Merge branch 'feature/jobs' into 'develop'

Job Queue

See merge request !732

config/config.exs

@@ -332,14 +332,16 @@
     "web"
   ]
 
-config :pleroma, Pleroma.Web.Federator, max_jobs: 50
-
 config :pleroma, Pleroma.Web.Federator.RetryQueue,
   enabled: false,
   max_jobs: 20,
   initial_timeout: 30,
   max_retries: 5
 
+config :pleroma, Pleroma.Jobs,
+  federator_incoming: [max_jobs: 50],
+  federator_outgoing: [max_jobs: 50]
+
 # Import environment specific config. This must remain at the bottom
 # of this file so it overrides the configuration defined above.
 import_config "#{Mix.env()}.exs"

config/test.exs

@@ -44,6 +44,8 @@
     "BLH1qVhJItRGCfxgTtONfsOKDc9VRAraXw-3NsmjMngWSh7NxOizN6bkuRA7iLTMPS82PjwJAr3UoK9EC1IFrz4",
   private_key: "_-XZ0iebPrRfZ_o0-IatTdszYa8VCH1yLN-JauK7HHA"
 
+config :pleroma, Pleroma.Jobs, testing: [max_jobs: 2]
+
 try do
   import_config "test.secret.exs"
 rescue

docs/config.md

@@ -36,14 +36,15 @@ This filter replaces the filename (not the path) of an upload. For complete obfu
 
 An example for Sendgrid adapter:
 
-```
+```exs
 config :pleroma, Pleroma.Mailer,
   adapter: Swoosh.Adapters.Sendgrid,
   api_key: "YOUR_API_KEY"
 ```
 
 An example for SMTP adapter:
-```
+
+```exs
 config :pleroma, Pleroma.Mailer,
   adapter: Swoosh.Adapters.SMTP,
   relay: "smtp.gmail.com",
@@ -209,7 +210,7 @@ their ActivityPub ID.
 
 An example:
 
-```
+```exs
 config :pleroma, :mrf_user_allowlist,
   "example.org": ["https://example.org/users/admin"]
 ```
@@ -238,18 +239,34 @@ the source code is here: https://github.com/koto-bank/kocaptcha. The default end
 
 Allows to set a token that can be used to authenticate with the admin api without using an actual user by giving it as the 'admin_token' parameter. Example:
 
-```
+```exs
 config :pleroma, :admin_token, "somerandomtoken"
 ```
 
 You can then do
-```
+
+```sh
 curl "http://localhost:4000/api/pleroma/admin/invite_token?admin_token=somerandomtoken"
 ```
 
-## Pleroma.Web.Federator
+## Pleroma.Jobs
+
+A list of job queues and their settings.
+
+Job queue settings:
+
+* `max_jobs`: The maximum amount of parallel jobs running at the same time.
+
+Example:
+
+```exs
+config :pleroma, Pleroma.Jobs,
+  federator_incoming: [max_jobs: 50],
+  federator_outgoing: [max_jobs: 50]
+```
+
+This config contains two queues: `federator_incoming` and `federator_outgoing`. Both have the `max_jobs` set to `50`.
 
-* `max_jobs`: The maximum amount of parallel federation jobs running at the same time.
 
 ## Pleroma.Web.Federator.RetryQueue
 

lib/pleroma/application.ex

@@ -108,9 +108,10 @@ def start(_type, _args) do
         hackney_pool_children() ++
         [
           worker(Pleroma.Web.Federator.RetryQueue, []),
-          worker(Pleroma.Web.Federator, []),
           worker(Pleroma.Stats, []),
-          worker(Pleroma.Web.Push, [])
+          worker(Pleroma.Web.Push, []),
+          worker(Pleroma.Jobs, []),
+          worker(Task, [&Pleroma.Web.Federator.init/0], restart: :temporary)
         ] ++
         streamer_child() ++
         chat_child() ++

lib/pleroma/jobs.ex

+# Pleroma: A lightweight social networking server
+# Copyright © 2017-2019 Pleroma Authors <https://pleroma.social/>
+# SPDX-License-Identifier: AGPL-3.0-only
+
+defmodule Pleroma.Jobs do
+  @moduledoc """
+  A basic job queue
+  """
+  use GenServer
+
+  require Logger
+
+  def init(args) do
+    {:ok, args}
+  end
+
+  def start_link do
+    queues =
+      Pleroma.Config.get(Pleroma.Jobs)
+      |> Enum.map(fn {name, _} -> create_queue(name) end)
+      |> Enum.into(%{})
+
+    state = %{
+      queues: queues,
+      refs: %{}
+    }
+
+    GenServer.start_link(__MODULE__, state, name: __MODULE__)
+  end
+
+  def create_queue(name) do
+    {name, {:sets.new(), []}}
+  end
+
+  @doc """
+  Enqueues a job.
+
+  Returns `:ok`.
+
+  ## Arguments
+
+  - `queue_name` - a queue name(must be specified in the config).
+  - `mod` - a worker module (must have `perform` function).
+  - `args` - a list of arguments for the `perform` function of the worker module.
+  - `priority` - a job priority (`0` by default).
+
+  ## Examples
+
+  Enqueue `Module.perform/0` with `priority=1`:
+
+      iex> Pleroma.Jobs.enqueue(:example_queue, Module, [])
+      :ok
+
+  Enqueue `Module.perform(:job_name)` with `priority=5`:
+
+      iex> Pleroma.Jobs.enqueue(:example_queue, Module, [:job_name], 5)
+      :ok
+
+  Enqueue `Module.perform(:another_job, data)` with `priority=1`:
+
+      iex> data = "foobar"
+      iex> Pleroma.Jobs.enqueue(:example_queue, Module, [:another_job, data])
+      :ok
+
+  Enqueue `Module.perform(:foobar_job, :foo, :bar, 42)` with `priority=1`:
+
+      iex> Pleroma.Jobs.enqueue(:example_queue, Module, [:foobar_job, :foo, :bar, 42])
+      :ok
+
+  """
+
+  def enqueue(queue_name, mod, args, priority \\ 1)
+
+  if Mix.env() == :test do
+    def enqueue(_queue_name, mod, args, _priority) do
+      apply(mod, :perform, args)
+    end
+  else
+    @spec enqueue(atom(), atom(), [any()], integer()) :: :ok
+    def enqueue(queue_name, mod, args, priority) do
+      GenServer.cast(__MODULE__, {:enqueue, queue_name, mod, args, priority})
+    end
+  end
+
+  def handle_cast({:enqueue, queue_name, mod, args, priority}, state) do
+    {running_jobs, queue} = state[:queues][queue_name]
+
+    queue = enqueue_sorted(queue, {mod, args}, priority)
+
+    state =
+      state
+      |> update_queue(queue_name, {running_jobs, queue})
+      |> maybe_start_job(queue_name, running_jobs, queue)
+
+    {:noreply, state}
+  end
+
+  def handle_info({:DOWN, ref, :process, _pid, _reason}, state) do
+    queue_name = state.refs[ref]
+
+    {running_jobs, queue} = state[:queues][queue_name]
+
+    running_jobs = :sets.del_element(ref, running_jobs)
+
+    state =
+      state
+      |> remove_ref(ref)
+      |> update_queue(queue_name, {running_jobs, queue})
+      |> maybe_start_job(queue_name, running_jobs, queue)
+
+    {:noreply, state}
+  end
+
+  def maybe_start_job(state, queue_name, running_jobs, queue) do
+    if :sets.size(running_jobs) < Pleroma.Config.get([__MODULE__, queue_name, :max_jobs]) &&
+         queue != [] do
+      {{mod, args}, queue} = queue_pop(queue)
+      {:ok, pid} = Task.start(fn -> apply(mod, :perform, args) end)
+      mref = Process.monitor(pid)
+
+      state
+      |> add_ref(queue_name, mref)
+      |> update_queue(queue_name, {:sets.add_element(mref, running_jobs), queue})
+    else
+      state
+    end
+  end
+
+  def enqueue_sorted(queue, element, priority) do
+    [%{item: element, priority: priority} | queue]
+    |> Enum.sort_by(fn %{priority: priority} -> priority end)
+  end
+
+  def queue_pop([%{item: element} | queue]) do
+    {element, queue}
+  end
+
+  defp add_ref(state, queue_name, ref) do
+    refs = Map.put(state[:refs], ref, queue_name)
+    Map.put(state, :refs, refs)
+  end
+
+  defp remove_ref(state, ref) do
+    refs = Map.delete(state[:refs], ref)
+    Map.put(state, :refs, refs)
+  end
+
+  defp update_queue(state, queue_name, data) do
+    queues = Map.put(state[:queues], queue_name, data)
+    Map.put(state, :queues, queues)
+  end
+end

lib/pleroma/web/activity_pub/activity_pub.ex

@@ -757,21 +757,19 @@ def publish(actor, activity) do
 
     public = is_public?(activity)
 
-    reachable_inboxes_metadata =
-      (Pleroma.Web.Salmon.remote_users(activity) ++ remote_followers)
-      |> Enum.filter(fn user -> User.ap_enabled?(user) end)
-      |> Enum.map(fn %{info: %{source_data: data}} ->
-        (is_map(data["endpoints"]) && Map.get(data["endpoints"], "sharedInbox")) || data["inbox"]
-      end)
-      |> Enum.uniq()
-      |> Enum.filter(fn inbox -> should_federate?(inbox, public) end)
-      |> Instances.filter_reachable()
-
     {:ok, data} = Transmogrifier.prepare_outgoing(activity.data)
     json = Jason.encode!(data)
 
-    Enum.each(reachable_inboxes_metadata, fn {inbox, unreachable_since} ->
-      Federator.enqueue(:publish_single_ap, %{
+    (Pleroma.Web.Salmon.remote_users(activity) ++ remote_followers)
+    |> Enum.filter(fn user -> User.ap_enabled?(user) end)
+    |> Enum.map(fn %{info: %{source_data: data}} ->
+      (is_map(data["endpoints"]) && Map.get(data["endpoints"], "sharedInbox")) || data["inbox"]
+    end)
+    |> Enum.uniq()
+    |> Enum.filter(fn inbox -> should_federate?(inbox, public) end)
+    |> Instances.filter_reachable()
+    |> Enum.each(fn {inbox, unreachable_since} ->
+      Federator.publish_single_ap(%{
         inbox: inbox,
         json: json,
         actor: actor,

lib/pleroma/web/activity_pub/activity_pub_controller.ex

@@ -155,13 +155,13 @@ def inbox(%{assigns: %{valid_signature: true}} = conn, %{"nickname" => nickname}
     with %User{} = user <- User.get_cached_by_nickname(nickname),
          true <- Utils.recipient_in_message(user.ap_id, params),
          params <- Utils.maybe_splice_recipient(user.ap_id, params) do
-      Federator.enqueue(:incoming_ap_doc, params)
+      Federator.incoming_ap_doc(params)
       json(conn, "ok")
     end
   end
 
   def inbox(%{assigns: %{valid_signature: true}} = conn, params) do
-    Federator.enqueue(:incoming_ap_doc, params)
+    Federator.incoming_ap_doc(params)
     json(conn, "ok")
   end
 

lib/pleroma/web/activity_pub/utils.ex

@@ -164,7 +164,7 @@ def maybe_federate(%Activity{local: true} = activity) do
         _ -> 5
       end
 
-    Pleroma.Web.Federator.enqueue(:publish, activity, priority)
+    Pleroma.Web.Federator.publish(activity, priority)
     :ok
   end
 

lib/pleroma/web/federator/federator.ex

@@ -3,8 +3,6 @@
 # SPDX-License-Identifier: AGPL-3.0-only
 
 defmodule Pleroma.Web.Federator do
-  use GenServer
-
   alias Pleroma.Activity
   alias Pleroma.User
   alias Pleroma.Web.WebFinger
@@ -16,45 +14,71 @@ defmodule Pleroma.Web.Federator do
   alias Pleroma.Web.ActivityPub.Utils
   alias Pleroma.Web.Federator.RetryQueue
   alias Pleroma.Web.OStatus
+  alias Pleroma.Jobs
 
   require Logger
 
   @websub Application.get_env(:pleroma, :websub)
   @ostatus Application.get_env(:pleroma, :ostatus)
 
-  def init(args) do
-    {:ok, args}
+  def init() do
+    # 1 minute
+    Process.sleep(1000 * 60 * 1)
+    refresh_subscriptions()
   end
 
-  def start_link do
-    spawn(fn ->
-      # 1 minute
-      Process.sleep(1000 * 60)
-      enqueue(:refresh_subscriptions, nil)
-    end)
+  # Client API
+
+  def incoming_doc(doc) do
+    Jobs.enqueue(:federator_incoming, __MODULE__, [:incoming_doc, doc])
+  end
+
+  def incoming_ap_doc(params) do
+    Jobs.enqueue(:federator_incoming, __MODULE__, [:incoming_ap_doc, params])
+  end
+
+  def publish(activity, priority \\ 1) do
+    Jobs.enqueue(:federator_outgoing, __MODULE__, [:publish, activity], priority)
+  end
+
+  def publish_single_ap(params) do
+    Jobs.enqueue(:federator_outgoing, __MODULE__, [:publish_single_ap, params])
+  end
 
-    GenServer.start_link(
-      __MODULE__,
-      %{
-        in: {:sets.new(), []},
-        out: {:sets.new(), []}
-      },
-      name: __MODULE__
-    )
+  def publish_single_websub(websub) do
+    Jobs.enqueue(:federator_outgoing, __MODULE__, [:publish_single_websub, websub])
   end
 
-  def handle(:refresh_subscriptions, _) do
+  def verify_websub(websub) do
+    Jobs.enqueue(:federator_outgoing, __MODULE__, [:verify_websub, websub])
+  end
+
+  def request_subscription(sub) do
+    Jobs.enqueue(:federator_outgoing, __MODULE__, [:request_subscription, sub])
+  end
+
+  def refresh_subscriptions() do
+    Jobs.enqueue(:federator_outgoing, __MODULE__, [:refresh_subscriptions])
+  end
+
+  def publish_single_salmon(params) do
+    Jobs.enqueue(:federator_outgoing, __MODULE__, [:publish_single_salmon, params])
+  end
+
+  # Job Worker Callbacks
+
+  def perform(:refresh_subscriptions) do
     Logger.debug("Federator running refresh subscriptions")
     Websub.refresh_subscriptions()
 
     spawn(fn ->
       # 6 hours
       Process.sleep(1000 * 60 * 60 * 6)
-      enqueue(:refresh_subscriptions, nil)
+      refresh_subscriptions()
     end)
   end
 
-  def handle(:request_subscription, websub) do
+  def perform(:request_subscription, websub) do
     Logger.debug("Refreshing #{websub.topic}")
 
     with {:ok, websub} <- Websub.request_subscription(websub) do
@@ -64,7 +88,7 @@ def handle(:request_subscription, websub) do
     end
   end
 
-  def handle(:publish, activity) do
+  def perform(:publish, activity) do
     Logger.debug(fn -> "Running publish for #{activity.data["id"]}" end)
 
     with actor when not is_nil(actor) <- User.get_cached_by_ap_id(activity.data["actor"]) do
@@ -90,7 +114,7 @@ def handle(:publish, activity) do
     end
   end
 
-  def handle(:verify_websub, websub) do
+  def perform(:verify_websub, websub) do
     Logger.debug(fn ->
       "Running WebSub verification for #{websub.id} (#{websub.topic}, #{websub.callback})"
     end)
@@ -98,12 +122,12 @@ def handle(:verify_websub, websub) do
     @websub.verify(websub)
   end
 
-  def handle(:incoming_doc, doc) do
+  def perform(:incoming_doc, doc) do
     Logger.info("Got document, trying to parse")
     @ostatus.handle_incoming(doc)
   end
 
-  def handle(:incoming_ap_doc, params) do
+  def perform(:incoming_ap_doc, params) do
     Logger.info("Handling incoming AP activity")
 
     params = Utils.normalize_params(params)
@@ -128,11 +152,11 @@ def handle(:incoming_ap_doc, params) do
     end
   end
 
-  def handle(:publish_single_salmon, params) do
+  def perform(:publish_single_salmon, params) do
     Salmon.send_to_user(params)
   end
 
-  def handle(:publish_single_ap, params) do
+  def perform(:publish_single_ap, params) do
     case ActivityPub.publish_one(params) do
       {:ok, _} ->
         :ok
@@ -142,7 +166,7 @@ def handle(:publish_single_ap, params) do
     end
   end
 
-  def handle(
+  def perform(
         :publish_single_websub,
         %{xml: _xml, topic: _topic, callback: _callback, secret: _secret} = params
       ) do
@@ -155,74 +179,11 @@ def handle(
     end
   end
 
-  def handle(type, _) do
+  def perform(type, _) do
     Logger.debug(fn -> "Unknown task: #{type}" end)
     {:error, "Don't know what to do with this"}
   end
 
-  if Mix.env() == :test do
-    def enqueue(type, payload, _priority \\ 1) do
-      if Pleroma.Config.get([:instance, :federating]) do
-        handle(type, payload)
-      end
-    end
-  else
-    def enqueue(type, payload, priority \\ 1) do
-      if Pleroma.Config.get([:instance, :federating]) do
-        GenServer.cast(__MODULE__, {:enqueue, type, payload, priority})
-      end
-    end
-  end
-
-  def maybe_start_job(running_jobs, queue) do
-    if :sets.size(running_jobs) < Pleroma.Config.get([__MODULE__, :max_jobs]) && queue != [] do
-      {{type, payload}, queue} = queue_pop(queue)
-      {:ok, pid} = Task.start(fn -> handle(type, payload) end)
-      mref = Process.monitor(pid)
-      {:sets.add_element(mref, running_jobs), queue}
-    else
-      {running_jobs, queue}
-    end
-  end
-
-  def handle_cast({:enqueue, type, payload, _priority}, state)
-      when type in [:incoming_doc, :incoming_ap_doc] do
-    %{in: {i_running_jobs, i_queue}, out: {o_running_jobs, o_queue}} = state
-    i_queue = enqueue_sorted(i_queue, {type, payload}, 1)
-    {i_running_jobs, i_queue} = maybe_start_job(i_running_jobs, i_queue)
-    {:noreply, %{in: {i_running_jobs, i_queue}, out: {o_running_jobs, o_queue}}}
-  end
-
-  def handle_cast({:enqueue, type, payload, _priority}, state) do
-    %{in: {i_running_jobs, i_queue}, out: {o_running_jobs, o_queue}} = state
-    o_queue = enqueue_sorted(o_queue, {type, payload}, 1)
-    {o_running_jobs, o_queue} = maybe_start_job(o_running_jobs, o_queue)
-    {:noreply, %{in: {i_running_jobs, i_queue}, out: {o_running_jobs, o_queue}}}
-  end
-
-  def handle_cast(_, state) do
-    {:noreply, state}
-  end
-
-  def handle_info({:DOWN, ref, :process, _pid, _reason}, state) do
-    %{in: {i_running_jobs, i_queue}, out: {o_running_jobs, o_queue}} = state
-    i_running_jobs = :sets.del_element(ref, i_running_jobs)
-    o_running_jobs = :sets.del_element(ref, o_running_jobs)
-    {i_running_jobs, i_queue} = maybe_start_job(i_running_jobs, i_queue)
-    {o_running_jobs, o_queue} = maybe_start_job(o_running_jobs, o_queue)
-
-    {:noreply, %{in: {i_running_jobs, i_queue}, out: {o_running_jobs, o_queue}}}
-  end
-
-  def enqueue_sorted(queue, element, priority) do
-    [%{item: element, priority: priority} | queue]
-    |> Enum.sort_by(fn %{priority: priority} -> priority end)
-  end
-
-  def queue_pop([%{item: element} | queue]) do
-    {element, queue}
-  end
-
   def ap_enabled_actor(id) do
     user = User.get_by_ap_id(id)
 

lib/pleroma/web/ostatus/ostatus_controller.ex

@@ -87,7 +87,7 @@ def salmon_incoming(conn, _) do
     {:ok, body, _conn} = read_body(conn)
     {:ok, doc} = decode_or_retry(body)
 
-    Federator.enqueue(:incoming_doc, doc)
+    Federator.incoming_doc(doc)
 
     conn
     |> send_resp(200, "")

lib/pleroma/web/salmon/salmon.ex

@@ -229,7 +229,7 @@ def publish(%{info: %{keys: keys}} = user, %{data: %{"type" => type}} = activity
       |> Enum.each(fn remote_user ->
         Logger.debug(fn -> "Sending Salmon to #{remote_user.ap_id}" end)
 
-        Pleroma.Web.Federator.enqueue(:publish_single_salmon, %{
+        Pleroma.Web.Federator.publish_single_salmon(%{
           recipient: remote_user,
           feed: feed,
           poster: poster,

lib/pleroma/web/websub/websub.ex

@@ -13,6 +13,7 @@ defmodule Pleroma.Web.Websub do
   alias Pleroma.Web.Endpoint
   alias Pleroma.Web.OStatus
   alias Pleroma.Web.Router.Helpers
+  alias Pleroma.Web.Federator
   require Logger
 
   import Ecto.Query
@@ -87,7 +88,7 @@ def publish(topic, user, %{data: %{"type" => type}} = activity)
         unreachable_since: reachable_callbacks_metadata[sub.callback]
       }
 
-      Pleroma.Web.Federator.enqueue(:publish_single_websub, data)
+      Federator.publish_single_websub(data)
     end)
   end
 
@@ -119,7 +120,7 @@ def incoming_subscription_request(user, %{"hub.mode" => "subscribe"} = params) d
 
       websub = Repo.update!(change)
 
-      Pleroma.Web.Federator.enqueue(:verify_websub, websub)
+      Federator.verify_websub(websub)
 
       {:ok, websub}
     else
@@ -269,7 +270,7 @@ def refresh_subscriptions(delta \\ 60 * 60 * 24) do
     subs = Repo.all(query)
 
     Enum.each(subs, fn sub ->
-      Pleroma.Web.Federator.enqueue(:request_subscription, sub)
+      Federator.request_subscription(sub)
     end)
   end
 

lib/pleroma/web/websub/websub_controller.ex

@@ -84,7 +84,7 @@ def websub_incoming(conn, %{"id" => id}) do