OK, Thx. Did not know there was a balancer. So adding a server adds also HDFS space right, because this is actually what I first want to do.
2010/10/10 CatOs Mandros <[email protected]> > yes it should if you configured hadoop correctly, just run the hadoop > balancer afterwards to balance the HDFS use. > > $ bin/start-balancer.sh > > > Have a good day :) > > > On Fri, Oct 8, 2010 at 7:36 PM, MilleBii <[email protected]> wrote: > > > for Q2 I'm already using HDFS but in pseudo distributed, ie I have hadoop > > running on a single server. > > So if I had a server it should run ? That would save a lot of time. > > > > -Raymond- > > > > 2010/10/8 CatOs Mandros <[email protected]> > > > > > Hi MilleBii, > > > > > > On Fri, Oct 8, 2010 at 7:21 PM, MilleBii <[email protected]> wrote: > > > > > > > How easy is it scale-up nutch following demand ? > > > > > > > > > > > Its kinda easy when you already have a distributed environement > working. > > > > > > > > > > Q1 : Suppose I have 2 servers and I want to add a 3 third one. Modulo > > the > > > > adequate configuration will hadoop just add the server and manage to > > > > distribute the data on hdfs: or I need to do something special. > > > > > > > > > > You have to add the machine to the cluster and hadoop will pick it up. > > > > > > Just configure the machine, > > > > > > start the services: > > > $ cd path/to/hadoop > > > $ bin/hadoop-daemon.sh start datanode > > > $ bin/hadoop-daemon.sh start tasktracker > > > > > > and you should be up and running. > > > > > > > > > > > > > > > > > Q2 : other question, from single server/pseudo distributed to a 2 > > server > > > > distributed. It is just a matter of configuring hadoop or is there > more > > > to > > > > it ? > > > > > > > > > > > Just configuring hadoop and telling to nutch to use the distributed > mode. > > > You'll have to migrate your data to HDFS tough. > > > > > > -- > > > > -MilleBii- > > > > > > > > > > > > > > > -- > > -MilleBii- > > > -- -MilleBii-
