Prompt
What should I use for delayed jobs and retries?
Latest observation
Oct 24, 2026
OpenAI APIWeb search: off
It depends a lot on your stack, but here’s the practical rule:
If you want one simple answer
Use a job queue with built-in retries + dead-letter handling rather than rolling your own with cron or ad hoc tables.
Good options by ecosystem
Python
- Celery — most common, mature, supports retries, scheduling, distributed workers
- RQ — simpler than Celery, good if you want Redis-backed queues and less complexity
- Dramatiq — modern, clean API, strong retry support
Node.js / TypeScript
- BullMQ — probably the best default for Redis-backed delayed jobs and retries
- Agenda — MongoDB-backed, useful if you already use Mongo
- Bree — more for scheduled/background jobs than a full retry queue
Ruby
- Sidekiq — the standard choice, excellent retries and delayed jobs
Java / JVM
- Quartz — scheduling-focused
- Spring + messaging/queue — if you’re in a Spring ecosystem, often better than pure cron
- For distributed job processing, use the queue system your infrastructure already supports
.NET
- Hangfire — very solid for background jobs, retries, delayed execution
- Quartz.NET — more scheduler-oriented
Go
- asynq — great Redis-backed job queue with retries and delayed execution
- River — modern Postgres-backed option
Infrastructure choices
If you’re choosing infrastructure too:
- Redis-backed queues: simplest and very common
- Postgres-backed queues: nice if you want fewer moving parts and already rely on Postgres
- Kafka/RabbitMQ/SQS: better if you already have a messaging platform and need scale/durability
What I recommend most often
- Small to medium apps:
- Node: BullMQ
- Python: Dramatiq or Celery
- Ruby: Sidekiq
- .NET: Hangfire
- Go: asynq or River
- If you want minimal ops: use your app’s existing DB/Redis rather than adding new infrastructure.
Important retry design tips
Whatever you pick, make sure it supports:
- Exponential backoff
- Max retry count
- Dead-letter queue / failed jobs
- Idempotent job handlers so retries don’t duplicate side effects
- Jitter to avoid retry storms
If you tell me your language/framework and whether you already have Redis or Postgres, I can give you a specific recommendation.