 12:26 /1
dr-xr-xr-x   - root root      45056 2016-12-30 13:06 /bin
dr-xr-xr-x   - root root       4096 2016-12-29 20:09 /boot
drwxr-xr-x   - root root       3120 2017-01-06 18:31 /dev
drwxr-xr-x   - root root       8192 2017-01-06 18:32 /etc
drwxr-xr-x   - root root         19 2016-11-05 23:38 /home
dr-xr-xr-x   - root root       4096 2016-12-30 12:29 /lib
dr-xr-xr-x   - root root      81920 2016-12-30 13:04 /lib64
drwxr-xr-x   - root root          6 2016-11-05 23:38 /media
# ...

# 等同 ls -l  /home/jungle/hadoop/hadoop-local
hadoop fs -ls /home/jungle/hadoop/hadoop-local
Found 11 items
-rw-r--r--   1 jungle jungle      84854 2017-01-06 14:56 /home/jungle/hadoop/hadoop-local/LICENSE.txt
-rw-r--r--   1 jungle jungle      14978 2017-01-06 14:56 /home/jungle/hadoop/hadoop-local/NOTICE.txt
-rw-r--r--   1 jungle jungle       1366 2017-01-06 14:56 /home/jungle/hadoop/hadoop-local/README.txt
drwxr-xr-x   - jungle jungle       4096 2017-01-06 15:06 /home/jungle/hadoop/hadoop-local/bin
drwxrwxr-x   - jungle jungle         31 2017-01-06 16:53 /home/jungle/hadoop/hadoop-local/dataLocal
drwxr-xr-x   - jungle jungle         19 2017-01-06 14:56 /home/jungle/hadoop/hadoop-local/etc
drwxr-xr-x   - jungle jungle        101 2017-01-06 14:56 /home/jungle/hadoop/hadoop-local/include
drwxr-xr-x   - jungle jungle         19 2017-01-06 14:56 /home/jungle/hadoop/hadoop-local/lib
drwxr-xr-x   - jungle jungle       4096 2017-01-06 14:56 /home/jungle/hadoop/hadoop-local/libexec
drwxr-xr-x   - jungle jungle       4096 2017-01-06 14:56 /home/jungle/hadoop/hadoop-local/sbin
drwxr-xr-x   - jungle jungle         29 2017-01-06 14:56 /home/jungle/hadoop/hadoop-local/share

从上面可以看出。hadoop fs -ls /home/jungle/hadoop/hadoop-local和linux的命令ls /home/jungle/hadoop/hadoop-local是等效的。

2.2 准备数据

下面基于上次实例的原始数据，将其copy到hdfs上。

hadoop fs -mkdir -p ./dataHdfs/input 

hadoop fs -ls .
Found 12 items
drwxrwxr-x   - jungle jungle         18 2017-01-06 18:44 dataHdfs
drwxrwxr-x   - jungle jungle         31 2017-01-06 16:53 dataLocal
# ...

hadoop fs -ls ./dataHdfs/
Found 1 items
drwxrwxr-x   - jungle jungle          6 2017-01-06 18:44 dataHdfs/input

hadoop fs -put
-put: Not enough arguments: expected 1 but got 0
Usage: hadoop fs [generic options] -put [-f] [-p] [-l] <localsrc> ... <dst>

# 将本地文件，put到hdfs上，实际效果等同于linux下的copy
hadoop fs -put dataLocal/input/ ./dataHdfs/
ls -l dataHdfs/
total 0
drwxrwxr-x. 2 jungle jungle 80 Jan  6 18:51 input

ls -l dataHdfs/input/
total 8
-rw-r--r--. 1 jungle jungle 37 Jan  6 18:51 file1.txt
-rw-r--r--. 1 jungle jungle 70 Jan  6 18:51 file2.txt

hadoop fs -ls  ./dataHdfs/
Found 1 items
drwxrwxr-x   - jungle jungle         80 2017-01-06 18:51 dataHdfs/input

hadoop fs -ls  ./dataHdfs/input/
Found 2 items
-rw-r--r--   1 jungle jungle         37 2017-01-06 18:51 dataHdfs/input/file1.txt
-rw-r--r--   1 jungle jungle         70 2017-01-06 18:51 dataHdfs/input/file2.txt

2.3 执行wordcount

hadoop jar /home/jungle/hadoop/hadoop-local/share/hadoop/mapreduce/hadoop-mapreduce-examples-2.7.3.jar wordcount dataHdfs/input/ dataHdfs/output
# 这里的input, output目录，即可以理解成hdfs里的目录，也可以理解成linux里的目录。 

cat dataHdfs/output/part-r-00000
I	1
am	1
bye	2
great	1
hadoop.	3
hello	3
is	1
jungle.	2
software	1
the	1
world.	2

md5sum dataLocal/outout/part-r-00000  dataHdfs/output/part-r-00000
68956fd01404e5fc79e8f84e148f19e8  dataLocal/outout/part-r-00000
68956fd01404e5fc79e8f84e148f19e8  dataHdfs/output/part-r-00000

可见与上篇中 dataLocal/下的结果是相同的。