Custom Entities TTL Clean Up on Cluster

When the ttl option is used to save a custom entity, there is a cronjob-like using to clean up those records and the ones inserted in the ttls table. I have been trying to find information on how this behaves on a kong cluster with multiple nodes but I haven’t.

Is there any documentation regarding this topic? This link just talks about the admin objects and how the data is cached. But I haven’t seen something describing how that job (do_clean_ttl) in posgres.lua will handle race conditions between kong nodes?