Description

Installs and configures Graphite http://graphite.wikidot.com/

Requirements

• Ubuntu 10.04 / Ubuntu 12.04
• Debian
• RHEL and derivatives (Centos, Amazon Linux, Oracle Linux, Scientific Linux)
• Fedora

Attributes

• node['graphite']['version'] - version of graphite to install (defaults to 0.9.10)
• node['graphite']['password'] - password for graphite root user (default to change_me and is only used if encrypted databag isn't)
• node['graphite']['chef_role'] - chef role name for graphite instances, used by the federated recipe (defaults to "graphite")
• node['graphite']['url'] - url of the graphite server (defaults to graphite)
• node['graphite']['url_aliases'] - array of url aliases (defaults to nil)
• node['graphite']['listen_port'] - port to listen on (defaults to 80)
• node['graphite']['base_dir'] = "/opt/graphite"
• node['graphite']['doc_root'] = "/opt/graphite/webapp"
• node['graphite']['storage_dir'] = "/opt/graphite/storage"
• node['graphite']['django_root'] = "@DJANGO_ROOT@" - configurable path to your django installation
• node['graphite']['timezone'] - set the timezone for the graphite web interface, defaults to America/Los_Angeles
• node['graphite']['whisper']['uri'] - download url for whisper
• node['graphite']['whisper']['checksum'] - checksum of the whisper download
• node['graphite']['encrypted_data_bag']['name'] - the name of the encrypted data bag containing the default password for the graphite "root" user. If this attribute is set it will not use node['graphite']['password'].

carbon-cache.py attributes

• node['graphite']['storage_schemas'] - an array with retention rates for storing metrics, used to generate the storage-schemas.conf file (see the example below)
• node['graphite']['carbon']['uri'] - download url for carbon
• node['graphite']['carbon']['checksum'] - checksum for the carbon download
• node['graphite']['carbon']['line_receiver_interface'] - line interface IP (defaults to 0.0.0.0)
• node['graphite']['carbon']['line_receiver_port'] - line interface port (defaults to 2003)
• node['graphite']['carbon']['enable_udp_listener'] - set this to "True" to enable the UDP listener (defaults to "False")
• node['graphite']['carbon']['udp_receiver_interface'] - line interface IP for UDP listener (defaults to 0.0.0.0)
• node['graphite']['carbon']['udp_receiver_port'] - line interface port for UDP listener (defaults to 2003)
• node['graphite']['carbon']['pickle_receiver_interface'] - pickle receiver IP (defaults to 0.0.0.0)
• node['graphite']['carbon']['pickle_receiver_port'] - pickle receiver port (defaults to 2004)
• node['graphite']['carbon']['use_insecure_unpickler'] - set this to "True" to use the old-fashioned insecure unpickler (defaults to "False")
• node['graphite']['carbon']['cache_query_interface'] - cache query IP (defaults to 0.0.0.0)
• node['graphite']['carbon']['cache_query_port'] - cache query port (defaults to 7002)
• node['graphite']['carbon']['use_flow_control'] - set this to "False" to drop datapoints received after the cache reaches MAX_CACHE_SIZE (defaults to "True")
• node['graphite']['carbon']['max_cache_size'] - max size of the carbon cache (defaults to "inf")
• node['graphite']['carbon']['max_creates_per_second'] - max number of new metrics to create per second (defaults to "inf")
• node['graphite']['carbon']['max_updates_per_second'] - max updates to carbon per second (defaults to "1000")
• node['graphite']['carbon']['log_whisper_updates'] - log updates to whisper (defaults to "False")
• node['graphite']['carbon']['whisper_autoflush'] - set this option to "True" if you want whisper to write synchronously (defaults to "False")
• node['graphite']['carbon']['service_type'] - init service to use for carbon (defaults to runit)

carbon-relay.py attributes

• node['graphite']['relay_rules'] - an array with relay rules for sending metrics to a certain backends, used to generate the relay-rules.conf file (see the example below)
• node['graphite']['carbon']['relay']['line_receiver_interface'] - line interface IP (defaults to 0.0.0.0)
• node['graphite']['carbon']['relay']['line_receiver_port'] - line interface port (defaults to 2013)
• node['graphite']['carbon']['relay']['pickle_receiver_interface'] - pickle receiver IP (defaults to 0.0.0.0)
• node['graphite']['carbon']['relay']['pickle_receiver_port'] - pickle receiver port (defaults to 2014)
• node['graphite']['carbon']['relay']['relay_method'] - choose between consistent-hashing and rules (defaults to "rules")
• node['graphite']['carbon']['relay']['replication_factor'] - used to replicate datapoint data to more than one machine (defaults to 1)
• node['graphite']['carbon']['relay']['destinations'] - list of carbon daemons to send metrics to
• node['graphite']['carbon']['relay']['max_datapoints_per_message'] - maximum datapoints to send in a message between carbon daemons (defaults to 500)
• node['graphite']['carbon']['relay']['max_queue_size'] - maximum queue of messages used to comunicate to other carbon daemons (defaults to 10000)
• node['graphite']['carbon']['relay']['use_flow_control'] - set this to "False" to drop datapoints received after the cache reaches MAX_CACHE_SIZE (defaults to "True")

carbon-aggregator.py attributes

• node['graphite']['storage_aggregation'] - an array with rules to configure how to aggregate data to lower-precision retentions, used to generate the storage-aggregation.conf file
• node['graphite']['aggregation_rules'] - an array with rules that allow you to add several metrics together, used to generate the aggregation-rules.conf file (see the example below)
• node['graphite']['carbon']['aggregator']['line_receiver_interface'] - line interface IP (defaults to 0.0.0.0)
• node['graphite']['carbon']['aggregator']['line_receiver_port'] - line interface port (defaults to 2023)
• node['graphite']['carbon']['aggregator']['pickle_receiver_interface'] - pickle receiver IP (defaults to 0.0.0.0)
• node['graphite']['carbon']['aggregator']['pickle_receiver_port'] - pickle receiver port (defaults to 2024)
• node['graphite']['carbon']['aggregator']['destinations'] - list of carbon daemons to send metrics to
• node['graphite']['carbon']['aggregator']['replication_factor'] - used to add redundancy to your data by replicating every datapoing to more than one machinne (defaults to 1)
• node['graphite']['carbon']['aggregator']['max_queue_size'] - maximum queue of messages used to comunicate to other carbon daemons (defaults to 10000)
• node['graphite']['carbon']['aggregator']['use_flow_control'] - set this to "False" to drop datapoints received after the cache reaches MAX_CACHE_SIZE (defaults to "True")
• node['graphite']['carbon']['aggregator']['max_datapoints_per_message'] - maximum datapoints to send in a message between carbon daemons (defaults to 500)
• node['graphite']['carbon']['aggregator']['max_aggregation_intervals'] - sets how many datapoints the aggregator remembers for each metric (defaults to 5)

graphite-web attributes

• node['graphite']['web']['uri'] - download url for the graphite web ui
• node['graphite']['web']['checksum'] - checksum for the graphite web ui download
• node['graphite']['web']['debug'] - debug mode (defaults to "False")
• node['graphite']['web']['admin_email'] - admin contact email (defaults to "admin@org.com")
• node['graphite']['web']['cluster_servers'] - IP address (and optionally port) of the webapp on each remote server in the cluster
• node['graphite']['web']['carbonlink_hosts'] - list the IP address, cache query port and instance name of each carbon cache instance on the local machine
• node['graphite']['web_server'] - defaults to apache. Anything else will use uswsgi instead of apache
• node['graphite']['user_account'] - user (default node['apache']['user'])
• node['graphite']['group_account'] - group (default node['apache']['group'])
• node['graphite']['create_user']- should the user be created, boolean (defaults to false)
• node['graphite']['ssl']['enabled'] - enable ssl in the apache2 vhost
• node['graphite']['ssl']['cipher_suite'] - the cipher suite to use if ssl is enabled
• node['graphite']['ssl']['certificate_file'] - the path to the certificate file if ssl is enabled
• node['graphite']['ssl']['certificate_key_file'] - the path to the vertificate key file if ssl is enabled

storage_schemas example

node.default['graphite']['storage_schemas'] = [
  {
    'name' => 'carbon',
    'pattern' => /^carbon\./,
    'retentions' => '1m:10d'
  },
  {
    'name' => 'sensu',
    'pattern' => /^sensu\./,
    'retentions' => '1m:30d'
  },
  {
    'name' => 'everything_30s7d_15m1m',
    'match-all' => true,
    'retentions' => '30s:7d,15m:1m'
  }
]

relay_rules example

node.default['graphite']['relay_rules'] = [
  {
    'name' => 'example_pattern',
    'pattern' => /^mydata\.foo\..+/,
    'destinations' => [ '10.1.2.3', '10.1.2.4:2004', 'myserver.mydomain.com' ]
  },{
    'name' => 'example_default',
    'default' => true,
    'destinations' => [ '10.1.2.5:2004' ]
  }
]

aggregation_rules example

node.default['graphite']['aggregation_rules'] = [
  {
    'output_template' => '<env>.applications.<app>.all.requests',
    'frequency' => '60',
    'method' => 'sum',
    'input_pattern' => '<env>.applications.<app>.*.requests'
  },
  {
    'output_template' => '<env>.applications.<app>.all.latency',
    'frequency' => '60',
    'method' => 'sum',
    'input_pattern' => '<env>.applications.<app>.*.latency'
  },
]

Data Bags

This cookbook optionally uses an encrypted data bag to store the graphite password.
If this data bag is not present the cookbook will use node['graphite']['password']
instead. To use the encrypted data bag set node['graphite']['encrypted_data_bag']['name']
with the name of the data bag you wish to use.

Helper Scripts

The following helper scripts are included in the graphite/bin directory:

• whisper-clean-this-node.sh - this script cleans the whisper metrics that belong to other machines in the cluster. Usually used after synchronizing the storage/whisper directory. Uses the whisper-clean.py script internally

Usage

recipe[graphite] should build a stand-alone Graphite installation.

recipe[graphite::ganglia] integrates with Ganglia. You'll want at
least one monitor node (i.e. recipe[ganglia]) node to be running
to use it.