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