A statsd client implementation for Elixir.
Add ExStatsD as a dependency for your application.
defp deps do
[{:ex_statsd, ">= 0.5.1"}]
end
You should also update your applications list to include statsd:
def application do
[applications: [:ex_statsd]]
end
Configure ex_statsd
using Mix.Config
as usual (probably in your
config/
):
use Mix.Config
config :ex_statsd,
host: "your.statsd.host.com",
port: 1234,
namespace: "your-app"
The defaults are:
- host: 127.0.0.1
- port: 8125
- namespace: nil
The following are the basic metric types. Additional features are described in "Extensions," below.
Counters can be manipulated with ExStatsD.increment/1
and
ExStatsD.decrement/1
for simple counts:
if passed? do
ExStatsD.increment("cases.passed")
end
if cancelled_account? do
ExStatsD.decrement("users")
end
You can also provide a sample_rate
with ExStatsD.increment/2
and
ExStatsD.decrement/2
. For example, in this case an increment for
cart.added
will only be sent 50% of the time:
ExStatsD.increment("cart.added", sample_rate: 0.5)
To set a counter explicitly, use ExStatsD.counter/2
:
3 |> ExStatsD.counter("cart.removed")
You can also send a sample_rate
:
3 |> ExStatsD.counter("cart.removed", sample_rate: 0.25)
Note that the function returns the value (eg, 3
here), making it
suitable for pipelining.
Manually timed values can be recorded with ExStatsD.timer/2
:
elapsed_ms = # something manually timed in
elapsed_ms |> ExStatsD.timer("foobar")
The value passed to ExStatsD.timer/2
(eg, elapsed_ms
here) is
returned from it, making this suitable for pipelining.
For convenience, you can also time a function call with
ExStatsD.timing/2
:
ExStatsD.timing "foo.bar", fn ->
# Time something
end
To sample (ie, 50% of the time), pass a sample_rate
:
ExStatsD.timing "foo.bar", fn ->
# Time something, some of the time
end, sample_rate: 0.5
Note that, regardless of the sample rate, the function is always called -- it's just not always measured. Also note that the return value of the measured function is returned, making this suitable for pipelining.
A value can be recorded in a set with ExStatsD.set/2
:
user_id |> ExStatsD.set("users")
Note that the function returns the value, making it suitable for pipelining.
All metrics support the
Datadog-specific tags extension
to StatsD. If you are using DogStatsD, you may provide a tags
option, eg:
ExStatsD.increment("cart.added", tags: ~w(foo bar))
The histogram extension to StatsD is supported for DogStatsD:
42 |> ExStatsD.histogram("database.query.time", tags: ["db", "perf"])
Note that the function returns the value, making it suitable for pipelining.
A histogram version of the ExStatsD.timing function is supported for DogStatsD.
ExStatsD.histogram_timing "foo.bar", fn ->
# Time something
end
The decorators allow you to quickly and easily time function calls in
your code. Simply replace def
with deftimed
for those functions
you wish to time.
defmodule MyModule.Data do
use ExStatsD.Decorator
deftimed slow_function do
# This is a suspect function we wish to time.
end
end
Now all calls to MyModule.Data.slow_function/0
will be timed and
reported to your statsd server. By default the metric key used for
each call will be PREFIX.function_call.MODULE.FUNCTION_ARITY
. So in
this example it would have been
function_call.mymodule.data.slow_function_0
.
You can change the metric name by setting the @metric
attribute
ahead of the function. The metric will apply to other function
definitions of the same arity unless specifically changed again. Other
following functions of different name or arity will use the default.
deftimed init, do: nil # PREFIX.function_call.mymodule.data.init_0
@metric "trace.some_function"
deftimed some_function(1), do: nil # PREFIX.trace.some_function
deftimed some_function(2), do: nil # PREFIX.trace.some_function
@metric "trace.some_function_catchall"
deftimed some_function(x) when is_list(x), do: nil # PREFIX.trace.some_function_catchall
deftimed some_function(x), do: nil # PREFIX.trace.some_function_catchall
deftimed some_function(x,y), do: nil # PREFIX.function_call.mymodule.data.some_function_2
You can set options using the @metric_options
attribute. This follows the same rules as with the @metric
example abobe.
Here we use Datadog's "tag" extension to StatD:
@metric_options [tags: ["basic"]]
deftimed some_function(), do: nil
There are 2 global options available. Both will apply to all functions that follow it unless locally overridden.
@default_metric_options
: Metric options to use unless overridden with@metric_options
. Defaults to [].@use_histogram
: Send results using histograms instead of gauges. For use with Datadog's DogStatD. Defaults to false.
The MIT License (MIT)
Copyright (c) 2014 CargoSense, Inc.
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.