arq

Package: z4j-arq - arq 0.25+.

Install

pip install z4j-arq

What it captures

arq doesn’t expose signals. z4j uses worker on_job_start / on_job_end hooks and wraps the enqueue_job method.

Source	z4j event
ArqRedis.enqueue_job	task_sent
on_job_start hook	task_started
on_job_end (success=True)	task_succeeded
on_job_end (success=False)	task_failed
retries (tried > 1)	task_retry (synthetic)

Wiring

arq’s config is a WorkerSettings class. z4j injects hooks:

from z4j_arq import z4j_worker_settings

class WorkerSettings(z4j_worker_settings(MyBaseSettings)):
    functions = [my_task, ...]
    # z4j's on_startup / on_shutdown / on_job_start / on_job_end are merged

Or manual:

from z4j_arq import on_startup, on_shutdown, on_job_start, on_job_end

class WorkerSettings:
    functions = [...]
    on_startup = on_startup
    on_shutdown = on_shutdown
    on_job_start = on_job_start
    on_job_end = on_job_end

Actions

Verb	How
retry	polyfill - arq_redis.enqueue_job(function_name, *args) with original payload
cancel	arq_redis.abort_job(job_id) - works pre-start; mid-run relies on your function’s cancellation cooperation
purge_queue	arq_redis.flushdb() (scoped to arq prefix)

Cron jobs

arq’s cron jobs are defined in WorkerSettings.cron_jobs. Same as Huey - code-only discovery, read-only in v1.0. See scheduler: arq-cron.

Caveats

• No chord/group.
• Max job size limited by Redis payload; large pickled args may be truncated in events (see redaction).
• Worker pool size appears as metadata.max_jobs in the agent drawer.

Config

# FastAPI env-based
Z4J_ARQ_REDIS = "redis://localhost:6379/0"

Or pass the ArqRedis instance explicitly via the adapter’s redis= kwarg if you need it.