實做四: Hadoop 程式編譯

範例練習 0 使用範例

$ cd /opt/hadoop

• 下載 nchc-example.jar

  $ wget http://trac.nchc.org.tw/cloud/raw-attachment/wiki/Hadoop_Lab4/nchc-example.jar
  

• 執行客製化的hadoop 程式

$ bin/hadoop jar nchc-example.jar

• 輸出結果

  ******************************************
  歡迎使用 NCHC.Hadoop 的運算功能
  指令：
    Hadoop jar nchc-example-*.jar <功能>
  功能：
    hello:     印出key-value的內容
    wordcount: 計算輸入資料夾內分別在每個檔案的字數統計
    mwc:       統合計算所有輸入檔的字數統計
    grep:      算出包指定字串的個數
    nchcgrep:  整合來源檔內的每個字與其所有出現的所在列
  ******************************************
  

以 Hello 為例

code

• 使用hello

  $ bin/hadoop jar nchc-example.jar hello
  

• 使用提示：

  hello <inDir> <outDir> <m> <r>
  

• 運作：

$ bin/hadoop dfs -put conf-local conf_input
$ bin/hadoop jar nchc-example.jar hello conf_input lab4_out6 1 1

• 看結果

練習 1 : Word Count 編譯+執行

code

• 上傳內容到hdfs內

$ cd /opt/hadoop
$ mkdir lab4_input
$ echo "I like NCHC Cloud Course." > lab4_input/input1
$ echo "I like nchc Cloud Course, and we enjoy this course." > lab4_input/input2
$ bin/hadoop fs -put lab4_input lab4_input
$ bin/hadoop fs -ls lab4_input

• 下載 WordCount.java 並存到/opt/hadoop;

  $ wget http://trac.nchc.org.tw/cloud/raw-attachment/wiki/Hadoop_Lab4/WordCount.java
  

• 編譯

$ mkdir MyJava
$ javac -classpath hadoop-*-core.jar -d MyJava WordCount.java
$ jar -cvf wordcount.jar -C MyJava .

• 運算

  $ bin/hadoop jar wordcount.jar WordCount lab4_input/ lab4_out1/
  $ bin/hadoop fs -cat lab4_out1/part-r-00000
  

• lab4_out1 執行結果

  Cloud 2
  Course, 1
  Course. 1
  I 2
  NCHC  1
  and 1
  course. 1
  enjoy 1
  like  2
  nchc  1
  this  1
  we  1
  

---

練習 2 : Word Count 進階版

code

$ echo "\." >pattern.txt && echo "\," >>pattern.txt
$ bin/hadoop fs -put pattern.txt ./
$ mkdir MyJava2

• 下載 WordCount2.java 並存到/opt/hadoop;

  $ wget http://trac.nchc.org.tw/cloud/raw-attachment/wiki/Hadoop_Lab4/WordCount2.java
  

$ javac -classpath hadoop-*-core.jar -d MyJava2 WordCount2.java
$ jar -cvf wordcount2.jar -C MyJava2 .
$ bin/hadoop jar wordcount2.jar WordCount2 lab4_input lab4_out2 -skip pattern.txt
$ bin/hadoop fs -cat lab4_out2/part-00000

• lab4_out2 執行結果

  Cloud 2
  Course  2
  I 2
  NCHC  1
  and 1
  course  1
  enjoy 1
  like  2
  nchc  1
  this  1
  we  1
  

$ bin/hadoop jar wordcount2.jar WordCount2 -Dwordcount.case.sensitive=false lab4_input lab4_out3 -skip pattern.txt
$ bin/hadoop fs -cat lab4_out3/part-00000

• lab4_out3 執行結果

  and 1
  cloud 2
  course  3
  enjoy 1
  i 2
  like  2
  nchc  2
  this  1
  we  1