hadoop和hbase接入snappy编译

1. 下载snappy源码包
   http://pkgs.fedoraproject.org/repo/pkgs/snappy/
2. 编译
   对jdk1.8.0_191/jre/lib/amd64/server/libjvm.so二进制包软链到/usr/local/lib/libjvm.so
   进入对应snappy包内
   ./autogen.sh
   ./configure
   make -j4 && make install
3. 编译hadoop对应的snappy包
   下载
   https://code.google.com/archive/p/hadoop-snappy/source/default/source
   编译
   cd trunk
   mvn clean package -DskipTests -Dsnappy.prefix=/usr/local/
4. 复制hadoop的snappy二进制包到hadoop的lib目录
   hadoop-snappy-0.0.1-SNAPSHOT.tar.gz解压
   cp -r hadoop-snappy-0.0.1-SNAPSHOT/lib/* <HADOOP_HOME>/lib
5. 修改环境变量

hadoop-env.sh

export LD_LIBRARY_PATH=LD_LIBRARY_PATH
export JAVA_LIBRARY_PATH=${HADOOP_HOME}/lib/native/Linux-amd64-64/
测试
/data/hadoop/bin/hadoop checknative -a

6. 对hbase新增snappy压缩
   将hadoop-snappy-0.0.1-SNAPSHOT.jar复制到{HADOOP_HOME}/lib/native LD_LIBRARY_PATH:HBASE_LIBRARY_PATH:$HBASE_HOME/lib/native/
   测试
   bin/hbase --config /data/hbase/conf org.apache.hadoop.util.NativeLibraryChecker