Struct tokio_util::task::task_tracker::TaskTracker
source · pub struct TaskTracker { /* private fields */ }
rt
only.Expand description
A task tracker used for waiting until tasks exit.
This is usually used together with CancellationToken
to implement graceful shutdown. The
CancellationToken
is used to signal to tasks that they should shut down, and the
TaskTracker
is used to wait for them to finish shutting down.
The TaskTracker
will also keep track of a closed
boolean. This is used to handle the case
where the TaskTracker
is empty, but we don’t want to shut down yet. This means that the
wait
method will wait until both of the following happen at the same time:
- The
TaskTracker
must be closed using theclose
method. - The
TaskTracker
must be empty, that is, all tasks that it is tracking must have exited.
When a call to wait
returns, it is guaranteed that all tracked tasks have exited and that
the destructor of the future has finished running. However, there might be a short amount of
time where JoinHandle::is_finished
returns false.
§Comparison to JoinSet
The main Tokio crate has a similar collection known as JoinSet
. The JoinSet
type has a
lot more features than TaskTracker
, so TaskTracker
should only be used when one of its
unique features is required:
- When tasks exit, a
TaskTracker
will allow the task to immediately free its memory. - By not closing the
TaskTracker
,wait
will be prevented from from returning even if theTaskTracker
is empty. - A
TaskTracker
does not require mutable access to insert tasks. - A
TaskTracker
can be cloned to share it with many tasks.
The first point is the most important one. A JoinSet
keeps track of the return value of
every inserted task. This means that if the caller keeps inserting tasks and never calls
join_next
, then their return values will keep building up and consuming memory, even if
most of the tasks have already exited. This can cause the process to run out of memory. With a
TaskTracker
, this does not happen. Once tasks exit, they are immediately removed from the
TaskTracker
.
§Examples
For more examples, please see the topic page on graceful shutdown.
§Spawn tasks and wait for them to exit
This is a simple example. For this case, JoinSet
should probably be used instead.
use tokio_util::task::TaskTracker;
#[tokio::main]
async fn main() {
let tracker = TaskTracker::new();
for i in 0..10 {
tracker.spawn(async move {
println!("Task {} is running!", i);
});
}
// Once we spawned everything, we close the tracker.
tracker.close();
// Wait for everything to finish.
tracker.wait().await;
println!("This is printed after all of the tasks.");
}
§Wait for tasks to exit
This example shows the intended use-case of TaskTracker
. It is used together with
CancellationToken
to implement graceful shutdown.
use tokio_util::sync::CancellationToken;
use tokio_util::task::TaskTracker;
use tokio::time::{self, Duration};
async fn background_task(num: u64) {
for i in 0..10 {
time::sleep(Duration::from_millis(100*num)).await;
println!("Background task {} in iteration {}.", num, i);
}
}
#[tokio::main]
async fn main() {
let tracker = TaskTracker::new();
let token = CancellationToken::new();
for i in 0..10 {
let token = token.clone();
tracker.spawn(async move {
// Use a `tokio::select!` to kill the background task if the token is
// cancelled.
tokio::select! {
() = background_task(i) => {
println!("Task {} exiting normally.", i);
},
() = token.cancelled() => {
// Do some cleanup before we really exit.
time::sleep(Duration::from_millis(50)).await;
println!("Task {} finished cleanup.", i);
},
}
});
}
// Spawn a background task that will send the shutdown signal.
{
let tracker = tracker.clone();
tokio::spawn(async move {
// Normally you would use something like ctrl-c instead of
// sleeping.
time::sleep(Duration::from_secs(2)).await;
tracker.close();
token.cancel();
});
}
// Wait for all tasks to exit.
tracker.wait().await;
println!("All tasks have exited now.");
}
Implementations§
source§impl TaskTracker
impl TaskTracker
sourcepub fn wait(&self) -> TaskTrackerWaitFuture<'_> ⓘ
pub fn wait(&self) -> TaskTrackerWaitFuture<'_> ⓘ
Waits until this TaskTracker
is both closed and empty.
If the TaskTracker
is already closed and empty when this method is called, then it
returns immediately.
The wait
future is resistant against ABA problems. That is, if the TaskTracker
becomes both closed and empty for a short amount of time, then it is guarantee that all
wait
futures that were created before the short time interval will trigger, even if they
are not polled during that short time interval.
§Cancel safety
This method is cancel safe.
However, the resistance against ABA problems is lost when using wait
as the
condition in a tokio::select!
loop.
sourcepub fn close(&self) -> bool
pub fn close(&self) -> bool
Close this TaskTracker
.
This allows wait
futures to complete. It does not prevent you from spawning new tasks.
Returns true
if this closed the TaskTracker
, or false
if it was already closed.
sourcepub fn reopen(&self) -> bool
pub fn reopen(&self) -> bool
Reopen this TaskTracker
.
This prevents wait
futures from completing even if the TaskTracker
is empty.
Returns true
if this reopened the TaskTracker
, or false
if it was already open.
sourcepub fn spawn<F>(&self, task: F) -> JoinHandle<F::Output>
pub fn spawn<F>(&self, task: F) -> JoinHandle<F::Output>
Spawn the provided future on the current Tokio runtime, and track it in this TaskTracker
.
This is equivalent to tokio::spawn(tracker.track_future(task))
.
sourcepub fn spawn_on<F>(&self, task: F, handle: &Handle) -> JoinHandle<F::Output>
pub fn spawn_on<F>(&self, task: F, handle: &Handle) -> JoinHandle<F::Output>
Spawn the provided future on the provided Tokio runtime, and track it in this TaskTracker
.
This is equivalent to handle.spawn(tracker.track_future(task))
.
sourcepub fn spawn_local<F>(&self, task: F) -> JoinHandle<F::Output>
pub fn spawn_local<F>(&self, task: F) -> JoinHandle<F::Output>
Spawn the provided future on the current LocalSet
, and track it in this TaskTracker
.
This is equivalent to tokio::task::spawn_local(tracker.track_future(task))
.
sourcepub fn spawn_local_on<F>(
&self,
task: F,
local_set: &LocalSet
) -> JoinHandle<F::Output>
pub fn spawn_local_on<F>( &self, task: F, local_set: &LocalSet ) -> JoinHandle<F::Output>
Spawn the provided future on the provided LocalSet
, and track it in this TaskTracker
.
This is equivalent to local_set.spawn_local(tracker.track_future(task))
.
sourcepub fn spawn_blocking<F, T>(&self, task: F) -> JoinHandle<T>
pub fn spawn_blocking<F, T>(&self, task: F) -> JoinHandle<T>
Spawn the provided blocking task on the current Tokio runtime, and track it in this TaskTracker
.
This is equivalent to tokio::task::spawn_blocking(tracker.track_future(task))
.
sourcepub fn spawn_blocking_on<F, T>(&self, task: F, handle: &Handle) -> JoinHandle<T>
pub fn spawn_blocking_on<F, T>(&self, task: F, handle: &Handle) -> JoinHandle<T>
Spawn the provided blocking task on the provided Tokio runtime, and track it in this TaskTracker
.
This is equivalent to handle.spawn_blocking(tracker.track_future(task))
.
sourcepub fn track_future<F: Future>(&self, future: F) -> TrackedFuture<F> ⓘ
pub fn track_future<F: Future>(&self, future: F) -> TrackedFuture<F> ⓘ
Track the provided future.
The returned TrackedFuture
will count as a task tracked by this collection, and will
prevent calls to wait
from returning until the task is dropped.
The task is removed from the collection when it is dropped, not when poll
returns
Poll::Ready
.
§Examples
Track a future spawned with tokio::spawn
.
use tokio_util::task::TaskTracker;
let tracker = TaskTracker::new();
tokio::spawn(tracker.track_future(my_async_fn()));
Track a future spawned on a JoinSet
.
use tokio::task::JoinSet;
use tokio_util::task::TaskTracker;
let tracker = TaskTracker::new();
let mut join_set = JoinSet::new();
join_set.spawn(tracker.track_future(my_async_fn()));
sourcepub fn token(&self) -> TaskTrackerToken
pub fn token(&self) -> TaskTrackerToken
Creates a TaskTrackerToken
representing a task tracked by this TaskTracker
.
This token is a lower-level utility than the spawn methods. Each token is considered to
correspond to a task. As long as the token exists, the TaskTracker
cannot complete.
Furthermore, the count returned by the len
method will include the tokens in the count.
Dropping the token indicates to the TaskTracker
that the task has exited.
sourcepub fn ptr_eq(left: &TaskTracker, right: &TaskTracker) -> bool
pub fn ptr_eq(left: &TaskTracker, right: &TaskTracker) -> bool
Returns true
if both task trackers correspond to the same set of tasks.
§Examples
use tokio_util::task::TaskTracker;
let tracker_1 = TaskTracker::new();
let tracker_2 = TaskTracker::new();
let tracker_1_clone = tracker_1.clone();
assert!(TaskTracker::ptr_eq(&tracker_1, &tracker_1_clone));
assert!(!TaskTracker::ptr_eq(&tracker_1, &tracker_2));
Trait Implementations§
source§impl Clone for TaskTracker
impl Clone for TaskTracker
source§fn clone(&self) -> TaskTracker
fn clone(&self) -> TaskTracker
Returns a new TaskTracker
that tracks the same set of tasks.
Since the new TaskTracker
shares the same set of tasks, changes to one set are visible in
all other clones.
§Examples
use tokio_util::task::TaskTracker;
#[tokio::main]
async fn main() {
let tracker = TaskTracker::new();
let cloned = tracker.clone();
// Spawns on `tracker` are visible in `cloned`.
tracker.spawn(std::future::pending::<()>());
assert_eq!(cloned.len(), 1);
// Spawns on `cloned` are visible in `tracker`.
cloned.spawn(std::future::pending::<()>());
assert_eq!(tracker.len(), 2);
// Calling `close` is visible to `cloned`.
tracker.close();
assert!(cloned.is_closed());
// Calling `reopen` is visible to `tracker`.
cloned.reopen();
assert!(!tracker.is_closed());
}
1.0.0 · source§fn clone_from(&mut self, source: &Self)
fn clone_from(&mut self, source: &Self)
source
. Read moresource§impl Debug for TaskTracker
impl Debug for TaskTracker
source§impl Default for TaskTracker
impl Default for TaskTracker
source§fn default() -> TaskTracker
fn default() -> TaskTracker
Creates a new TaskTracker
.
The TaskTracker
will start out as open.