簡介

• 啟用lzo

啟用lzo的壓縮方式對於小規模叢集是很有用處，壓縮比率大概能降到原始日誌大小的1/3。同時解壓縮的速度也比較快。

• 安裝lzo

lzo並不是linux系統原生支援，所以需要下載安裝軟體包。這裡至少需要安裝3個軟體包：lzo, lzop, hadoop-gpl-packaging。

• 增加索引

gpl-packaging的作用主要是對壓縮的lzo檔案建立索引，否則的話，無論壓縮檔案是否大於hdfs的block大小，都只會按照預設啟動2個map操作。

部署

安裝lzop native library

[root@localhost
 
 ~]#  wget http://www.oberhumer.com/opensource/lzo/download/lzo-2.06.tar.gz

[root@localhost ~]#  tar -zxvf lzo-2.06.tar.gz

[root@localhost ~]#  cd lzo-2.06

[root@localhost ~]#  export CFLAGS=-m64

[root@localhost ~]#  ./configure -enable-shared -prefix=/usr/local/hadoop/lzo/

[root@localhost ~]# make && sudo make install
 


編譯完lzo包之後，會在/usr/local/hadoop/lzo/生成一些檔案。

將/usr/local/hadoop/lzo目錄下的所有檔案打包，並同步到叢集中的所有機器上。

在編譯lzo包的時候，需要一些環境，可以用下面的命令安裝好lzo編譯環境

[root@localhost ~]#  yum -y install  lzo-devel  zlib-devel  gcc autoconf automake libtool

安裝hadoop-lzo

    這裡下載的是Twitter hadoop-lzo，可以用Maven(如何安裝Maven請參照本部落格的《Linux命令列下安裝Maven與配置》)進行編譯。

[
 
[email protected] ~]#  wget https://github.com/twitter/hadoop-lzo/archive/master.zip

下載後的檔名是master，它是一個zip格式的壓縮包，可以進行解壓：

[[email protected] ~]#  unzip master

解壓後的資料夾名為hadoop-lzo-master

當然，如果你電腦安裝了git，你也可以用下面的命令去下載

[[email protected] ~]#  git clone https://github.com/twitter/hadoop-lzo.git

hadoop-lzo中的pom.xml依賴了hadoop2.1.0-beta，由於我們這裡用到的是Hadoop 2.2.0，所以建議將hadoop版本修改為2.2.0：

<properties>
    <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
    <hadoop.current.version>2.2.0</hadoop.current.version>
    <hadoop.old.version>1.0.4</hadoop.old.version>
</properties>

然後進入hadoop-lzo-master目錄，依次執行下面的命令

[[email protected] ~]# export CFLAGS=-m64

[[email protected] ~]# export CXXFLAGS=-m64

[[email protected] ~]# export C_INCLUDE_PATH=/usr/local/hadoop/lzo/include

[[email protected] ~]# export LIBRARY_PATH=/usr/local/hadoop/lzo/lib

[[email protected] ~]# mvn clean package -Dmaven.test.skip=true

[[email protected] ~]# cd target/native/Linux-amd64-64

[[email protected] ~]# tar -cBf - -C lib . | tar -xBvf - -C ~

[[email protected] ~]# cp ~/libgplcompression* $HADOOP_HOME/lib/native/

[[email protected] ~]# cp target/hadoop-lzo-0.4.18-SNAPSHOT.jar $HADOOP_HOME/share/hadoop/common/

其實在tar -cBf – -C lib . | tar -xBvf – -C ~命令之後，會在~目錄下生成一下幾個檔案：

[[email protected] ~]# ls -l
1-rw-r--r--  1 libgplcompression.a
2-rw-r--r--  1 libgplcompression.la
3lrwxrwxrwx  1 libgplcompression.so -> libgplcompression.so.0.0.0
4lrwxrwxrwx  1 libgplcompression.so.0 -> libgplcompression.so.0.0.0
5-rwxr-xr-x  1 libgplcompression.so.0.0.0

其中libgplcompression.so和libgplcompression.so.0是連結檔案，指向libgplcompression.so.0.0.0，將剛剛生成的libgplcompression*和target/hadoop-lzo-0.4.18-SNAPSHOT.jar同步到叢集中的所有機器對應的目錄。

配置hadoop環境變數

1、在Hadoop中的$HADOOP_HOME/etc/hadoop/hadoop-env.sh加上下面配置：
export LD_LIBRARY_PATH=/usr/local/hadoop/lzo/lib
2、在$HADOOP_HOME/etc/hadoop/core-site.xml加上如下配置：
<property>
    <name>io.compression.codecs</name>
    <value>org.apache.hadoop.io.compress.GzipCodec,
           org.apache.hadoop.io.compress.DefaultCodec,
           com.hadoop.compression.lzo.LzoCodec,
           com.hadoop.compression.lzo.LzopCodec,
           org.apache.hadoop.io.compress.BZip2Codec
        </value>
</property>

<property>
    <name>io.compression.codec.lzo.class</name>
    <value>com.hadoop.compression.lzo.LzoCodec</value>

</property>
3、在$HADOOP_HOME/etc/hadoop/mapred-site.xml加上如下配置
<property>
    <name>mapred.compress.map.output</name>
    <value>true</value>
</property>

<property>
    <name>mapred.map.output.compression.codec</name>
    <value>com.hadoop.compression.lzo.LzoCodec</value>
</property>

<property>
    <name>mapred.child.env</name>
    <value>LD_LIBRARY_PATH=/usr/local/hadoop/lzo/lib</value>
</property>

將剛剛修改的配置檔案全部同步到叢集的所有機器上，並重啟Hadoop叢集，這樣就可以在Hadoop中使用lzo。

驗證lzo（通過hive測試）

建立lzo表

CREATE TABLE lzo (
ip STRING,
user STRING,
time STRING,
request STRING,
status STRING,
size STRING,
rt STRING,
referer STRING,
agent STRING,
forwarded String
)
partitioned by (
date string,
host string
)
row format delimited
fields terminated by '\t'
STORED AS INPUTFORMAT "com.hadoop.mapred.DeprecatedLzoTextInputFormat"
OUTPUTFORMAT "org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat";

匯入資料

LOAD DATA Local INPATH '/home/hadoop/data/access_20151230_25.log.lzo' INTO TABLE lzo PARTITION(date=20151229,host=25);
/home/hadoop/data/access_20151219.log檔案的格式如下:

xxx.xxx.xx.xxx  -       [23/Dec/2015:23:22:38 +0800]    "GET /ClientGetResourceDetail.action?id=318880&token=Ocm HTTP/1.1"   200     199     0.008   "xxx.com"        "Android4.1.2/LENOVO/Lenovo A706/ch_lenovo/80"   "-"
直接採用lzop  /home/hadoop/data/access_20151219.log即可生成lzo格式壓縮檔案/home/hadoop/data/access_20151219.log.lzo

索引LZO檔案

  1. 批量lzo檔案修改

$HADOOP_HOME/bin/hadoop jar 
/home/hadoop/hadoop-2.2.0/share/hadoop/common/hadoop-lzo-0.4.20-SNAPSHOT.jar 
com.hadoop.compression.lzo.DistributedLzoIndexer 
/user/hive/warehouse/lzo

  2. 單個lzo檔案修改

$HADOOP_HOME/bin/hadoop jar 
/home/hadoop/hadoop-2.2.0/share/hadoop/common/hadoop-lzo-0.4.20-SNAPSHOT.jar
com.hadoop.compression.lzo.LzoIndexer
/user/hive/warehouse/lzo/20151228/lzo_test_20151228.lzo

利用hive執行mr任務

set hive.exec.reducers.max=10;
set mapred.reduce.tasks=10;
select ip,rt from nginx_lzo limit 10;
在hive的控制檯能看到類似如下格式輸出，就表示正確了!
hive> set hive.exec.reducers.max=10;
hive> set mapred.reduce.tasks=10;
hive> select ip,rt from lzo limit 10;
Total MapReduce jobs = 1
Launching Job 1 out of 1
Number of reduce tasks is set to 0 since there's no reduce operator
Starting Job = job_1388065803340_0009, Tracking URL = http://mycluster:8088/proxy/application_1388065803340_0009/
Kill Command = /home/hadoop/hadoop-2.2.0/bin/hadoop job -kill job_1388065803340_0009
Hadoop job information for Stage-1: number of mappers: 1; number of reducers: 0
2013-12-27 09:13:39,163 Stage-1 map = 0%, reduce = 0%
2013-12-27 09:13:45,343 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 1.22 sec
2013-12-27 09:13:46,369 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 1.22 sec
MapReduce Total cumulative CPU time: 1 seconds 220 msec
Ended Job = job_1388065803340_0009
MapReduce Jobs Launched:
Job 0: Map: 1 Cumulative CPU: 1.22 sec HDFS Read: 63570 HDFS Write: 315 SUCCESS
Total MapReduce CPU Time Spent: 1 seconds 220 msec
OK
xxx.xxx.xx.xxx "XXX.com"
Time taken: 17.498 seconds, Fetched: 10 row(s)

修改使用中hive表的輸入輸出格式

ALTER TABLE lzo SET FILEFORMAT 
INPUTFORMAT 'com.hadoop.mapred.DeprecatedLzoTextInputFormat' 
OUTPUTFORMAT "org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat" 
SERDE "org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe";