Struct WorkerLoop 

pub struct WorkerLoop { /* private fields */ }

DB-backed queue worker.

Runs a reaper→claim→spawn cycle until signalled to stop. Panic isolation (D-11) ensures a panicking job handler never kills the loop. Graceful shutdown (D-10) drains in-flight jobs and resets claimed-but-unstarted rows.

Implementations

impl WorkerLoop

pub fn new(config: WorkerConfig) -> Self

Create a new worker loop.

A random UUID is generated as the worker ID. The worker reads the DB connection from Queue::connection() — no connection argument needed.

pub fn with_tenant_scope(self, provider: Arc<dyn TenantScopeProvider>) -> Self

Inject a tenant scope provider for tenant-aware job execution.

When set, jobs with a tenant_id are executed inside a tenant context scope. Jobs without a tenant_id or workers without a provider run in the default (system) scope.

pub fn register<J>(&mut self)

where J: Job + DeserializeOwned + 'static,

Register a job handler.

The handler closure deserializes the job from JSON, calls handle(), and also captures the per-job retry_delay(attempt) for use on failure.

Example

worker_loop.register::<SendEmailJob>();

pub fn from_registry(config: WorkerConfig) -> Self

Build a WorkerLoop and apply all job types registered via [Queue::register].

This is the entry point used by the framework’s server boot path to create the auto-started worker when at least one job type has been registered.

pub fn shutdown(&self)

Signal the worker loop to shut down gracefully.

Sets the same AtomicBool that the SIGTERM/Ctrl-C handler sets, so programmatic and signal-based shutdown share one path.

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

Run the worker loop until shutdown.

Loop behaviour:

1. Check shutdown flag — if set, drain in-flight jobs and requeue.
2. For each queue: run reaper, then attempt claim.
   • If a job is claimed, spawn it and loop immediately (no idle sleep).
3. If no jobs were found across all queues, sleep config.sleep_duration.