We have found HVR V6 (LDP) is connecting to Aurora MySQL with server_id=0 and this is causing Failover of Aurora MySQL instances frequently and root cause of the issue is a timing related condition when a binlog dump thread is connecting to a writer instance while another is disconnecting

in order to resolve these issues, we have two options.

1. Connect each binlog consumer once and keep that connection alive while the consumer requires binlog data
2. using unique non-zero replica ids for each HVR connection to the same Aurora cluster

Do we have any plan to implement any of the options in future release to overcome impact on source systems?

Thanks and Regards,

Krushikesh