Identify the actual queue adapter, Rails version and production worker command. Do not assume a new Rails default is also the configuration of an older application.

1. Configure the adapter and its required database, broker or other service. Apply any required schema setup through a reviewed deployment.

2. Start the supported worker or dispatcher through the approved supervisor. Give it the same relevant production configuration as the web application.

3. Submit a harmless test job and verify the real result. Confirm it is not being held only in process memory when durable execution is required.

4. Set queue selection, concurrency, retry and failure behavior according to the adapter. Design externally visible actions to tolerate redelivery.

5. Restart workers as part of deployment and monitor queued age and failed work. Document how to retry or discard a job after reviewing its effect.

Confirm a job completes after the submitting browser disconnects and that workers return after a controlled restart. Test delayed work as well as immediate jobs.

If jobs remain pending, check the adapter, queue name and worker environment. If a job failed after an external action, reconcile the result before retrying.

• Running Rails database preparation safely
• Preserving Active Storage uploads during deployment
• Using a queue for slow exports and notifications
• Building and serving Rails production assets

• Rails Active Job