git.squeep.com Git - akkoma/blob - lib/pleroma/web/telemetry.ex

   1 defmodule Pleroma.Web.Telemetry do
   2   use Supervisor
   3   import Telemetry.Metrics
   4   alias Pleroma.Stats
   5
   6   def start_link(arg) do
   7     Supervisor.start_link(__MODULE__, arg, name: __MODULE__)
   8   end
   9
  10   @impl true
  11   def init(_arg) do
  12     children = [
  13       {:telemetry_poller, measurements: periodic_measurements(), period: 10_000},
  14       {TelemetryMetricsPrometheus, metrics: prometheus_metrics(), plug_cowboy_opts: [ip: {127, 0, 0, 1}]}
  15     ]
  16
  17     Supervisor.init(children, strategy: :one_for_one)
  18   end
  19
  20   @doc """
  21   A seperate set of metrics for distributions because phoenix dashboard does NOT handle
  22   them well
  23   """
  24   defp distribution_metrics do
  25     [
  26       distribution(
  27         "phoenix.router_dispatch.stop.duration",
  28         # event_name: [:pleroma, :repo, :query, :total_time],
  29         measurement: :duration,
  30         unit: {:native, :second},
  31         tags: [:route],
  32         reporter_options: [
  33           buckets: [0.1, 0.2, 0.5, 1, 2.5, 5, 10, 25, 50, 100, 250, 500, 1000]
  34         ]
  35       ),
  36
  37       # Database Time Metrics
  38       distribution(
  39         "pleroma.repo.query.total_time",
  40         # event_name: [:pleroma, :repo, :query, :total_time],
  41         measurement: :total_time,
  42         unit: {:native, :millisecond},
  43         reporter_options: [
  44           buckets: [0.1, 0.2, 0.5, 1, 2.5, 5, 10, 25, 50, 100, 250, 500, 1000]
  45         ]
  46       ),
  47       distribution(
  48         "pleroma.repo.query.queue_time",
  49         # event_name: [:pleroma, :repo, :query, :total_time],
  50         measurement: :queue_time,
  51         unit: {:native, :millisecond},
  52         reporter_options: [
  53           buckets: [0.01, 0.025, 0.05, 0.1, 0.2, 0.5, 1, 2.5, 5, 10]
  54         ]
  55       ),
  56       distribution(
  57         "oban_job_exception",
  58         event_name: [:oban, :job, :exception],
  59         measurement: :duration,
  60         tags: [:worker],
  61         tag_values: fn tags -> Map.put(tags, :worker, tags.job.worker) end,
  62         unit: {:native, :second},
  63         reporter_options: [
  64           buckets: [0.01, 0.025, 0.05, 0.1, 0.2, 0.5, 1, 2.5, 5, 10]
  65         ]
  66       ),
  67       distribution(
  68         "tesla_request_completed",
  69         event_name: [:tesla, :request, :stop],
  70         measurement: :duration,
  71         tags: [:response_code],
  72         tag_values: fn tags -> Map.put(tags, :response_code, tags.env.status) end,
  73         unit: {:native, :second},
  74         reporter_options: [
  75           buckets: [0.01, 0.025, 0.05, 0.1, 0.2, 0.5, 1, 2.5, 5, 10]
  76         ]
  77       ),
  78       distribution(
  79         "oban_job_completion",
  80         event_name: [:oban, :job, :stop],
  81         measurement: :duration,
  82         tags: [:worker],
  83         tag_values: fn tags -> Map.put(tags, :worker, tags.job.worker) end,
  84         unit: {:native, :second},
  85         reporter_options: [
  86           buckets: [0.01, 0.025, 0.05, 0.1, 0.2, 0.5, 1, 2.5, 5, 10]
  87         ]
  88       )
  89     ]
  90   end
  91
  92   defp summary_metrics do
  93     [
  94       # Phoenix Metrics
  95       summary("phoenix.endpoint.stop.duration",
  96         unit: {:native, :millisecond}
  97       ),
  98       summary("phoenix.router_dispatch.stop.duration",
  99         tags: [:route],
 100         unit: {:native, :millisecond}
 101       ),
 102       summary("pleroma.repo.query.total_time", unit: {:native, :millisecond}),
 103       summary("pleroma.repo.query.decode_time", unit: {:native, :millisecond}),
 104       summary("pleroma.repo.query.query_time", unit: {:native, :millisecond}),
 105       summary("pleroma.repo.query.queue_time", unit: {:native, :millisecond}),
 106       summary("pleroma.repo.query.idle_time", unit: {:native, :millisecond}),
 107
 108       # VM Metrics
 109       summary("vm.memory.total", unit: {:byte, :kilobyte}),
 110       summary("vm.total_run_queue_lengths.total"),
 111       summary("vm.total_run_queue_lengths.cpu"),
 112       summary("vm.total_run_queue_lengths.io"),
 113
 114
 115       last_value("pleroma.local_users.total"),
 116       last_value("pleroma.domains.total"),
 117       last_value("pleroma.local_statuses.total")
 118     ]
 119   end
 120
 121   def prometheus_metrics, do: summary_metrics() ++ distribution_metrics()
 122   def live_dashboard_metrics, do: summary_metrics()
 123
 124   defp periodic_measurements do
 125     [
 126       {__MODULE__, :instance_stats, []}
 127     ]
 128   end
 129
 130   def instance_stats do
 131     stats = Stats.get_stats()
 132     :telemetry.execute([:pleroma, :local_users], %{total: stats.user_count}, %{})
 133     :telemetry.execute([:pleroma, :domains], %{total: stats.domain_count}, %{})
 134     :telemetry.execute([:pleroma, :local_statuses], %{total: stats.status_count}, %{})
 135   end
 136 end