i guess when you say source code you mean the hadoop binary jar. if not then you never need hadoop source code to run hadoop daemons. You just need the binary jars
yes you will need to put the binaries on all the nodes. ideally the hadoop cluster is set across the same unified environment. so its better you have same OS across all your machines to be on sure side that you don't run into weird issues which you can not isolate from your jobs and hadoop codebase On Tue, Apr 2, 2013 at 7:23 PM, Agarwal, Nikhil <[email protected]>wrote: > Hi,**** > > ** ** > > I wanted to ask that suppose I have Hadoop source code and Jonbtracker > running on my linux machine. If I want to start a TaskTracker daemon on > another machine then do I need to put Hadoop source code in that machine or > can I start TaskTracker without the source code also?**** > > ** ** > > Also, if my master node (where JobTracker is running) is linux then can I > have slave node as windows machine and start Tasktracker on it without > putting Hadoop source code?**** > > ** ** > > Thank you for the help.**** > > ** ** > > Regards,**** > > Nikhil**** > -- Nitin Pawar
