I was able to reproduce your reconnect issue with the ReliableClientSession. It appears that prior to 1.0.0, the reliable client would always reconnect, regardless of error. For resynchronisation errors this is required. For otherwise normal operation, this isn't. Since we can probably assume that a client exit would be system initiated and therefore not likely to elicit a reconnect sequence, we should restore the 0.10.0 behaviour.
Can you raise a ticket in Jira and we'll fix this in 1.0.2.