Exec { path => "/bin:/usr/bin", }

#
# Install Hadoop from tarball
#
class hadoop {

  exec { gethadoop:
    command => "wget -nc http://puppetmasterlocation/packages/hadoop-0.20.1.tar.gz",
    cwd => "/usr/local",
    path => "/usr/bin:/usr/sbin:/bin"
  }

  # untar Hadoop, but only do so if a tarball was fetched.
  exec { untar-hadoop:
    command => "tar xzf /usr/local/hadoop-0.20.1.tar.gz",
    cwd => "/usr/local", # put hadoop in /usr/local
    creates => "/usr/local/hadoop-0.20.1",
    subscribe => Exec[gethadoop],
    refreshonly => true;
  }


  # create a nice Hadoop symlink after the tarball is untarred.
  file { "/usr/local/hadoop":
    ensure => "/usr/local/hadoop-0.20.1",
    require => Exec[untar-hadoop];
  }

  # create the Hadoop user and group.
  user { "hadoop":
    shell => "/bin/bash",
    password => 'my-haoop-user-password',
    managehome => "true",
    home => "/home/hadoop",
    ensure => "present";
  }

  #
  # Ensure local directories exist for writing tasks
  #
  file { "/datadrivea/mapred":
    ensure => "directory",
    owner => "hadoop",
    group => "hadoop",
    path => "/datadrivea/mapred",
    recurse => true;
  }

 file { "/datadrivea/mapred/local":
    ensure => "directory",
    owner => "hadoop",
    group => "hadoop",
    path => "/datadrivea/mapred/local",
    recurse => true;
  }
  
  file { "/usr/local/hadoop/logs":
    ensure => "directory",
    owner => "hadoop",
    group => "hadoop",
    path => "/usr/local/hadoop/logs",
    recurse => true;
  }
  
  file { "/tmp/hadoop":
    ensure => "directory",
    owner => "hadoop",
    group => "hadoop",
    path => "/tmp/hadoop",
    recurse => true;
  }
  
  file { "/tmp/hadoop/mapred":
    ensure => "directory",
    owner => "hadoop",
    group => "hadoop",
    path => "/tmp/hadoop/mapred",
    recurse => true;
  }
  
  file { "/tmp/hadoop/mapred/system/":
    ensure => "directory",
    owner => "hadoop",
    group => "hadoop",
    path => "/tmp/hadoop/mapred/system/",
    recurse => true;
  }

#
  # Ensure writeable hdfs directories exist for storage
  #
  file { "/datadrivea/hdfs":
    ensure => "directory",
    owner => "hadoop",
    group => "hadoop",
    path => "/datadrivea/hdfs";
  }

  file { "/datadriveb/hdfs":
    ensure => "directory",
    owner => "hadoop",
    group => "hadoop",
    path => "/datadriveb/hdfs";
  }

  # Hand out authorized key (for hadoop1) which must be installed on all machines in the cluster
  ssh_authorized_key{ "hadoopsshauthkey":
   ensure => present,
   key => "My secret hadoop auth key",
   name => "hadoop@myhadoopserver",
   type => "ssh-dss",
   user => "hadoop",
   target => "/home/hadoop/.ssh/authorized_keys";
  }

  # Copy the Global Relay hadoop config files over the top of the default ones
  exec { gethadoopconfig:
    command => "wget -nd -N -l1 --no-parent -r http://puppetmasterlocation/packages/hadoopconf && rm index.html*",
    cwd => "/usr/local/hadoop/conf",
    user => "hadoop",
    group => "hadoop",
    path => "/usr/bin:/usr/sbin:/bin"
  }

  # change the permissions of the Hadoop installation.
  exec { chown-hadoop:
    command => "chown -R hadoop:hadoop /usr/local/hadoop*",
    require => [ User[hadoop], Exec[gethadoop] ],
    subscribe => Exec[untar-hadoop],
    refreshonly => true;
  }



}