Why does autoRun and onInit only trigger after the first HTTP request in Next.js production?

[shestakov2103]

Hi

I've noticed a significant difference in behavior between the CLI runner and the Web server in Payload 3.0.

If I run pnpm payload jobs:run --handle-schedules, everything starts immediately and works perfectly.

If I rely on autoRun inside the main Next.js app (npm run start), the worker and onInit hooks remain idle until I manually open the admin panel in a browser.

I understand that Next.js uses lazy initialization, but for background workers, this is a major issue. We expect the worker to start as soon as the Node process starts, not when the first user logs in.

Is there a way to force "Eager Initialization" for Payload within the Next.js process without resorting to running a separate CLI worker? Or is a separate CLI process now the officially recommended way for production background tasks?

[andershermansen]

We solved it with a health check so that it is initialized.

[shestakov2103]

Thanks! Any issues with the process going back to sleep if there are no requests for a while?

[andershermansen]

No, never happened to me.