View Source Task.Supervisor (Elixir v1.13.0)

A task supervisor.

This module defines a supervisor which can be used to dynamically supervise tasks.

A task supervisor is started with no children, often under a supervisor and a name:

children = [
  {Task.Supervisor, name: MyApp.TaskSupervisor}
]

Supervisor.start_link(children, strategy: :one_for_one)

The options given in the child specification are documented in start_link/1.

See the Task module for more examples.

Name registration

A Task.Supervisor is bound to the same name registration rules as a GenServer. Read more about them in the GenServer docs.

Link to this section Summary

Types

Option values used by start_link

Functions

Starts a task that can be awaited on.

Starts a task that can be awaited on.

Starts a task that can be awaited on.

Starts a task that can be awaited on.

Returns a stream that runs the given function fun concurrently on each element in enumerable.

Returns a stream where the given function (module and function) is mapped concurrently on each element in enumerable.

Returns a stream that runs the given function concurrently on each element in enumerable.

Returns a stream where the given function (module and function) is mapped concurrently on each element in enumerable.

Returns all children PIDs.

Starts a task as a child of the given supervisor.

Starts a task as a child of the given supervisor.

Starts a new supervisor.

Terminates the child with the given pid.

Link to this section Types

Option values used by start_link

Link to this section Functions

Link to this function

async(supervisor, fun, options \\ [])

View Source
@spec async(Supervisor.supervisor(), (() -> any()), Keyword.t()) :: Task.t()

Starts a task that can be awaited on.

The supervisor must be a reference as defined in Supervisor. The task will still be linked to the caller, see Task.async/3 for more information and async_nolink/3 for a non-linked variant.

Raises an error if supervisor has reached the maximum number of children.

Options

  • :shutdown - :brutal_kill if the tasks must be killed directly on shutdown or an integer indicating the timeout value, defaults to 5000 milliseconds.
Link to this function

async(supervisor, module, fun, args, options \\ [])

View Source
@spec async(Supervisor.supervisor(), module(), atom(), [term()], Keyword.t()) ::
  Task.t()

Starts a task that can be awaited on.

The supervisor must be a reference as defined in Supervisor. The task will still be linked to the caller, see Task.async/3 for more information and async_nolink/3 for a non-linked variant.

Raises an error if supervisor has reached the maximum number of children.

Options

  • :shutdown - :brutal_kill if the tasks must be killed directly on shutdown or an integer indicating the timeout value, defaults to 5000 milliseconds.
Link to this function

async_nolink(supervisor, fun, options \\ [])

View Source
@spec async_nolink(Supervisor.supervisor(), (() -> any()), Keyword.t()) :: Task.t()

Starts a task that can be awaited on.

The supervisor must be a reference as defined in Supervisor. The task won't be linked to the caller, see Task.async/3 for more information.

Raises an error if supervisor has reached the maximum number of children.

Options

  • :shutdown - :brutal_kill if the tasks must be killed directly on shutdown or an integer indicating the timeout value, defaults to 5000 milliseconds.

Compatibility with OTP behaviours

If you create a task using async_nolink inside an OTP behaviour like GenServer, you should match on the message coming from the task inside your GenServer.handle_info/2 callback.

The reply sent by the task will be in the format {ref, result}, where ref is the monitor reference held by the task struct and result is the return value of the task function.

Keep in mind that, regardless of how the task created with async_nolink terminates, the caller's process will always receive a :DOWN message with the same ref value that is held by the task struct. If the task terminates normally, the reason in the :DOWN message will be :normal.

Examples

Typically, you use async_nolink/3 when there is a reasonable expectation that the task may fail, and you don't want it to take down the caller. Let's see an example where a GenServer is meant to run a single task and track its status:

defmodule MyApp.Server do
  use GenServer

  # ...

  def start_task do
    GenServer.call(__MODULE__, :start_task)
  end

  # In this case the task is already running, so we just return :ok.
  def handle_call(:start_task, _from, %{ref: ref} = state) when is_reference(ref) do
    {:reply, :ok, state}
  end

  # The task is not running yet, so let's start it.
  def handle_call(:start_task, _from, %{ref: nil} = state) do
    task =
      Task.Supervisor.async_nolink(MyApp.TaskSupervisor, fn ->
        ...
      end)

    # We return :ok and the server will continue running
    {:reply, :ok, %{state | ref: task.ref}}
  end

  # The task completed successfully
  def handle_info({ref, answer}, %{ref: ref} = state) do
    # We don't care about the DOWN message now, so let's demonitor and flush it
    Process.demonitor(ref, [:flush])
    # Do something with the result and then return
    {:noreply, %{state | ref: nil}}
  end

  # The task failed
  def handle_info({:DOWN, ref, :process, _pid, _reason}, %{ref: ref} = state) do
    # Log and possibly restart the task...
    {:noreply, %{state | ref: nil}}
  end
end
Link to this function

async_nolink(supervisor, module, fun, args, options \\ [])

View Source
@spec async_nolink(Supervisor.supervisor(), module(), atom(), [term()], Keyword.t()) ::
  Task.t()

Starts a task that can be awaited on.

The supervisor must be a reference as defined in Supervisor. The task won't be linked to the caller, see Task.async/3 for more information.

Raises an error if supervisor has reached the maximum number of children.

Note this function requires the task supervisor to have :temporary as the :restart option (the default), as async_nolink/5 keeps a direct reference to the task which is lost if the task is restarted.

Link to this function

async_stream(supervisor, enumerable, fun, options \\ [])

View Source (since 1.4.0)
@spec async_stream(
  Supervisor.supervisor(),
  Enumerable.t(),
  (term() -> term()),
  keyword()
) :: Enumerable.t()

Returns a stream that runs the given function fun concurrently on each element in enumerable.

Each element in enumerable is passed as argument to the given function fun and processed by its own task. The tasks will be spawned under the given supervisor and linked to the caller process, similarly to async/3.

See async_stream/6 for discussion, options, and examples.

Link to this function

async_stream(supervisor, enumerable, module, function, args, options \\ [])

View Source (since 1.4.0)
@spec async_stream(
  Supervisor.supervisor(),
  Enumerable.t(),
  module(),
  atom(),
  [term()],
  keyword()
) :: Enumerable.t()

Returns a stream where the given function (module and function) is mapped concurrently on each element in enumerable.

Each element will be prepended to the given args and processed by its own task. The tasks will be spawned under the given supervisor and linked to the caller process, similarly to async/5.

When streamed, each task will emit {:ok, value} upon successful completion or {:exit, reason} if the caller is trapping exits. The order of results depends on the value of the :ordered option.

The level of concurrency and the time tasks are allowed to run can be controlled via options (see the "Options" section below).

If you find yourself trapping exits to handle exits inside the async stream, consider using async_stream_nolink/6 to start tasks that are not linked to the calling process.

Options

  • :max_concurrency - sets the maximum number of tasks to run at the same time. Defaults to System.schedulers_online/0.

  • :ordered - whether the results should be returned in the same order as the input stream. This option is useful when you have large streams and don't want to buffer results before they are delivered. This is also useful when you're using the tasks for side effects. Defaults to true.

  • :timeout - the maximum amount of time to wait (in milliseconds) without receiving a task reply (across all running tasks). Defaults to 5000.

  • :on_timeout - what do to when a task times out. The possible values are:

    • :exit (default) - the process that spawned the tasks exits.
    • :kill_task - the task that timed out is killed. The value emitted for that task is {:exit, :timeout}.
  • :shutdown - :brutal_kill if the tasks must be killed directly on shutdown or an integer indicating the timeout value. Defaults to 5000 milliseconds.

Examples

Let's build a stream and then enumerate it:

stream = Task.Supervisor.async_stream(MySupervisor, collection, Mod, :expensive_fun, [])
Enum.to_list(stream)
Link to this function

async_stream_nolink(supervisor, enumerable, fun, options \\ [])

View Source (since 1.4.0)
@spec async_stream_nolink(
  Supervisor.supervisor(),
  Enumerable.t(),
  (term() -> term()),
  keyword()
) :: Enumerable.t()

Returns a stream that runs the given function concurrently on each element in enumerable.

Each element in enumerable is passed as argument to the given function fun and processed by its own task. The tasks will be spawned under the given supervisor and will not be linked to the caller process, similarly to async_nolink/3.

See async_stream/6 for discussion and examples.

Link to this function

async_stream_nolink(supervisor, enumerable, module, function, args, options \\ [])

View Source (since 1.4.0)
@spec async_stream_nolink(
  Supervisor.supervisor(),
  Enumerable.t(),
  module(),
  atom(),
  [term()],
  keyword()
) :: Enumerable.t()

Returns a stream where the given function (module and function) is mapped concurrently on each element in enumerable.

Each element in enumerable will be prepended to the given args and processed by its own task. The tasks will be spawned under the given supervisor and will not be linked to the caller process, similarly to async_nolink/5.

See async_stream/6 for discussion, options, and examples.

@spec children(Supervisor.supervisor()) :: [pid()]

Returns all children PIDs.

Link to this function

start_child(supervisor, fun, options \\ [])

View Source
@spec start_child(Supervisor.supervisor(), (() -> any()), keyword()) ::
  DynamicSupervisor.on_start_child()

Starts a task as a child of the given supervisor.

Task.Supervisor.start_child(MyTaskSupervisor, fn ->
  IO.puts "I am running in a task"
end)

Note that the spawned process is not linked to the caller, but only to the supervisor. This command is useful in case the task needs to perform side-effects (like I/O) and you have no interest in its results nor if it completes successfully.

Options

  • :restart - the restart strategy, may be :temporary (the default), :transient or :permanent. :temporary means the task is never restarted, :transient means it is restarted if the exit is not :normal, :shutdown or {:shutdown, reason}. A :permanent restart strategy means it is always restarted.

  • :shutdown - :brutal_kill if the task must be killed directly on shutdown or an integer indicating the timeout value, defaults to 5000 milliseconds.

Link to this function

start_child(supervisor, module, fun, args, options \\ [])

View Source

Starts a task as a child of the given supervisor.

Similar to start_child/3 except the task is specified by the given module, fun and args.

Link to this function

start_link(options \\ [])

View Source
@spec start_link([option()]) :: Supervisor.on_start()

Starts a new supervisor.

Examples

A task supervisor is typically started under a supervision tree using the tuple format:

{Task.Supervisor, name: MyApp.TaskSupervisor}

You can also start it by calling start_link/1 directly:

Task.Supervisor.start_link(name: MyApp.TaskSupervisor)

But this is recommended only for scripting and should be avoided in production code. Generally speaking, processes should always be started inside supervision trees.

Options

  • :name - used to register a supervisor name, the supported values are described under the Name Registration section in the GenServer module docs;

  • :max_restarts, :max_seconds, and :max_children - as specified in DynamicSupervisor;

This function could also receive :restart and :shutdown as options but those two options have been deprecated and it is now preferred to give them directly to start_child.

Link to this function

terminate_child(supervisor, pid)

View Source
@spec terminate_child(Supervisor.supervisor(), pid()) :: :ok | {:error, :not_found}

Terminates the child with the given pid.