Question: Separate Instances for Scheduling and Executing Jobs

[JakeCG]

Hi,

Is it possible to configure two separate instances of db-scheduler in a Spring context - one dedicated to scheduling jobs and the other dedicated to executing them?

Ideally, this separation would be controlled via environment variables, allowing us to disable job execution on instances that serve end-user traffic. This would help prevent performance degradation from memory-intensive jobs on those instances.

Is this setup supported or recommended? If so, are there any best practices for implementing it?

Thanks!

[piotrooo]

Hey @JakeCG 👋
I'm a bit curious about your use case. Why am I asking? Because I have exactly the same issue, and I’ve been handling it by manually creating two separate tasks. One is for scheduling — it's a recurring task, and each application (pod) has exactly one listener to perform the scheduling. The second task is used for one-time jobs, which are scheduled by the first task. Each of these tasks has its own table for scheduled executions.

Right now, we’re considering migrating to a more "standard" setup with just one table. To make sure the scheduler isn’t starved by other jobs, we’re planning to introduce task priorities (added in v15.0.0). The scheduler task would be given the highest priority.

This is just a theoretical approach for now — we haven’t tested it yet.
What do you think about this idea?

Maybe @kagkarlsson has some thoughts?