Threadmill vs JobRunr
JobRunr is the best-known job library for Java, with an open-source edition and a commercial Pro edition. Threadmill is fully open source. This page compares them feature by feature so you can choose.
JobRunr facts are from its public pricing and documentation pages, checked on . If something here is out of date, open an issue and we will fix it.
Short version
- Choose Threadmill if you want priority queues, transactional enqueue, mutexes, per-key concurrency and timeouts without a commercial licence, you run PostgreSQL, Oracle or Redis, and you are on a current Java and Spring Boot.
- Choose JobRunr OSS if you need Java 8 to 21, Quarkus or Micronaut, or a database Threadmill does not support, and the open-source feature set is enough.
- Choose JobRunr Pro if you need batches, rate limiting, dashboard single sign-on or a multi-cluster dashboard today.
Checked on . Sources: JobRunr pricing, feature comparison, storage documentation, recurring jobs documentation, JobRunr documentation.
Feature comparison
| Threadmill | JobRunr OSS | JobRunr Pro | |
|---|---|---|---|
| License | Apache 2.0 | LGPL 3.0 | Commercial |
| Price | Free | Free | From €9,000 per year per production cluster |
| Fire-and-forget, delayed and recurring jobs | Yes | Yes | Yes |
| Automatic retries | Yes | Yes | Yes |
| Priority queues | Yes | No | Yes |
| Transactional enqueue | Yes | No | Yes |
| Mutexes | Yes | No | Yes |
| Per-key concurrency limits | Yes | No | Yes |
| Job timeouts | Yes | No | Yes |
| Run jobs on tagged servers | Yes | No | Yes |
| Missed recurring runs after downtime | Configurable per task | Skipped | Yes |
| Recurring jobs | No fixed limit | About 100 | 5,000, or unlimited on Enterprise |
| New-job pickup | Wake signal, 500 ms poll fallback | 5-second minimum poll | Real-time scheduling |
| Dashboard | Secured by default, redaction, audit hooks | Basic | Search, filters; SSO on Enterprise |
| Job chains | Parent and child steps | No | Yes |
| Atomic batches | Bulk enqueue only | No | Yes |
| Rate limiting | No | No | Yes |
| Durable steps inside a job | No | Yes | Yes |
| Redis store | Yes | Removed in v8 | Removed in v8 |
| Frameworks | Spring Boot | Spring Boot, Quarkus, Micronaut | Spring Boot, Quarkus, Micronaut |
| Commercial support | Available | No | Included |
Sources: JobRunr pricing and feature comparison, storage documentation, recurring jobs documentation. Checked on .
How the two differ in design
Jobs as data, not as lambdas. JobRunr captures a lambda and stores the method call. Threadmill stores a typed payload record and the name of its handler. The payload is an explicit contract you can version and review.
Redis as a first-class store. Threadmill supports Redis in standalone, Sentinel and Cluster setups, tested against the same contract as PostgreSQL.
One engine for schedules and queues. Lanes give each queue its own capacity, so a flood of bulk work cannot starve latency-sensitive jobs.
Where JobRunr is ahead
JobRunr supports more Java versions, more frameworks and more databases. It has six years of production use and a larger community. Its runStepOnce API checkpoints steps inside a job, and Pro offers batches, rate limiting and a richer dashboard. If you need those today, JobRunr is the right choice.
Migrating
Most JobRunr jobs map directly: the method body becomes a handler, its arguments become a payload record, and @Recurring stays @Recurring.
Checked on . Sources: JobRunr pricing, feature comparison, storage documentation, recurring jobs documentation, JobRunr documentation.