Using service discovery - How connections to service discovery agents are handled? Can we limit it?

I am using this lib with service discovery based on Consul. I have recentry hit Consul max connection limit per IP.
It seems that every gRPC thread has its own connection to Consul. Is it how it works? Is there an option to share connections and lower number of concurrent connections to service discovery agent?

Or maybe it is entirely spring-related topic?

No files or tests are named. Start by tracing how service-discovery connections are handled for each gRPC thread and determine whether connections can be shared or limited; done means documenting the current behavior and agreeing on the scope of any change.

Written by the indexing model from the issue text.