Data Daemon

An Elixir StatsD client made for DataDog.

Quick Setup

# In your config/config.exs file
config :my_app, Sample.DataDog,
  url: "statsd+udp://localhost:8125"

# In your application code
defmodule Sample.DataDog do
  @moduledoc ~S"My DataDog reporter."
  use DataDaemon,
    otp_app: :my_app,
    extensions: [:datadog, :erlang_vm]
end

defmodule Sample.App do
  alias Sample.DataDog

  def send_metrics do
    Sample.DataDog.start_link()

    tags = [zone: "us-east-1a"]

    DataDog.gauge("request.queue_depth", 12, tags: tags)

    DataDog.distribution("connections", 123, tags: tags)
    DataDog.histogram("request.file_size", 1034, tags: tags)

    DataDog.timing("request.duration", 34, tags: tags)

    DataDog.increment("request.count_total", tags: tags)
    DataDog.decrement("request.count_total", tags: tags)
    DataDog.count("request.count_total", 2, tags: tags)
  end
end

Installation

The package can be installed by adding data_daemon to your list of dependencies in mix.exs:

def deps do
  [
    {:data_daemon, "~> 0.4"}
  ]
end

The docs can be found at https://hexdocs.pm/data_daemon.

Extensions

DataDog

A DataDog extension is available offering the following functionality:

distribution/3 metric.
event/3 event sending over UDP.
error_handler is available as option and can be used to send errors as events.

Events

Create an event for the DataDog event stream by passing a title and message to event/3.

The following options are also supported:

Option	Description
`:timestamp` (optional)	Add a timestamp to the event. Default is the current timestamp.
`:hostname` (optional)	Add a hostname to the event. No default.
`:aggregation_key` (optional)	Add an aggregation key to group the event with others that have the same key. No default.
`:priority` (optional)	Set to `:normal` or `:low`. Default `:normal`.
`:source_type_name` (optional)	Add a source type to the event. No default.
`:alert_type` (optional)	Set to `:error`, `:warning`, `:info` or `:success`. Default `:info`.

Example

defmodule Sample.DataDog do
  @moduledoc ~S"My DataDog reporter."
  use DataDaemon,
    otp_app: :my_app,
    extensions: [:datadog],
    error_handler: true
end

defmodule Sample.App do
  alias Sample.DataDog

  def send_events do
    tags = [zone: "us-east-1a"]

    DataDog.event("Event Title", "Event body.\nMore details", tags: tags)
  end
end

All event options are support, for more details see: []

Erlang VM

An Erlang VM extension is available logging Erlang VM stats/metrics every minute.

The reporting interval can be configured with the :rate (in millisecond) inside the :erlang_vm config.

Example: config :my_app, MyDaemon, erlang_vm: [rate: 1_000] for updates every second.

The following metrics are tracked:

vm.process.count, the current process count.
vm.process.limit, the current process limit.
vm.process.queue, the current amount of processes queued for running.
vm.port.count, the current port count.
vm.port.limit, the current port limit.
vm.atom.count, the current atom count.
vm.atom.limit, the current atom limit.
vm.error.queue, the amount of process messages queued for the error logger.
vm.uptime, erlang uptime.
vm.refresh, the amount of time (ms) since last stat check.
vm.reductions, amount of reductions.
vm.message.queue, total queued messages over all processes.
vm.modules, current amount of loaded modules.
vm.memory.total, total memory use in Kb.
vm.memory.processes, total process memory chunk use in Kb.
vm.memory.processes_used, total process memory use in Kb.
vm.memory.system, total system memory use in Kb.
vm.memory.atom, total atom memory chunk use in Kb.
vm.memory.atom_used, total atom memory use in Kb.
vm.memory.binary, total binary memory use in Kb.
vm.memory.code, total code memory use in Kb.
vm.memory.ets, total ets memory use in Kb.
vm.io.in, total IO input in Kb.
vm.io.out, total IO output in Kb.
vm.garbage_collection.count, number of garbage collections.
vm.garbage_collection.words, number of words garbage.

Changelog

0.4.5 (2024-04-25)

Fix recompilation when IP does not change.

0.4.3 (2023-11-17)

Added test helper clear function. (by @punchcafe)

0.4.3 (2023-10-23)

Respect runtime configuration.
Error with {:error, :not_started} when Daemon is not started.

0.4.1 (2022-08-09)

Fix resolving url with IP as host. Reduce dependency on Plug. Namespace logging.

0.4.0 (2020-08-11)

Improved metrics throughput by sharing named sockets/ports.

0.3.4 (2020-03-15)

Elixir 1.10 support. (Purely spec updates.)

0.3.3 (2020-02-06)

Bug fixes:

UDP socket leak on DNS refresh.

0.3.2 (2020-02-01)

New:

Full documentation and specs for using module.
Doctest support for target module.

Bug fixes:

DataDog distribution custom metric identifier.

0.3.1 (2020-01-31)

Changes:

Custom metric type passing. (binary)
Extended module based config options.

Bug fixes:

DataDog specs. (Dialyzer)

0.3.0 (2020-01-30)

Changes:

New socket logic for Erlang OTP 22 and up.

Bug fixes:

Hound spec. (Dialyzer)
DataDaemon spec. (Dialyzer)

0.2.4 (2020-01-29)

Bug fixes:

Make hound resolve and open a new socket on some errors.

0.2.3 (2019-04-30)

Bug fixes:

Make hound update async to prevent pool-resolver deadlock.

0.2.2 (2019-04-29)

New features:

vm.refresh erlang vm metric that tracks time (ms) since last stat check.

Bug fixes:

Fixes issue where the vm.uptime metric wouldn't actually track uptime, but time since last vm check.

0.2.1 (2019-04-27)

New features:

:minimum_ttl option to set a minimum TTL to prevent excessive refreshing. (Default: 1_000)

Optimizations:

:erlang_vm extension runs as additional child to restart during crashes.
Perform actual resolve in async process to prevent deadlock during excessive DNS refresh.
DNS resolves in separate process removing use of :timer and only sending an update to workers if IP changes.

Bug fixes:

Fixes issue where the :erlang_vm would link its process not to the DataDaemon, but the calling process.

0.2.0 (2019-04-17)

New features:

:erlang_vm extension. Logs Erlang VM stats/metrics every minute. (Can be configured)

Optimizations:

Make config more dynamic.
DNS resolves in separate process removing use of :timer and only sending an update to workers if IP changes.

0.1.4 (2019-03-25)

New features:

Allow config overwrite by passing keyword settings to the child spec or start link.

Optimizations:

Keep using old IP in case DNS refresh fails.
Added extra logging around DNS failures.

0.1.3 (2018-11-03)

New features:

The :hound setting now allows you to set the pool settings.

Optimizations:

DNS lookup only updates header on change.

Fixes:

Functions with default arguments are now properly overwritten.

0.1.2 (2018-10-09)

New features:

Test mode now supported as configuration.
The :error_handler setting now allows you to set a minimum level. (Default: :info, possible: :debug, :info, :warn, and :error)
Add :dsn_refresh config for refreshing the host name. (Default: :ttl, possible: :ttl and <integer>. (seconds))

Optimizations:

DNS lookup only updates header on change.

Fixes:

Functions with default arguments are now properly overwritten.

0.1.0 (2018-10-07)

New features:

Extension system to allow for different StatsD extensions.
Add new tag and value formats:
- iodata now supported for tags and values.
- {:config, app, key} now supported for tags.
DataDog Events are now supported with event/3.
DataDog can now be used as error handler by setting error_handler: true in module.

Optimizations:

Plug reported optimized for detached user response and compile time optimizations.