1.需要安装jdk、maven、protoc2.5.0环境,下载hadoop2.2.0、
protoc2.5.0等
源码包
2.jdk、maven安装并配置环境变量很简单,这里不介绍。
3.安装protoc依赖的gcc等支持:
(1)yum install gcc
(2)yum intall gcc-c++
(3)yum install make
(4)yum install cmake
(5)yum install openssl-devel
(6)yum install ncurses-devel
4.安装protoc2.5.0并配置环境变量:
(1)tar -zxvf protobuf-2.5.0.tar.gz
(2)cd protobuf-2.5.0
(3)./configure --prefix=/opt/protoc/
(4)make && make install
(5)vi /etc/profile,在最后面添加如下内容:
export PROTOC_HOME=/opt/protoc
export PATH=$PROTOC_HOME/bin:$PATH
(6)source /etc/profile
5.最新的Hadoop 2.2.0 的Source Code 压缩包解压出来的code有个bug 需要patch后才能编译:
(1)wget https://issues.apache.org/jira/secure/attachment/12614482/HADOOP-10110.patch; patch -p0 < HADOOP-10110.patch
6.编译hadoop:
(1)cd hadoop-2.2.0-src
(2)mvn package -DskipTests -Pdist,native -Dtar
7.替换掉32位的native库,用
./hadoop-dist/target/hadoop-2.2.0/lib/native替换掉
/hadoop-2.2.0/lib/native
(1)
rm -rf ~/local/opt/hadoop-2.2.0/lib/native
(2)cp ./hadoop-dist/target/hadoop-2.2.0/lib/native ~/local/opt/hadoop-2.2.0/lib/
注:确保机器能连接互联网,maven配置的setting.xml文件最好将中心仓库修改为国内maven仓库,因为中心仓库网址在国外访问可能很慢或访问失败