Xandra uses DBConnection under the hood so you can call DBConnection.close/3 like so:

DBConnection.close(conn, prepared)

A better approach is to have the Xandra process managed by the supervision tree. So you can leverage the pool for better use of resources, automatic reconnection, and automatic clean up when the application finishes.

To do that add to your workers list at the application entrypoint:

 workers = [
    ...,
   {Xandra, name: MyXandra, nodes: [Application.get_env(:diamond, :env_vars)[:cass_host]],
      authentication: {Xandra.Authenticator.Password, 
        [
          username: Application.get_env(:diamond, :env_vars)[:cass_host], 
          password: Application.get_env(:diamond, :env_vars)[:cass_pass]
        ]},
      pool: DBConnection.Poolboy,
      pool_size: 10}
     .....,
  ]

Then to use this process to execute the queries you will reference by the name you gave to it, in the example it was MyXandra:

def find_by_id(audit_log_id) do
    
    stm = "SELECT * FROM audit_log_keyspace.rules_engine_audit_logs WHERE id=:id;"

    prepared = MyXandra |> Xandra.prepare!(stm)
   
    MyXandra
    |> Xandra.execute!(prepared, %{ "id" => audit_log_id })
    |> Enum.fetch!(0)
end