Metrics collection and reporting for Elixir applications.
Alchemetrics makes life easier for anyone who wants to report and distribute metrics from an Elixir application. The metrics can help you to measure performance, detect errors or track information about your application.
Alchemetrics makes use of GenStage to ensure that collecting and submitting metrics will not impact application performance.
Each metric report creates a new event that follows the GenStage flow. With the help of the ConsumerSupervisor, your metrics are distributed with little impact to the application performance.
Documentation is available at HexDocs
Alchemetrics is available on Hex. All you have to do is to add it to mix.exs as a test dependency.
def deps do
[{:alchemetrics, "~> 0.5.2"}]
endReports are made via calls to Alchemetrics functions. It is through reports that Alchemetrics stores a value to be measured and sent to the backends. Further details about reports can be found in the documentation available in HexDocs.
Collected metrics are typically stored in some type of datastore, such as Logstash and Influxdb. Alchemetrics uses the concept of backends to distribute the metrics to these data stores. More details can be found in the documentation about custom backends.
When a dataset is created, it subscribes to all backends enabled on the application. Datasets created before a backend is enabled will not subscribe to the new backend. Also, when a backend is disabled, all datasets will unsubscribe from it.
Alchemetrics comes with a built-in backend called Alchemetrics.ConsoleBackend. This backend sends yor metrics to the console and is very useful when debugging.
You can enable Alchemetrics.ConsoleBackend when your application boot by adding it to yor backend list:
# config/config.exs
config :alchemetrics,
backends: [
{Alchemetrics.ConsoleBackend, []}
]You can also enable it on the console:
# iex -S mix
iex(1)> Alchemetrics.ConsoleBackend.enable
Starting Elixir.Alchemetrics.ConsoleBackend with following options: []
:ok
iex(2)> Alchemetrics.report(100, :test)
:ok
iex(3)> %{datapoint: :avg, name: :test, options: [], value: 100}
%{datapoint: :max, name: :test, options: [], value: 100}
%{datapoint: :min, name: :test, options: [], value: 100}
%{datapoint: :p95, name: :test, options: [], value: 100}
%{datapoint: :p99, name: :test, options: [], value: 100}
%{datapoint: :last_interval, name: :test, options: [], value: 100}
%{datapoint: :total, name: :test, options: [], value: 100}Alchemetrics automatically collects some information about Erlang's VM, like memory and run queue. This is disabled by default, but you can enable it on the configs:
# config/config.exs
config :alchemetrics, instrument_beam: trueThis will enable the report of some data from Erlang's VM:
%{datapoint: :memory_atom, options: [], type: :memory, value: 621465}
%{datapoint: :memory_binary, options: [], type: :memory, value: 392504}
%{datapoint: :memory_ets, options: [], type: :memory, value: 1365728}
%{datapoint: :memory_processes, options: [], type: :memory, value: 10513080}
%{datapoint: :memory_total, options: [], type: :memory, value: 38397232}
%{datapoint: :system_runqueue, options: [], type: :system, value: 1}Let's show an example of how Alchemetrics could be used to instrument a Phoenix application with the help of Plug:
In this example application we will measure:
- The number of requests for each route of the application grouped by the status of the response;
- The response times (
average,percentiles,maxandmin) of each route;
The collected metrics will be printed to the console by Alchemetrics.ConsoleBackend. To do this, we need to enable it in the settings:
# config/config.exs
config :alchemetrics,
backends: [
[module: Alchemetrics.ConsoleBackend, []]
]Let's create the RequestInstrumentor plug:
# lib/my_app_web/plugs/request_count.ex
defmodule RequestInstrumentor do
@behaviour Plug
def init(opts \\ []), do: opts
def call(conn, opts), do: count_request(conn)
defp count_request(conn) do
start = System.monotonic_time()
Plug.Conn.register_before_send(conn, fn conn ->
stop = System.monotonic_time()
diff = System.convert_time_unit(stop - start, :native, :micro_seconds)
# report throughput
Alchemetrics.increment(request_count: %{method: conn.method, path: conn.request_path, status: conn.status})
# report request time
Alchemetrics.report(diff, request_time: %{method: conn.method, path: conn.request_path})
conn
end)
end
endPlug it on Phoenix Endpoint:
defmodule MyAppWeb.Endpoint do
use Phoenix.Endpoint, otp_app: :my_app
plug RequestInstrumentor
...
endRequest your application:
$ mix phx.server
$ curl localhost:4000/Information about the request will show on application console:
%{datapoint: :last_interval, options: [], requests_for: %{method: "GET", path: "/", status: 200}, value: 1}
%{datapoint: :total, options: [], requests_for: %{method: "GET", path: "/", status: 200}, value: 1}
%{datapoint: :avg, options: [], request_time: %{method: "GET", path: "/", status: 200}, value: 44069}
%{datapoint: :max, options: [], request_time: %{method: "GET", path: "/", status: 200}, value: 44069}
%{datapoint: :min, options: [], request_time: %{method: "GET", path: "/", status: 200}, value: 44069}
%{datapoint: :p95, options: [], request_time: %{method: "GET", path: "/", status: 200}, value: 44069}
%{datapoint: :p99, options: [], request_time: %{method: "GET", path: "/", status: 200}, value: 44069}
%{datapoint: :last_interval, options: [], request_time: %{method: "GET", path: "/", status: 200}, value: 44069}
%{datapoint: :total, options: [], request_time: %{method: "GET", path: "/", status: 200}, value: 44069}If you request an inexistent route, the reports will show 404 status code:
%{datapoint: :last_interval, options: [], requests_for: %{method: "GET", path: "/invalid_route", status: 404}, value: 1}
%{datapoint: :total, options: [], requests_for: %{method: "GET", path: "/invalid_route", status: 404}, value: 1}
%{datapoint: :avg, options: [], request_time: %{method: "GET", path: "/invalid_route", status: 404}, value: 39558}
%{datapoint: :max, options: [], request_time: %{method: "GET", path: "/invalid_route", status: 404}, value: 39558}
%{datapoint: :min, options: [], request_time: %{method: "GET", path: "/invalid_route", status: 404}, value: 39558}
%{datapoint: :p95, options: [], request_time: %{method: "GET", path: "/invalid_route", status: 404}, value: 39558}
%{datapoint: :p99, options: [], request_time: %{method: "GET", path: "/invalid_route", status: 404}, value: 39558}
%{datapoint: :last_interval, options: [], request_time: %{method: "GET", path: "/invalid_route", status: 404}, value: 39558}
%{datapoint: :total, options: [], request_time: %{method: "GET", path: "/invalid_route", status: 404}, value: 39558}For more details about reports, metrics, datasets, backends and all Alchemetrics concepts, take a look at the docs.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent