| goncordia | River | Asynq | Machinery | |
|---|---|---|---|---|
| Backends | Postgres, MySQL, SQLite, MongoDB, Redis, Cassandra, ClickHouse, DynamoDB, Firestore, memory | PostgreSQL only | Redis only | Redis, RabbitMQ, SQS, MongoDB |
| Transactional insert (outbox pattern) | ✅ where backend supports tx | ✅ Postgres only | ❌ | ❌ |
Generic Driver[TTx] interface | ✅ | ✅ (Postgres only) | — | — |
| Scheduled / cron jobs | ✅ | ✅ | ✅ | ✅ |
| Unique jobs | ✅ | ✅ | ✅ | ❌ |
| Priority queues | ✅ | ✅ | ✅ | ❌ |
| Push notifications | ✅ LISTEN/NOTIFY, Change Streams, Pub/Sub | ✅ LISTEN/NOTIFY | ✅ Pub/Sub | ❌ polling |
| Web UI | ✅ embeddable | ✅ (River UI, Pro) | ✅ (asynqmon) | ❌ |
| Await a job’s typed result | ✅ future.Func.Call / Await | ❌ | stores results, no await | ✅ AsyncResult.Get |
| Unit tests without Docker | ✅ in-memory driver + gontest | needs Postgres | needs Redis | needs a broker |
| Workflow primitives (chains, chords) | ❌ | ❌ | ❌ | ✅ |
Choose goncordia when you want the transactional outbox pattern and don’t want to introduce Redis or a dedicated Postgres instance — it works with whatever database your application already uses. It also fits when a caller needs the value a background job returns (future.Func.Call), or when you want unit tests of enqueue and worker code without Docker.
Choose River if you’re on Postgres and want a mature, battle-tested library with a polished UI and a professional support tier.
Choose Asynq if Redis is already in your stack, you don’t need transactional inserts, and you want a built-in web dashboard.
Choose Machinery if you need Celery-style workflow primitives (chains, chords, groups) across RabbitMQ or SQS.