Question about symbol server configs

We’re trying to set up a symbol server service for our on-premise sentry cluster.

In order to properly config the symbol server, can you tell me what are the purposes for symbol_dir and the s3 bucket respectively?

The symbol server README.md says symbol_dir is /path/to/symbol/directory, and the Sentry Debug Symbols webpage says the s3 bucket is “where symbols are synchronized from”. So it sounds like both are used to save symbols. I’m a little confused what differences are between the two (the symbol_dir and the s3 bucket).

Also deployment wise, can we put multiple symbol servers behind an ELB and have them autoscale, or is that not the right way to operate them?

Thanks!