Google Git
Sign in
apache / bigtop / refs/heads/branch-0.6 / . / bigtop-deploy / puppet
tree: 42f5c1c8fe8cb14e0c3225358eba7195c4d52960 [path history] [tgz]
  1. config/
  2. manifests/
  3. modules/
  4. README.md
bigtop-deploy/puppet/README.md

Puppet classes for deploying Hadoop

Configuration

manifests/init.pp expects configuration to live in CSV at $confdir/config/site.csv, which takes the form

An example is provided at config/site.csv.example. These values are loaded using puppet's extlookup() mechanism.

Any options not defined there will revert to a default value defined in manifests/cluster.pp, with the following exceptions (which are required):

  • hadoop_head_node: must be set to the FQDN of the name node of your cluster (which will also become its job tracker and gateway)
  • bigtop_yumrepo_uri: uri of a repository containing packages for hadoop as built by Bigtop.

$confdir is the directory that puppet will look into for its configuration. On most systems, this will be either /etc/puppet/ or /etc/puppetlabs/puppet/. You may override this value by specifying --confdir=path/to/config/dir on the puppet command line.

For other options that may be set here, look for calls to extlookup() in manifests/cluster.pp. Note that if hadoop_storage_dirs is left unset, puppet will attempt to guess which directories to use.

Usage

Make sure that the bigtop-deploy directory is available on every node of your cluster, and then run the following on those nodes: