Hadoop Windows Cygwin error tmp folder file permissions

Error
java.io.IOException: Failed to set permissions of path: file:/tmp/hadoop-jj/mapred/staging/jj-1931875024/.staging to 0700
The details for my Hadoop were 0.20.203
Solution
This is known issue in 203 version , just download the previous stable version from hadoop website
OR
Download the latest stable version from Cloudera website , incase you are using that flavour
Details
Also you can read the details for possible causes for the above error at below link
http://lucene.472066.n3.nabble.com/Problem-debugging-MapReduce-job-under-Windows-td2913135.html

 You can also develop using eclipse in Windows , for more details on how to setup eclipse development environment for hadoop please follow the link http://jugnu-life.blogspot.com/2011/11/hadoop-development-environment-in.html

Hadoop Setup in Windows

The below post will explain quickly how to install hadoop on windows for development

I used Windows XP  , you can follow the steps to configure standalone setup for hadoop

 

  1. Download the latest stable release from hadoop website
  2. If you are on Windows you also need Cgywin , download from its website and choose to install openssh and rsync packages
    (When you download hadoop in the file /src/core/overview.html , You can find the list of softwares required )
  3. Add path varible in windows as ;c:\cygwin\bin;c:\cygwin\usr\bin
  4. Copy hadoop zip file to /usr/local directory , rename the hadoop.x.x folder to simple name such as hadoop
  5. To make the JAVA_HOME setting up easy , i copied java directly into usr\local\lib\jdk1.7.0 folder
  6. Open the file called conf/hadoop-env.sh and go to JAVA_HOME variable
    Set something like
    # The java implementation to use.  Required.
    export JAVA_HOME=/usr/local/lib/jdk1.7.0
    If you are copying the JAVA in default windows path like C:\Program Files\Java then You need to escale space with \
  7. To test that configuration has been done properly do this on cgywin command prompt

$> cd /usr/local/hadoop
$> bin/hadoop version

It would show output like

Hadoop 0.20.203.0
Subversion http://svn.apache.org/....
Compiled by XXX on XXX

Congrats :) Its done