Each servent keeps track of the ones it finds so it can find some of them again.
They all pass along IP addresses to each other to help each other build up their caches.
Most of them simply use text files with nothing but these numbers, since no other info is needed.

Special host cache sites also exist which do nothing but handle this task, as well as directing
servents to other host caches. This is a backup source of info when local caches run dry.

The main thing to keep in mind is that this is all decentralized and nobody is coordinating it.