Just to clear it our, the behavior of a replicated backup when it cannot contact its live server. Should it become 'live' after a time-out or should it just shutdown?
The reason I am going about this is that we have old tests that attempt to (1) stop the backup, (2) crash the live, and (3) restart the backup. I don't think this makes sense for the remote replication case, as the journal may not be up to date.
FWIW, I am changing the current code to at least closing the backup server in case of initial connection (instead of simply leaving everything half open). It is -at the very least- more consistent.
It should be configurable based on retries.
If the user configured the connection with infinite-retries.. .then it will loop on getConnection at the connector forever.
If the user configured a max.. then it will give up after some time, and on that case if you the connection is returned null, you should just call server.stop()