Why brod_client:get_connection/3 restarts a dead connection instead of checking for other active connections?

[alsbaz]

In the state, I can see two connections.

[{conn, {"hostname1",port}, {dead_since,{1753,269753,411438},{shutdown,ssl_closed}}}, {conn,{<<"hostname2">>, port},<11258.1602.0>}]

When a regular process (a health check for our kafka connection) calls brod_client:get_connection/3, brod restarts the connection to the 1st host, instead of finding the second one. This results in an {shutdown, ssl_closed} error every 10 minutes, as the kafka disconnects the idle (only used for the health check) connection, which shortly after reconnects due to the health check. This produces another log when it reconnects.

Shouldn't the get_connection/3 function check the live connection first, without restarting a previous one, if it is able to?

Or maybe I'm misunderstanding the point of this function, but wanted to check, maybe someone else has ran into this issue.