Go to file
Graham Floyd c0fa6af51b Add disque plugin 2015-07-31 14:46:46 -05:00
cmd/telegraf Add -pidfile and Commit variable 2015-06-18 11:31:16 -07:00
etc add MongoDB plugin 2015-07-07 11:25:34 -05:00
plugins Add disque plugin 2015-07-31 14:46:46 -05:00
scripts systemd unit support 2015-07-14 21:42:33 +03:00
testdata Tivan is dead, long live Telegraf. Fixes #3 2015-05-22 16:45:14 -07:00
testutil Merge pull request #49 from marcosnils/container_services 2015-07-30 16:29:44 -07:00
.gitignore Add release.sh and Vagrantfile 2015-04-07 09:31:28 -07:00
CHANGELOG.md Update CHANGELOG.md 2015-07-09 12:19:49 -06:00
LICENSE Initial commit 2015-04-01 09:28:44 -07:00
Makefile Add --no-recreate option to prepare target 2015-07-06 21:17:44 -03:00
README.md Merge pull request #49 from marcosnils/container_services 2015-07-30 16:29:44 -07:00
Vagrantfile Add release.sh and Vagrantfile 2015-04-07 09:31:28 -07:00
accumulator.go protect accumulator values with a mutex 2015-06-21 21:16:46 -04:00
agent.go Use string for InfluxDB timeout duration config. 2015-07-12 18:05:44 +02:00
agent_test.go Tivan is dead, long live Telegraf. Fixes #3 2015-05-22 16:45:14 -07:00
config.go Use string for InfluxDB timeout duration config. 2015-07-12 18:05:44 +02:00
docker-compose.yml Use postgres default configuration 2015-07-06 03:46:53 -03:00
package.sh Telegraf should have its own directories. 2015-07-05 18:38:22 -07:00
release.sh Add linux/arm to list of built binaries 2015-06-19 21:53:58 -07:00

README.md

Telegraf - A native agent for InfluxDB

Telegraf is an agent written in Go for collecting metrics from the system it's running on or from other services and writing them into InfluxDB.

Design goals are to have a minimal memory footprint with a plugin system so that developers in the community can easily add support for collecting metrics from well known services (like Hadoop, or Postgres, or Redis) and third party APIs (like Mailchimp, AWS CloudWatch, or Google Analytics).

We'll eagerly accept pull requests for new plugins and will manage the set of plugins that Telegraf supports. See the bottom of this doc for instructions on writing new plugins.

Quickstart

  • Build from source or download telegraf:

Linux packages for Debian/Ubuntu and RHEL/CentOS:

http://get.influxdb.org/telegraf/telegraf_0.1.4_amd64.deb
http://get.influxdb.org/telegraf/telegraf-0.1.4-1.x86_64.rpm

OSX via Homebrew:

brew update
brew install telegraf

How to use it:

  • Run telegraf -sample-config > telegraf.toml to create an initial configuration
  • Edit the configuration to match your needs
  • Run telegraf -config telegraf.toml -test to output one full measurement sample to STDOUT
  • Run telegraf -config telegraf.toml to gather and send metrics to InfluxDB

Telegraf Options

Telegraf has a few options you can configure under the agent section of the config. If you don't see an agent section run telegraf -sample-config > telegraf.toml to create a valid initial configuration:

  • hostname: The hostname is passed as a tag. By default this will be the value retured by hostname on the machine running Telegraf. You can override that value here.
  • interval: How ofter to gather metrics. Uses a simple number + unit parser, ie "10s" for 10 seconds or "5m" for 5 minutes.
  • debug: Set to true to gather and send metrics to STDOUT as well as InfluxDB.

Supported Plugins

Telegraf currently has support for collecting metrics from:

  • System (memory, CPU, network, etc.)
  • Docker
  • MySQL
  • Prometheus (client libraries and exporters)
  • PostgreSQL
  • Redis
  • Elasticsearch
  • RethinkDB
  • Kafka
  • MongoDB

We'll be adding support for many more over the coming months. Read on if you want to add support for another service or third-party API.

Plugin Options

There are 3 configuration options that are configurable per plugin:

  • pass: An array of strings that is used to filter metrics generated by the current plugin. Each string in the array is tested as a prefix against metrics and if it matches, the metric is emitted.
  • drop: The inverse of pass, if a metric matches, it is not emitted.
  • interval: How often to gather this metric. Normal plugins use a single global interval, but if one particular plugin should be run less or more often, you can configure that here.

Plugins

This section is for developers that want to create new collection plugins. Telegraf is entirely plugin driven. This interface allows for operators to pick and chose what is gathered as well as makes it easy for developers to create new ways of generating metrics.

Plugin authorship is kept as simple as possible to promote people to develop and submit new plugins.

Guidelines

  • A plugin must conform to the plugins.Plugin interface.
  • Telegraf promises to run each plugin's Gather function serially. This means developers don't have to worry about thread safety within these functions.
  • Each generated metric automatically has the name of the plugin that generated it prepended. This is to keep plugins honest.
  • Plugins should call plugins.Add in their init function to register themselves. See below for a quick example.
  • To be available within Telegraf itself, plugins must add themselves to the github.com/influxdb/telegraf/plugins/all/all.go file.
  • The SampleConfig function should return valid toml that describes how the plugin can be configured. This is include in telegraf -sample-config.
  • The Description function should say in one line what this plugin does.

Plugin interface

type Plugin interface {
  SampleConfig() string
  Description() string
  Gather(Accumulator) error
}

type Accumulator interface {
  Add(measurement string, value interface{}, tags map[string]string)
  AddValuesWithTime(measurement string, values map[string]interface{}, tags map[string]string, timestamp time.Time)
}

Accumulator

The way that a plugin emits metrics is by interacting with the Accumulator.

The Add function takes 3 arguments:

  • measurement: A string description of the metric. For instance bytes_read or faults.
  • value: A value for the metric. This accepts 5 different types of value:
    • int: The most common type. All int types are accepted but favor using int64 Useful for counters, etc.
    • float: Favor float64, useful for gauges, percentages, etc.
    • bool: true or false, useful to indicate the presence of a state. light_on, etc.
    • string: Typically used to indicate a message, or some kind of freeform information.
    • time.Time: Useful for indicating when a state last occurred, for instance light_on_since.
  • tags: This is a map of strings to strings to describe the where or who about the metric. For instance, the net plugin adds a tag named "interface" set to the name of the network interface, like "eth0".

The AddValuesWithTime allows multiple values for a point to be passed. The values used are the same type profile as value above. The timestamp argument allows a point to be registered as having occurred at an arbitrary time.

Let's say you've written a plugin that emits metrics about processes on the current host.


type Process struct {
  CPUTime float64
  MemoryBytes int64
  PID int
}

func Gather(acc plugins.Accumulator) error {
  for _, process := range system.Processes() {
    tags := map[string]string {
      "pid": fmt.Sprintf("%d", process.Pid),
    }

    acc.Add("cpu", process.CPUTime, tags)
    acc.Add("memory", process.MemoryBytes, tags)
  }
}

Example

package simple

// simple.go

import "github.com/influxdb/telegraf/plugins"

type Simple struct {
  Ok bool
}

func (s *Simple) Description() string {
  return "a demo plugin"
}

func (s *Simple) SampleConfig() string {
  return "ok = true # indicate if everything is fine"
}

func (s *Simple) Gather(acc plugins.Accumulator) error {
  if s.Ok {
    acc.Add("state", "pretty good", nil)
  } else {
    acc.Add("state", "not great", nil)
  }

  return nil
}

func init() {
  plugins.Add("simple", func() plugins.Plugin { return &Simple{} })
}

Testing

As Telegraf collects metrics from several third-party services it becomes a difficult task to mock each service as some of them have complicated protocols which would take some time to replicate.

To overcome this situation we've decided to use docker containers to provide a fast and reproducible environment to test those services which require it. For other situations (i.e: https://github.com/influxdb/telegraf/blob/master/plugins/redis/redis_test.go ) a simple mock will suffice.

To execute Telegraf tests follow this simple steps:

  • Install docker compose following these instructions
  • execute make test