c++ - 在 Hadoop 2.x 中运行 C++ 代码

Question

我已经在CentOS VM中安装了hadoop 2.5.2。我对 hadoop 很陌生，正在尝试基于 tutorials 在 hadoop 2.x 中执行 C++ 代码

我发现在 Hadoop 2.x 版本中，没有文件夹 (HADOOP_INSTALL)/c++/$(PLATFORM)/。我看到 $(HADOOP_INSTALL)/include 下有包含文件，以及 $HADOOP_INSTALL/lib/native/libhadooppipes.a 下的 hadoopipes.a 等库文件。我像这样调整了我的 makefile:

 CC = g++
    HADOOP_INSTALL = /usr/local/hadoop
    CPPFLAGS = -m32 -I$(HADOOP_INSTALL)/include

    wordcount: WordCount.cpp
        $(CC) $(CPPFLAGS) $< -Wall -L$(HADOOP_INSTALL)/lib/native -lhadooppipes \
        -lhadooputils -lpthread -g -O2 -o $@

and changed in code ( rest of code is same as in the link above )

      #include "Pipes.hh" 
      #include "TemplateFactory.hh" 
      #include "StringUtils.hh"

when I compile I get
    $ make wordcount
    g++ -m32 -I/usr/local/hadoop/include WordCount.cpp -Wall -L/usr/local/hadoop/lib/native -lhadooppipes \
        -lhadooputils -lpthread -g -O2 -o wordcount
    /usr/bin/ld: skipping incompatible /usr/local/hadoop/lib/native/libhadooppipes.a when searching for -lhadooppipes
    /usr/bin/ld: cannot find -lhadooppipes
    collect2: ld returned 1 exit status

谁能指点一下如何在 hadoop 2.x 上编译 c++。 (不过我可以在hadoop 1.x中执行c++程序。我感兴趣的是，如何在hadoop 2.x中使用hadoop管道执行c++程序)提前感谢

Answer 1

问题:

How to compile 32 bit binary by linking it to hadoop archive ( file format elf64-x86-64 architecture: i386:x86-64).Please advice

答案:为此，您需要multilib。如果它已经安装在 Ubuntu 14.04 或 Arch Linux 上，则必须在 64 位系统上启用 32 位应用程序。

例如，像这样source说-

Enabling the multilib repository allows the user to run and build 32-bit applications on 64-bit installations of Arch Linux. multilib creates a directory containing 32-bit instruction set libraries inside /usr/lib32/, which 32-bit binary applications may need when executed.

否则，您必须安装multilib -

sudo apt-get install gcc-multilib

和

sudo apt-get install ia32-libs-dev

这个link也可以提供帮助。 Google 还提供了有关多库支持的更多信息。