On Mon, Jun 1, 2009 at 12:35 AM, HRoger <hanxianyongro...@163.com> wrote: > > You should do thart in the right way as the follow steps: > 1.create a new file named as excludes under $HADOOP_HOME with the datanode > hostname(IP) in it by one name every line.
do these in master? What I mean is that I executes the following commands just in Namenode (JobTracker): $ echo <ip_of_fail_node> > excludes > 2.edit the hadoop-site.xml by adding > <property> > <name>dfs.hosts.exclude</name> > <value>excludes</ value> > </property> > and save it. which hadoop-site.xml I would edit, in Namenode or in Datanode or all nodes? > 3.execute the command "bin/hadoop dfsadmin -refreshNodes" in the namenode > host. > 4.when the step 3 finished,you can run "bin/hadoop dfsadmin -report and > check the result. >
