Preface

The hadoop-bigtop Roll installs and configures hadoop.

• Cluster Planning: To understand the default value of some attributes, this roll roll assumes the following minimal partition scheme by default:

  / (reformatted at kickstart) /var (reformatted at kickstart) /scratch (a partition to host local dirs to be shared in cluster) /hadoop (a partition to be used by hadoop)

  To change the default partition/dir used by hadoop, set changes in rocks attrs BIGTOP_HadoopDataDir and/or BIGTOP_HadoopData. For example

  ###global #/opt/rocks/bin/rocks set attr BIGTOP_HadoopDataDir value="/bigdata" ### or host specific #/opt/rocks/bin/rocks set host attr compute-0-0 BIGTOP_HadoopDataDir value="/bigdata" #/opt/rocks/bin/rocks set host attr compute-0-0 BIGTOP_HadoopData value="/bigdata/mydata"

Please visit the hadoop-bigtop site to learn more about their release and the individual software components.