The MySQL connector should be able to connect to a MySQL HA cluster and when needed reestablish a connection to the cluster and continue reading where it left off. For example, if the server that it is using is removed from the cluster, the connect should connect to a different server.

This likely requires adding support for reading the GTID, recording the GTID in the offsets, and when (re)connecting to find the binlog position using the last known GTID. (Note that GTIDs are only available in recent MySQL releases, which might make this feature useable only on those recent versions.)