一.Hadoop支持Snappy

1.重新编译Hadoop 2.7.2源代码,使其支持Snappy解压压缩库  http://blog.itpub.net/30089851/viewspace-2120631/

2.查看libsnappy.so.1.2.0

[root@sht-sgmhadoopnn-01 ~]# ll $HADOOP_HOME/lib/native/

total 4880

-rw-r--r-- 1 root root 1211196 Jun 21 19:11 libhadoop.a

-rw-r--r-- 1 root root 1485756 Jun 21 19:12 libhadooppipes.a

lrwxrwxrwx 1 root root      18 Jun 21 19:45 libhadoop.so -> libhadoop.so.1.0.0

-rwxr-xr-x 1 root root  717060 Jun 21 19:11 libhadoop.so.1.0.0

-rw-r--r-- 1 root root  582128 Jun 21 19:12 libhadooputils.a

-rw-r--r-- 1 root root  365052 Jun 21 19:11 libhdfs.a

lrwxrwxrwx 1 root root      16 Jun 21 19:45 libhdfs.so -> libhdfs.so.0.0.0

-rwxr-xr-x 1 root root  229289 Jun 21 19:11 libhdfs.so.0.0.0

-rw-r--r-- 1 root root  233538 Jun 21 19:11 libsnappy.a

-rwxr-xr-x 1 root root     953 Jun 21 19:11 libsnappy.la

lrwxrwxrwx 1 root root      18 Jun 21 19:45 libsnappy.so -> libsnappy.so.1.2.0

lrwxrwxrwx 1 root root      18 Jun 21 19:45 libsnappy.so.1 -> libsnappy.so.1.2.0

-rwxr-xr-x 1 root root  147726 Jun 21 19:11 libsnappy.so.1.2.0 

 [root@sht-sgmhadoopnn-01 ~]#
###假如集群已经安装好

3.修改$HADOOP_HOME/etc/hadoop/hadoop-env.sh,添加

export HADOOP_OPTS="$HADOOP_OPTS -Djava.library.path=$HADOOP_HOME/lib:$HADOOP_HOME/lib/native"

############会解决Warn:” Unable to load native-hadoop library ”################################################

 [root@sht-sgmhadoopnn-01 ~]# hadoop fs -ls /

  16/06/21 15:08:24 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using     builtin-java classes where applicable

##################################################################################################

4.修改$HADOOP_HOME/etc/hadoop/core-site.xml

5.修改 $HADOOP_HOME/etc/hadoop/mapred-site.xml 中有关压缩属性,测试snappy

6.新增$HADOOP_HOME/etc/hadoop/yarn-site.xml(是否启用日志聚集功能、yarn日志服务的地址、配置yarn的memory和cpu)

点击(此处)折叠或打开

1. <property>
   
2.           <name>yarn.log-aggregation-enable</name>
   
3.          <value>true</value>
   
4. </property>
   
5. <property>
   
6.          <name>yarn.log.server.url</name>
   
7.          <value>http://sht-sgmhadoopnn-01:19888/jobhistory/logs</value>
   
8. </property>
   
9. 
   
10. <property>
    
11.     <name>yarn.nodemanager.resource.memory-mb</name>
    
12.     <value>10240</value>
    
13. </property>
    
14. <property>
    
15.     <name>yarn.scheduler.minimum-allocation-mb</name>
    
16.     <value>1500</value>
    
17.     <discription>单个任务可申请最少内存，默认1024MB</discription>
    
18. </property>
    
19.   
    
20. <property>
    
21.     <name>yarn.scheduler.maximum-allocation-mb</name>
    
22.     <value>2500</value>
    
23.     <discription>单个任务可申请最大内存，默认8192MB</discription>
    
24.  </property>
    
25. <property>
    
26.      <name>yarn.nodemanager.resource.cpu-vcores</name>
    
27.     <value>2</value>
    
28. </property>

7.将hadoop-env.sh, core-site.xml, mapred-site.xml,yarn-site.xml 同步到集群其他节点

8.重启Hadoop集群

9.验证1: hadoop checknative

[root@sht-sgmhadoopnn-01 ~]# hadoop checknative

16/06/25 12:58:13 WARN bzip2.Bzip2Factory: Failed to load/initialize native-bzip2 library system-native, will use pure-Java version

16/06/25 12:58:13 INFO zlib.ZlibFactory: Successfully loaded & initialized native-zlib library

Native library checking:

hadoop:  true /hadoop/hadoop-2.7.2/lib/native/libhadoop.so.1.0.0

zlib:    true /usr/local/lib/libz.so.1

snappy:  true /hadoop/hadoop-2.7.2/lib/native/libsnappy.so.1

lz4:     true revision:99

bzip2:   false

openssl: true /usr/lib64/libcrypto.so

[root@sht-sgmhadoopnn-01 ~]#

####支持本地native,支持snappy

10.验证2

[root@sht-sgmhadoopnn-01 ~]# vi test.log

a

c d

c d d d a

1 2

a

[root@sht-sgmhadoopnn-01 ~]# hadoop fs -mkdir /input

[root@sht-sgmhadoopnn-01 ~]#hadoop fs -put test.log /input/

[root@sht-sgmhadoopnn-01 ~]#hadoop jar /hadoop/hadoop/share/hadoop/mapreduce/hadoop-mapreduce-examples-*.jar wordcount /input /output1

为了验证是否成功，往hdfs上传一个文本文件，敲入一些词组，运行wordcount程序。如果map部分100%完成，即说明我们hadoop snappy安装成功。

因为hadoop没有像HBase一样提供util.CompressionTest类(或者是我没有找到)，所以只能按照这种方法来测试。接下来，将详细列出HBase使用Snappy的配置过程。

二.HBase支持Snappy

1.复制hadoop的native文件夹至hbase的lib目录下

[root@sht-sgmhadoopnn-01 ~]# cp -r $HADOOP_HOME/lib/native $HBASE_HOME/lib/

2.修改$HBASE_HOME/conf/hbase-env.sh,添加

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$HADOOP_HOME/lib/native/

export HBASE_LIBRARY_PATH=$HBASE_LIBRARY_PATH:$HBASE_HOME/lib/native/

3.将hbase-env.sh同步到集群其他节点

4.重启Hbase集群

5.验证1: hbase org.apache.hadoop.hbase.util.CompressionTest hdfs://mycluster/output1/part-r-00000 snappy

 [root@sht-sgmhadoopnn-01 ~]#hbase org.apache.hadoop.hbase.util.CompressionTest hdfs://mycluster/output1/part-r-00000 snappy 

#使用 CompressionTest 来查看snappy是否 enabled 并且能成功 loaded.

/output1/part-r-00000 为我们在验证hadoop snappy的时候，wordcount的输出

6.验证2:建表和put插入数据