arq

v1.0 (new)

Async Redis queue for FastAPI-era Python.

arq has no native re-enqueue-by-id primitive, so retry_task is polyfilled brain-side via submit_task using captured args. Event capture chains onto arq's WorkerSettings on_job_start / on_job_end hooks.

Engine docs PyPI: z4j-arq

Install

bash

pip install z4j-arq

Native actions

submit_task
cancel_task (Job.abort)

Brain-polyfilled actions

retry_task
bulk_retry
requeue_dead_letter

The brain fetches the original args from its tasks table and lowers to submit_task transparently. The dashboard shows the same button regardless.

Not supported

purge_queue
restart_worker (graceful self-exit)
rate_limit

Engine-level limitations, not a roadmap gap. These actions require primitives the engine does not expose.

Wiring

How z4j plugs in

Event capture

attach_to_worker_settings() chains onto on_job_start / on_job_end. Optional; lifecycle still works without.

Reconciliation

arq.jobs.Job.status() + result_info(): clean deferred/queued/in_progress/complete mapping.

Paired scheduler

arq cron_jobs

Read-only surface for arq's WorkerSettings.cron_jobs.

z4j-arqcron

arq cron jobs are statically configured in WorkerSettings.cron_jobs. Same read-only posture as Huey. This adapter exists for visibility, last-fire tracking, and per-job drill-down.

Learn more

More engines