Hey, On Thu, Feb 24, 2011 at 10:13 AM, Adarsh Sharma <adarsh.sha...@orkash.com> wrote: > Dear all, > > I am confused about the concepts used while running map-reduce jobs in > Hadoop Cluster. > I attached a program that is used to run in Hadoop Cluster. Please find the > attachment. > My PATH Variable shows that it includes all libraries as > > [hadoop@cuda1 ~]$ echo $PATH > /usr/kerberos/bin:/usr/local/bin:/bin:/usr/bin:/home/hadoop/project/hadoop-0.20.2/jcuda.jar:/usr/local/cuda/lib:/home/hadoop/bin > [hadoop@cuda1 ~]$
AFAIK, on Linux, java.library.path is loaded from LD_LIBRARY_PATH and on Windows it is PATH. Also, build/lib/native or lib/native folders are added if available to Hadoop's java.library.path if launched using the bin scripts. You can place your native libraries there (under the proper Java platform). -- Harsh J www.harshj.com
