Installs and configures collectd. Much of the work in this cookbook reflects work done by coderanger and realityforge.

• Amazon 2012.09
• CentOS 6
• CentOS 7
• Ubuntu 12.04
• Ubuntu 14.04

• build-essential
• yum

• node["collectd"]["version"] - Version of collectd to install.
• node["collectd"]["dir"] - Base directory for collectd.
• node["collectd"]["plugins_conf_dir"]- Plugin directory for collectd.
• node["collectd"]["url"] - URL to the collectd archive.
• node["collectd"]["checksum"] - Checksum for the collectd archive.
• node["collectd"]["interval"] - Number of seconds to wait between data reads.
• node["collectd"]["read_threads"] - Number of threads performing data reads.
• node["collectd"]["write_queue_limit_high"] - Upper bound on write queue size.
• node["collectd"]["write_queue_limit_low"] - Lower bound on write queue size.
• node["collectd"]["collect_internal_stats"] - Flag to collect internal collectd statistics.
• node["collectd"]["name"] - Name of the node reporting statstics.
• node["collectd"]["fqdnlookup"] - Flag to determine if the node should determine its own FQDN.
• node["collectd"]["plugins"] - Mash of plugins for installation.
• node["collectd"]["python_plugins"] - Mash of Python plugins for installation.
• node["collectd"]["plugins_conf_dir"] - Directory for collectd plugins configuration files.
• node["collectd"]["graphite_role"] – Role assigned to Graphite server for search.
• node["collectd"]["graphite_ipaddress"] – IP address to Graphite server if you're trying to target one that isn't searchable.
• node["collectd"]["packages"] – List of collectd packages.
• node["collectd"]["configure_flag"] – Flag for enabling non-default collectd packages.

• recipe[collectd-ng] will install collectd from source.
• recipe[collectd-ng::attribute_driven] will install collectd via node attributes.
• recipe[collectd-ng::packages] will install collectd (and other plugins) from packages.
• recipe[collectd-ng::recompile] will attempt to recompile collectd.

Note: You need to include the default recipe[collectd-ng] when using recipe[collectd-ng:attribute_driven] to install collectd.

By default this cookbook will attempt to download collectd from collectd.org. If your HTTP request includes Chef as the user agent, collectd.org returns an HTTP response with a message asking you to please stop using their downloads via Chef. It is highly recommended that you override node["collectd"]["url"] with your own download location for collectd.

A list of alternative download locations for collectd:

• https://s3.amazonaws.com/collectd-5.4.1/collectd-5.4.1.tar.gz (@ahochman)

In order to configure collectd via attributes, setup your roles like:

default_attributes(
  "collectd" => {
    "plugins" => {
      "syslog" => {
        "config" => { "LogLevel" => "Info" }
      },
      "disk"      => { },
      "swap"      => { },
      "memory"    => { },
      "cpu"       => { },
      "interface" => {
        "config" => { "Interface" => "lo", "IgnoreSelected" => true }
      },
      "df"        => {
        "config" => {
          "ReportReserved" => false,
          "FSType" => [ "proc", "sysfs", "fusectl", "debugfs", "devtmpfs", "devpts", "tmpfs" ],
          "IgnoreSelected" => true
        }
      },
      "write_graphite" => {
        "config" => {
          "Prefix" => "servers."
        }
      }
    }
  }
)