Hao, This only affects your startup, and once its up you should have no further issues, and was fixed via https://issues.apache.org/jira/browse/HDFS-1835.
This was also previously discussed at http://search-hadoop.com/m/0n3W12wbQRQ As a workaround, you can try adding inside conf/hadoop-env.sh: export HADOOP_OPTS="$HADOOP_OPTS -Djava.security.egd=file:/dev/./urandom" On Tue, Feb 14, 2012 at 4:51 AM, Hao Lin <hlin0...@gmail.com> wrote: > Hi, > > I am beginning with Hadoop. I have been experiencing a trouble of waiting a > long time before I can add (put) my input data to HDFS in Hadoop v0.20.2. > > I followed configuration step and format the HDFS. I am able to create new > directory but when I am try to add input data set to DFS, I am stuck by a > very common error: > > "java.io.IOException: File ... could only be replicated to 0 nodes, > instead of 1" > > then I use command "hadoop dfsadmin -report", and it says there is 0 > available datanode. > > Usually, I wait for another 5 or 10 minutes and then datanode is ready and I > can continue to put on the data file. So is there a way that I could avoid > waiting for the 5 minutes? I mean it happens also in the pseudo-cluster > mode. > > Thanks. > > -- > best wishes > > Hao Lin (林昊) > Research Assistant > School of Electrcial & Computer Eng > Purdue University, West Lafayette, IN > Personal Homepage: http://web.ics.purdue.edu/~lin116/ > -- Harsh J Customer Ops. Engineer Cloudera | http://tiny.cloudera.com/about
