Compiling Hadoop from source code

The procedure is actually very simple, but the document is not very detailed, leading to the whole process of exploration, to organize a share for everyone.


1, download URL:


http://git.apache.org/


2, necessary software maven


This should be noted, do not download the latest 3.1.1, but download 3.0.5, because 3.1.1 a bug can cause trouble.


This is also Red Hat, IBM does not use the latest version of the reason it, known as stable in fact there are major bugs.


http://jira.codehaus.org/browse/MSITE-683


I put maven in usr, take this as an example, modify/etc/profile add:


Export m2_home=/usr/local/apache-maven-3.0.5


export path= $PATH: $M 2_home/bin


3, necessary software protocolbuffer, download 2.5.0


http://code.google.com/p/protobuf/


after decompression


/configure && Make


is very simple, there is nothing to say


3, necessary software findbugs, turn over the wall or think of other ways


http://findbugs.sourceforge.net/


decompression to/usr/local/, modify/etc/profile:


Export findbugs_home=/usr/local/findbugs-2.0.2


export path= $PATH: $FINDBUGS _home/bin


4, Source/etc/profile (do not know how to ask Google)


5, the necessary software cmake, it seems that all Linux distributions have, installation can be


6, enter the download complete Hadoop source code directory:


mvn Package-dskiptests-pdist,native,docs-dtar


maven I'm not very familiar with it, please read its documentation.


7, at this point, you can find the Hadoop-dist/target inside the installation package completed, the next steps and download binary installation is no different.