I started multiple processes in the production environment over time 'PROMETHEUS_ MULTIPROC_ ‘ There will be more and more files under dir. How can I establish cleaning rules and the volume of responses returned by the 'metrics' endpoint will eventually exceed the connection time. Any help would be greatly appreciated.