z4j-celery 1.3.0

z4j Celery queue engine adapter (Apache 2.0)

z4j-celery

License: Apache 2.0 Status: v1.0.0 - first public release. The flagship queue-engine adapter.

The Celery queue-engine adapter for z4j. Captures every task lifecycle event via Celery's signal system, executes control actions (retry, cancel, bulk retry, purge, pool restart) against your live Celery app, and discovers tasks via a five-layer pipeline that handles signal registration, broker events, and static AST scans.

Install

pip install z4j-celery z4j-celerybeat

Pair with a framework adapter:

pip install z4j-django z4j-celery z4j-celerybeat      # Django
pip install z4j-flask  z4j-celery z4j-celerybeat      # Flask
pip install z4j-fastapi z4j-celery z4j-celerybeat     # FastAPI
pip install z4j-bare  z4j-celery z4j-celerybeat       # Bare worker

What it ships on day 1

Capability	Status	Notes
Event capture	✅	Dual-capture: Celery signals (task_received, task_succeeded, task_failed, task_retried, task_revoked) + broker events for multi-worker deployments
Discovery	✅	Five-layer pipeline: runtime celery_app.tasks, static tasks.py AST scan, signal registration, merged task registry
submit_task / retry_task / cancel_task	✅	Direct against the Celery app
bulk_retry	✅	Filter-driven; re-enqueues matching tasks
purge_queue	✅	With confirm-token + Z4J_PURGE_THRESHOLD guard
requeue_dead_letter	✅	From the configured DLX
restart_worker	✅	Broadcast pool_restart - zero task loss, child processes drain in-flight tasks before respawn
pool_grow / pool_shrink	✅	Via Celery's control API
rate_limit	✅	Broker-side via Celery's control channel - the only engine where this works correctly
reconcile_task	✅	Via AsyncResult against the result backend

This is the widest feature coverage of any adapter. Celery's rich remote- control surface lets z4j offer capabilities no other engine can match (pool restart with zero task loss, broker-side rate limiting).

Reliability

• No exception from the adapter propagates to Celery signal handlers.
• Events buffer to an in-memory ring; a background thread flushes to the brain. Signal handlers never do network I/O.
• When the brain is unreachable, events buffer locally (via z4j-bare's SQLite buffer). Your Celery workers never slow down or block.

Scheduler pairing

Use z4j-celerybeat for Celery Beat / django-celery-beat schedules on the Schedules page.

Documentation

• Celery engine guide
• Architecture
• Adapter protocol

License

Apache 2.0 - see LICENSE. Your application is never AGPL-tainted by importing z4j_celery.

Links

• Homepage: https://z4j.com
• Documentation: https://z4j.dev
• Issues: https://github.com/z4jdev/z4j-celery/issues
• Changelog: CHANGELOG.md
• Security: [e-mail] (see SECURITY.md)