Do the 12 hosts have no identity/address known? AFAIK, you need to install Hadoop to a local disk in each of the nodes in order to have separate configuration for each of them. You can run them off a common place too I guess, but the configuration would have to be common always.
By specifying the right hostnames/ip-addresses in the masters and slaves files you should be fine. It could work this way if you just have two machines: masters => snoopy.domain.etc.etc slaves => snoopy.domain.etc.etc bonobo.domain.etc.etc Provided your nodes can talk to each other without issues and ssh works fine between them. If you don't want to use the ssh easiness, you may start the nodes manually on each node via hadoop <node-type> or via the JAR, etc. by logging onto them and issuing so. On Wed, Oct 13, 2010 at 9:24 PM, maha <[email protected]> wrote: > Thanks for the link HTH, that's a great link, but actually hadoop is in my > account, so whether I accessed 'snoopy' or 'booboo' I can see hadoop > installed i. How can I specify that the hadoop in my account is for one of > them and install another for the other computer? is it through the > hadoop.tmp.dir by including 'snoopy.cs.ucsb.edu' and 'booboo.cs.ucsb.edu' as > hosts?master and slave? > > Thanks, > Maha > > > On Oct 12, 2010, at 9:04 PM, Medha Atre wrote: > >> http://www.michael-noll.com/wiki/Running_Hadoop_On_Ubuntu_Linux_%28Multi-Node_Cluster%29 >> >> HTH. >> >> On Tue, Oct 12, 2010 at 6:11 PM, maha <[email protected]> wrote: >>> Hi, >>> >>> I'm trying to run Hadoop on two computers out of 12 network-connected >>> computers. I know only know the 2 computer names ( 'snoopy','booboo') as >>> established by the department and the host name is "cs.ucsb.edu". >>> when I ssh to my account, I want to login to 'snoopy', install hadoop and >>> specify 'booboo' to be a slave. >>> >>> Can any one help? or guide me to a website to read? >>> >>> Thank you, >>> Maha > > -- Harsh J www.harshj.com
