Googlielmo's blog

In my last post I described the process to successfully build Hadoop on a Windows machine. Before coming to a stable and reproducible procedure I faced some errors which messages don't highlight the real problem. Browsing the web I didn't found so much info about them, so I am going to share the solutions I found. In this list I am going to skip some errors due to missing execution of one or more steps of the overall build procedure (like those due to missing JDK or Maven, or missing paths in the PATH variable, or missing environment variables, etc.) that could be easily fixed. I am still referring to the release 2.7.1 of Hadoop. So most part of the proposed solutions have been tested for this release only. Building through an incompatible version of the .NET framework. Error message : LINK : fatal error LNK1123: failure during conversion to COFF: file invalid or corrupt [C:\OpenSourcePrograms\Hadoop\hadoop-2.7.1-src\hadoop-common-project\hadoop-common\src\main\winutils\w

This post will describe how to build Apache Hadoop ( https://hadoop.apache.org/ ) on a Windows machine. Sometimes you're required to develop or test on Windows machines and in those cases you will need to build Hadoop from scratch because the available pre-built binaries for this OS family don't work fine. The process described in this post has been tested on Windows Vista and Windows Server 2008 64-bit. I started from the current stable version (2.7.1) of the Hadoop source code. Steps:  - Download a JDK 7 (IBM and Oracle work both).  - Set the JAVA_HOME variable.  - Download the Windows SDK 7.1 from one of the following links:     http://www.microsoft.com/en-in/download/details.aspx?id=8442     (DVD ISO)     or     http://www.microsoft.com/en-us/download/confirmation.aspx?id=8279     (Web installer)     and install it. This SDK comes with the .NET framework 4: it is mandatory to use this release in order to successfully complete the Hadoop building process.  - Loc