after moving the includes and excludes file from /root/ to $HADOOP_HOME/conf
problem resolved. really strange .... 在 2012年4月10日 上午10:45，air <cnwe...@gmail.com>写道： > > > ---------- 已转发邮件 ---------- > 发件人： air <cnwe...@gmail.com> > 日期： 2012年4月10日 上午9:46 > 主题： CDH3u3 problem when decommission nodes > 收件人： CDH Users <cdh-u...@cloudera.org> > > > > all operations on the (JT + NN) node. > > I create a file with all (DN + TT) nodes listed in it called hosts.include > I also create a file with all the nodes need to be decommissioned in it > called hosts.exclude > > and then I setting the* mapred.hosts* and *dfs.hosts* to *hosts.include*and > set > * mapred.hosts.exclude* and* dfs.hosts.exclude* to *hosts.exclude* > > and then I restart he JT > > and then refresh nodes with sudo -u hdfs hadoop dfsadmin -refreshNodes > > after that, there is no effect, the cluster still there without any > changes. > > do I forget something ? I successfully decommissioned 3 nodes last week > through the same way , but, why it has no effect today ? > > thank you for your reply in advance ! > -- > 不学习，不知道 > > > > > -- > 不学习，不知道 > > -- 不学习，不知道
