Struct WorkerPool 

pub struct WorkerPool { /* private fields */ }

Atomic worker/task coordinator. All clones share one revisioned state lock.

Implementations

impl WorkerPool

pub fn new() -> Self

pub fn with_limits(limits: DistributedLimits) -> Result<Self, DistributedError>

pub fn snapshot(&self) -> Result<StateSnapshot, DistributedError>

pub fn task_queue(&self) -> TaskQueue

Return a cloneable task facade guaranteed to share this pool’s state.

pub fn register_worker( &self, expected_revision: u64, now_secs: u64, spec: WorkerSpec, ) -> Result<Transition<WorkerNode>, DistributedError>

Register a never-before-seen worker. Duplicate IDs never overwrite.

pub fn reactivate_worker( &self, expected_revision: u64, now_secs: u64, expected_generation: u64, spec: WorkerSpec, ) -> Result<Transition<WorkerNode>, DistributedError>

Reactivate an offline worker under a new generation fence.

pub fn update_worker( &self, expected_revision: u64, now_secs: u64, worker_id: &str, worker_generation: u64, observation: WorkerObservation, ) -> Result<Transition<WorkerNode>, DistributedError>

Update heartbeat, eligibility state, and integer resource observations.

pub fn get_available_worker( &self, now_secs: u64, ) -> Result<Option<WorkerNode>, DistributedError>

Return the deterministically selected eligible worker at logical time.

pub fn assign_task( &self, expected_revision: u64, now_secs: u64, task_id: &str, worker_id: &str, lease_ttl_secs: u64, ) -> Result<Transition<TaskLease>, DistributedError>

Atomically assign a specific queued task to a specific worker.

pub fn assign_next( &self, expected_revision: u64, now_secs: u64, worker_id: &str, lease_ttl_secs: u64, ) -> Result<Transition<TaskLease>, DistributedError>

Atomically assign the highest-priority FIFO task to a specific worker.

pub fn assign_next_available( &self, expected_revision: u64, now_secs: u64, lease_ttl_secs: u64, ) -> Result<Transition<TaskLease>, DistributedError>

Atomically choose both task and worker. Equal worker keys choose the lexicographically smallest worker ID.

pub fn start_task( &self, expected_revision: u64, now_secs: u64, lease: &TaskLease, ) -> Result<Transition<StartOutcome>, DistributedError>

Transition a lease from Leased to Running. Exact replay is a no-op.

pub fn complete_task( &self, expected_revision: u64, now_secs: u64, lease: &TaskLease, ) -> Result<Transition<CompletionOutcome>, DistributedError>

Complete only the current lease. Exact replay returns the same receipt without changing revision or worker counters.

pub fn cancel_task( &self, expected_revision: u64, now_secs: u64, ownership: &TaskOwnership, ) -> Result<Transition<CancellationOutcome>, DistributedError>

Cancel queued or leased work only with its exact ownership fence. Exact replay is a no-op; a different token fails closed.

pub fn fail_task( &self, expected_revision: u64, now_secs: u64, lease: &TaskLease, ) -> Result<Transition<FailureOutcome>, DistributedError>

Fail the current attempt. Retry policy is fixed in DistributedLimits.

pub fn deregister_worker( &self, expected_revision: u64, now_secs: u64, worker_id: &str, worker_generation: u64, ) -> Result<Transition<RecoverySummary>, DistributedError>

Mark a worker offline, fence its generation, and recover every lease it owned in original lease order under the fixed retry policy.

pub fn recover_expired_leases( &self, expected_revision: u64, now_secs: u64, ) -> Result<Transition<RecoverySummary>, DistributedError>

Recover every lease at or beyond its deadline (now >= expires_at) under the fixed retry policy.

pub fn prune_dead_workers( &self, expected_revision: u64, now_secs: u64, ) -> Result<Transition<RecoverySummary>, DistributedError>

Mark heartbeat-stale workers offline and recover their leases. Stale workers are already ineligible for assignment before this command runs.

pub fn expire_old_tasks( &self, expected_revision: u64, now_secs: u64, task_ttl_secs: u64, ) -> Result<Transition<usize>, DistributedError>

Terminally expire every active task whose age reaches task_ttl_secs.

pub fn get_workers(&self) -> Result<Vec<WorkerNode>, DistributedError>

Return workers in stable worker-ID order.

pub fn get_worker( &self, worker_id: &str, ) -> Result<Option<WorkerNode>, DistributedError>

pub fn check_invariants(&self) -> Result<(), DistributedError>

Actively verify all cross-record invariants.

Trait Implementations

impl Clone for WorkerPool

fn clone(&self) -> WorkerPool

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Default for WorkerPool

fn default() -> Self

Returns the “default value” for a type.