Hadoop2.x源码－编译剖析

1.概述

　　最近，有小伙伴涉及到源码编译。然而，在编译期间也是遇到各种坑，在求助于搜索引擎，技术博客，也是难以解决自身所遇到的问题。笔者在被询问多次的情况下，今天打算为大家来写一篇文章来剖析下编译的细节，以及遇到编译问题后，应该如何去解决这样类似的问题。因为，编译的问题，对于后期业务拓展，二次开发，编译打包是一个基本需要面临的问题。

2.编译准备

　　在编译源码之前，我们需要准备编译所需要的基本环境。下面给大家列举本次编译的基础环境，如下所示：

• 硬件环境

操作系统	CentOS6.6
CPU	I7
内存	16G
硬盘	闪存
核数	4核

• 软件环境

JDK	1.7
Maven	3.2.3
ANT	1.9.6
Protobuf	2.5.0

　　在准备好这些环境之后，我们需要去将这些环境安装到操作系统当中。步骤如下：

2.1 基础环境安装

　　关于JDK，Maven，ANT的安装较为简单，这里就不多做赘述了，将其对应的压缩包解压，然后在/etc/profile文件当中添加对应的路径到PATH中即可。下面笔者给大家介绍安装Protobuf，其安装需要对Protobuf进行编译，故我们需要编译的依赖环境gcc、gcc-c++、cmake、openssl-devel、ncurses-devel，安装命令如下所示：

yum -y install gcc
yum -y install gcc-c++
yum -y install cmake
yum -y install openssl-devel
yum -y install ncurses-devel

　　验证GCC是否安装成功，命令如下所示：

　　验证Make核CMake是否安装成功，命令如下所示：

　　在准备完这些环境之后，开始去编译Protobuf，编译命令如下所示：

[hadoop@nna ~]$ cd protobuf-2.5./
[hadoop@nna protobuf-2.5.]$ ./configure --prefix=/usr/local/protoc
[hadoop@nna protobuf-2.5.]$ make
[hadoop@nna protobuf-2.5.]$ make install

　　PS：这里安装的时候有可能提示权限不足，若出现该类问题，使用sudo进行安装。

　　验证Protobuf安装是否成功，命令如下所示：

　　下面，我们开始进入编译环境，在编译的过程当中会遇到很多问题，大家遇到问题的时候，要认真的去分析这些问题产生的原因，这里先给大家列举一些可以避免的问题，在使用Maven进行编译的时候，若使用默认的JVM参数，在编译到hadoop-hdfs模块的时候，会出现溢出现象。异常信息如下所示：

java.lang.OutOfMemoryError: Java heap space 

　　这里，我们在编译Hadoop源码之前，可以先去环境变量中设置其参数即可，内容修改如下：

export MAVEN_OPTS="-Xms256m -Xmx512m"

　　接下来，我们进入到Hadoop的源码，这里笔者使用的是Hadoop2.6的源码进行编译，更高版本的源码进行编译，估计会有些许差异，编译命令如下所示：

[hadoop@nna tar]$ cd hadoop-2.6.-src
[hadoop@nna tar]$ mvn package -DskipTests -Pdist,native

　　PS：这里笔者是直接将其编译为文件夹，若需要编译成tar包，可以在后面加上tar的参数，命令为 mvn package -DskipTests -Pdist,native -Dtar

　　笔者在编译过程当中，出现过在编译KMS模块时，下载tomcat不完全的问题，Hadoop采用的tomcat是apache-tomcat-6.0.41.tar.gz，若是在此模块下出现异常，可以使用一下命令查看tomcat的文件大小，该文件正常大小为6.9M左右。查看命令如下所示：

[hadoop@nna downloads]$ du -sh *

　　若出现只有几K的tomcat安装包，表示tomcat下载失败，我们将其手动下载到/home/hadoop/tar/hadoop-2.6.0-src/hadoop-common-project/hadoop-kms/downloads目录下即可。在编译成功后，会出现以下信息：

[INFO] ------------------------------------------------------------------------
[INFO] Reactor Summary:
[INFO]
[INFO] Apache Hadoop Main ................................. SUCCESS [  1.162 s]
[INFO] Apache Hadoop Project POM .......................... SUCCESS [  0.690 s]
[INFO] Apache Hadoop Annotations .......................... SUCCESS [  1.589 s]
[INFO] Apache Hadoop Assemblies ........................... SUCCESS [  0.164 s]
[INFO] Apache Hadoop Project Dist POM ..................... SUCCESS [  1.064 s]
[INFO] Apache Hadoop Maven Plugins ........................ SUCCESS [  2.260 s]
[INFO] Apache Hadoop MiniKDC .............................. SUCCESS [  1.492 s]
[INFO] Apache Hadoop Auth ................................. SUCCESS [  2.233 s]
[INFO] Apache Hadoop Auth Examples ........................ SUCCESS [  2.102 s]
[INFO] Apache Hadoop Common ............................... SUCCESS [: min]
[INFO] Apache Hadoop NFS .................................. SUCCESS [  3.891 s]
[INFO] Apache Hadoop KMS .................................. SUCCESS [  5.872 s]
[INFO] Apache Hadoop Common Project ....................... SUCCESS [  0.019 s]
[INFO] Apache Hadoop HDFS ................................. SUCCESS [: min]
[INFO] Apache Hadoop HttpFS ............................... SUCCESS [: min]
[INFO] Apache Hadoop HDFS BookKeeper Journal .............. SUCCESS [: min]
[INFO] Apache Hadoop HDFS-NFS ............................. SUCCESS [  2.492 s]
[INFO] Apache Hadoop HDFS Project ......................... SUCCESS [  0.020 s]
[INFO] hadoop-yarn ........................................ SUCCESS [  0.018 s]
[INFO] hadoop-yarn-api .................................... SUCCESS [: min]
[INFO] hadoop-yarn-common ................................. SUCCESS [: min]
[INFO] hadoop-yarn-server ................................. SUCCESS [  0.029 s]
[INFO] hadoop-yarn-server-common .......................... SUCCESS [: min]
[INFO] hadoop-yarn-server-nodemanager ..................... SUCCESS [: min]
[INFO] hadoop-yarn-server-web-proxy ....................... SUCCESS [  1.810 s]
[INFO] hadoop-yarn-server-applicationhistoryservice ....... SUCCESS [  4.041 s]
[INFO] hadoop-yarn-server-resourcemanager ................. SUCCESS [ 11.739 s]
[INFO] hadoop-yarn-server-tests ........................... SUCCESS [  3.332 s]
[INFO] hadoop-yarn-client ................................. SUCCESS [  4.762 s]
[INFO] hadoop-yarn-applications ........................... SUCCESS [  0.017 s]
[INFO] hadoop-yarn-applications-distributedshell .......... SUCCESS [  1.586 s]
[INFO] hadoop-yarn-applications-unmanaged-am-launcher ..... SUCCESS [  1.233 s]
[INFO] hadoop-yarn-site ................................... SUCCESS [  0.018 s]
[INFO] hadoop-yarn-registry ............................... SUCCESS [  3.270 s]
[INFO] hadoop-yarn-project ................................ SUCCESS [  2.164 s]
[INFO] hadoop-mapreduce-client ............................ SUCCESS [  0.032 s]
[INFO] hadoop-mapreduce-client-core ....................... SUCCESS [ 13.047 s]
[INFO] hadoop-mapreduce-client-common ..................... SUCCESS [ 10.890 s]
[INFO] hadoop-mapreduce-client-shuffle .................... SUCCESS [  2.534 s]
[INFO] hadoop-mapreduce-client-app ........................ SUCCESS [  6.429 s]
[INFO] hadoop-mapreduce-client-hs ......................... SUCCESS [  4.866 s]
[INFO] hadoop-mapreduce-client-jobclient .................. SUCCESS [: min]
[INFO] hadoop-mapreduce-client-hs-plugins ................. SUCCESS [  1.183 s]
[INFO] Apache Hadoop MapReduce Examples ................... SUCCESS [  3.655 s]
[INFO] hadoop-mapreduce ................................... SUCCESS [  1.775 s]
[INFO] Apache Hadoop MapReduce Streaming .................. SUCCESS [ 11.478 s]
[INFO] Apache Hadoop Distributed Copy ..................... SUCCESS [ 15.399 s]
[INFO] Apache Hadoop Archives ............................. SUCCESS [  1.359 s]
[INFO] Apache Hadoop Rumen ................................ SUCCESS [  3.736 s]
[INFO] Apache Hadoop Gridmix .............................. SUCCESS [  2.822 s]
[INFO] Apache Hadoop Data Join ............................ SUCCESS [  1.791 s]
[INFO] Apache Hadoop Ant Tasks ............................ SUCCESS [  1.350 s]
[INFO] Apache Hadoop Extras ............................... SUCCESS [  1.858 s]
[INFO] Apache Hadoop Pipes ................................ SUCCESS [  5.805 s]
[INFO] Apache Hadoop OpenStack support .................... SUCCESS [  3.061 s]
[INFO] Apache Hadoop Amazon Web Services support .......... SUCCESS [: min]
[INFO] Apache Hadoop Client ............................... SUCCESS [  2.986 s]
[INFO] Apache Hadoop Mini-Cluster ......................... SUCCESS [  0.053 s]
[INFO] Apache Hadoop Scheduler Load Simulator ............. SUCCESS [  2.917 s]
[INFO] Apache Hadoop Tools Dist ........................... SUCCESS [  5.702 s]
[INFO] Apache Hadoop Tools ................................ SUCCESS [  0.015 s]
[INFO] Apache Hadoop Distribution ......................... SUCCESS [  8.587 s]
[INFO] ------------------------------------------------------------------------
[INFO] BUILD SUCCESS
[INFO] ------------------------------------------------------------------------
[INFO] Total time: : min
[INFO] Finished at: --22T15::+:
[INFO] Final Memory: 89M/451M
[INFO] ------------------------------------------------------------------------

　　在编译完成之后，会在Hadoop源码的dist目录下生成编译好的文件，如下图所示：

　　图中hadoop-2.6.0即表示编译好的文件。

3.总结

　　在编译的过程当中，会出现各种各样的问题，有些问题可以借助搜索引擎去帮助我们解决，有些问题搜索引擎却难以直接的给出解决方案，这时，我们需要冷静的分析编译错误信息，大胆的去猜测，然后去求证我们的想法。简而言之，解决问题的方法是有很多的。当然，大家也可以在把遇到的编译问题，贴在评论下方，供后来者参考或借鉴。

4.结束语

　　这篇文章就和大家分享到这里，如果大家在研究和学习的过程中有什么疑问，可以加群进行讨论或发送邮件给我，我会尽我所能为您解答，与君共勉！