Many scheduled tasks -- cleanups, report generation, health checks -- run as a single unit of work on one machine. The standalone execution model handles this by dispatching each job instance to exactly one worker, keeping scheduling simple and resource usage predictable.
How it works
When a job triggers, the scheduler selects one worker at random from the available worker pool and dispatches the job instance to it. No other workers participate in that execution.
One trigger, one worker, one execution.
When to use standalone
Use standalone for tasks that complete on a single machine without parallel processing:
| Scenario | Example |
|---|---|
| Scheduled cleanups | Delete expired records, rotate logs, purge temporary files |
| Data aggregation | Compute daily summaries, generate reports, refresh materialized views |
| Notifications | Send batch emails, push alerts, trigger webhook callbacks |
| Health checks | Poll an external API or service endpoint at regular intervals |
Supported job types
Standalone supports all job types available in MSE.