PHP-FPM workers can stay alive across many requests, while background queue workers may run for hours. Both can continue using previously loaded resources after a change. Replacing them is often necessary when applying PHP security updates or configuration changes.
The correct operation depends on how PHP is hosted. A system service may be restarted or reloaded, a hosting panel may manage a pool, and a containerized site may require a new deployment. A worker pool restart is not the same thing as restarting nginx, and a queue worker may belong to another service altogether.
Some reloads replace workers gracefully while allowing in-flight requests to finish. Other restart procedures briefly interrupt service. The actual behavior depends on the process manager and its settings.
Use the deployment method supported by your environment, check error logs and health checks, and verify that no old PHP workers still handle traffic after the change.