[ https://issues.apache.org/jira/browse/HADOOP-7518?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13080084#comment-13080084 ]
Eric Yang commented on HADOOP-7518: ----------------------------------- Hi Vinod, Hadoop trunk supports two layouts. The original layout where common, hdfs, mapreduce are in separated directories. The new layout which merges common, hdfs, mapreduce into a single deploy directory. (Recommend for production usage). For original layout, use these command: {noformat} mvn package -Ptar -Dmaven.test.skip.exec=true ant veryclean tar -Dresolvers=internal export HADOOP_COMMON_HOME=/tmp/hadoop-common-0.23.0-SNAPSHOT export HADOOP_HDFS_HOME=/tmp/hadoop-hdfs-0.23.0-SNAPSHOT {noformat} For the merged layout, use these command: {noformat} mvn package -Pbintar -Dmaven.test.skip.exec=true ant veryclean binary -Dresolvers=internal {noformat} untar both tarballs into the same directory, i.e. /tmp/hadoop. Make sure HADOOP_*_HOME environment are unset. Let me know if you have any questions. Thanks. > Unable to start HDFS cluster on trunk (after the common mavenisation) > --------------------------------------------------------------------- > > Key: HADOOP-7518 > URL: https://issues.apache.org/jira/browse/HADOOP-7518 > Project: Hadoop Common > Issue Type: Bug > Components: scripts > Reporter: Vinod Kumar Vavilapalli > Attachments: Patch > > > This is what I do: > In common directory: > mvn package -Pbintar -Dmaven.test.skip.exec=true > cp target/hadoop-common-0.23.0-SNAPSHOT-bin.tar.gz ~/tmp/common/ > In hdfs directory: > ant veryclean binary -Dresolvers=internal > cp build/hadoop-hdfs-0.23.0-SNAPSHOT-bin.tar.gz ~/tmp/hdfs/ > Untar the tarballs, and start namenode as follows: > {quote} > export > HADOOP_COMMON_HOME=~vinodkv/tmp/common/hadoop-common-0.23.0-SNAPSHOT-bin; > export HADOOP_CONF_DIR=~vinodkv/tmp/conf; > export HADOOP_HDFS_HOME=~vinodkv/tmp/hdfs/hadoop-hdfs-0.23.0-SNAPSHOT; > $HADOOP_COMMON_HOME/sbin/hadoop-daemon.sh start namenode > {quote} > The last one simply says "Hadoop common not found." and exits. -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira