On Feb 5, 2006, at 9:35 AM, Bernd Fehling wrote:
A basic question:
Is it possible to install nutch with NDFS on a single machine
or do I need at least two maschines?
Yes, it is possible. I just ran a Hadoop map/reduce example on a
single machine using Hadoop DFS. On a single node, I ran one instance
of all 4 servers (namenode, datanode, jobtracker, and tasktracker). I
was able to run a map/reduce application with reading the inputs from
DFS and writing the output to DFS.
Note that effectively, this configuration is only useful for testing
because you are wasting time using the distributed framework for a
single node. As a test, it was very useful. *smile*
Note that you do want to change the value of dfs.replication to 1.
I've never run the indexing part of Nutch, so I can't help you on that
side. The Hadoop framework works fine in that configuration.
-- Owen
