db-scheduler

Uses a single database table with optimistic locking or select-for-update to ensure tasks are executed by only one instance in a cluster, preventing duplicates in distributed environments.

Benchmarked to handle 2,000–10,000 executions per second with configurable polling strategies like lock-and-fetch, making it suitable for high-load scenarios as noted in user testimonials.

Depends only on SLF4J, keeping it lightweight and easy to embed into existing Java applications without bloating the dependency graph.

Supports static and dynamic recurring tasks, one-time tasks, and custom schedules via cron, fixed delay, or daily intervals, with persistent task data for multi-tenant use cases.