Struct Supervisor 

pub struct Supervisor { /* private fields */ }

The central supervisor.

Manages the lifecycle of all TradingService implementations using TaskTracker for structured concurrency and CancellationToken for graceful shutdown propagation.

Memory safety

Unlike JoinSet, TaskTracker does not accumulate task return values. Completed task memory is reclaimed immediately, making this safe for long-running processes that may restart services hundreds of times over weeks of operation.

Example

let supervisor = Supervisor::new(SupervisorConfig::default());
supervisor.spawn_service(Box::new(MyService));
supervisor.run_until_shutdown().await?;

Implementations

impl Supervisor

pub fn new(config: SupervisorConfig) -> Supervisor

Create a new supervisor with the given configuration.

pub fn with_defaults() -> Supervisor

Create a new supervisor with default configuration.

pub fn cancel_token(&self) -> &CancellationToken

Get a reference to the supervisor’s cancellation token.

Useful for external code that needs to observe or trigger shutdown.

pub fn metrics(&self) -> &Arc<SupervisorMetrics>

Get a reference to the supervisor’s metrics.

pub async fn lifecycle_snapshots(&self) -> Vec<ServiceLifecycleSnapshot>

Snapshot all service lifecycles.

pub async fn service_lifecycle( &self, name: &str, ) -> Option<ServiceLifecycleSnapshot>

Lifecycle snapshot for a specific service by name.

pub async fn service_count(&self) -> usize

Number of services currently tracked (alive + terminated).

pub fn trigger_shutdown(&self)

Trigger a graceful shutdown of all managed services.

pub fn is_shutting_down(&self) -> bool

Returns true if the supervisor’s shutdown has been triggered.

pub fn spawn_service(&self, service: Box<dyn TradingService>)

Spawn a service with default options.

pub fn spawn_service_with_options( &self, service: Box<dyn TradingService>, options: SpawnOptions, )

Spawn a service with custom per-service options.

pub async fn wait_for_drain(&self)

Close the tracker and wait for all tasks to complete, with timeout.

Call after triggering shutdown (or after run_until_shutdown returns) to ensure all tasks have drained.

pub async fn run_until_shutdown(&self) -> Result<(), Error>

Run the supervisor until a shutdown signal is received.

If install_signal_handler is true (the default), listens for Ctrl-C / SIGTERM and triggers shutdown. Returns after all services have drained (or the shutdown timeout has elapsed).