if all initial servers become unreachable, operations being failing #74
Description
First, I want to stress two things. I use this client with AWS Keyspaces, which is "Cassandra", but does have some unusual behaviors compared to a regular Cassandra cluster.
Second, I have not really been able to use this client a lot, because the lack of sigv4 signing is a pretty big barrier.
Here's what happens. Keyspaces seems to change out the visible servers regularly. And it can happen that all of the initial servers end up becoming unreachable. Should that happen, the underlying C++ driver used to become unhappy. This was a fairly common problem with Keyspaces, and also used to cause the Go driver to fail.