Workers

A worker pops payloads, restores jobs, runs middleware + handle(), and either acks success, releases for retry, or logs a failed job.

./grind queue:work
./grind queue:work --connection=redis --queue=default
./grind queue:work --once
./grind queue:work --stop-when-empty --max=100
./grind queue:work --daemon --sleep=3

Flags

Flag	Meaning
--connection=	Queue connection name
--queue=	Queue name on that connection
--max=	Stop after N successful process attempts
--once	Process at most one job
--stop-when-empty	Exit when the queue has no available jobs (default unless --daemon / --sleep)
--daemon	Keep polling when empty
--sleep=	Seconds to sleep when empty in daemon mode
--timeout=	Per-job timeout in seconds (0 disables)
--memory=	Stop if used heap reaches this many MB (0 disables)
--tries=	Override job try count
--backoff=	Override backoff seconds on retry

Programmatically:

Worker(
    manager,
    connectionName = "array",
    queueName = "default",
    options = WorkerOptions(once = true, timeoutSeconds = 30),
).run()

steam

Boot the app with ./grind steam and call Bus / Worker while developing. Prefer a dedicated queue:work process for real workloads.

Retries

On exception, if attempts remain below tries, the worker releases the payload with backoffSeconds (via availableAt). Otherwise the failure is recorded — see Failed jobs.