Struct sc_service::TaskManager

pub struct TaskManager { /* fields omitted */ }

Helper struct to manage background/async tasks in Service.

Implementations

impl TaskManager

pub fn spawn_handle(&self) -> SpawnTaskHandle

Get a handle for spawning tasks.

pub fn spawn_essential_handle(&self) -> SpawnEssentialTaskHandle

Get a handle for spawning essential tasks.

pub fn clean_shutdown(self) -> Pin<Box<dyn Future<Output = ()> + Send>>

Send the signal for termination, prevent new tasks to be created, await for all the existing tasks to be finished and drop the object. You can consider this as an async drop.

It’s always better to call and await this function before exiting the process as background tasks may be running in the background. If the process exit and the background tasks are not cancelled, this will lead to objects not getting dropped properly.

This is an issue in some cases as some of our dependencies do require that we drop all the objects properly otherwise it triggers a SIGABRT on exit.

pub fn future<'a>(
    &'a mut self
) -> Pin<Box<dyn Future<Output = Result<(), Error>> + Send + 'a>>

Return a future that will end with success if the signal to terminate was sent (self.terminate()) or with an error if an essential task fails.

Warning

This function will not wait until the end of the remaining task. You must call and await clean_shutdown() after this.

pub fn terminate(&mut self)

Signal to terminate all the running tasks.

pub fn keep_alive<T: 'static + Send + Sync>(&mut self, to_keep_alive: T)

Set what the task manager should keep alive, can be called multiple times.

pub fn add_child(&mut self, child: TaskManager)

Register another TaskManager to terminate and gracefully shutdown when the parent terminates and gracefully shutdown. Also ends the parent future() if a child’s essential task fails. (But don’t end the parent if a child’s normal task fails.)