steps:
1. start brand new connector with 100 tables
2. table 99 fails during the initial snapshot with NOT_ENOUGH_REPLICAS error
3. delete the pod (don't touch the topics)
4. observe that new pod starts and snapshotting restarts from table 1,2,3...etc
5. in between log messages of exporting x rows.... will also see log messages of "Couldn't commit processed log positions with the source database due to a concurrent connector shutdown or restart"

I wonder if this is related to similar sounding Postgres issues (https://github.com/debezium/debezium/pull/3393 , https://github.com/debezium/debezium/pull/2660/files, https://github.com/debezium/debezium/pull/3620/files)

logs attached

as per https://issues.redhat.com/browse/DBZ-5750 i do notice some rows not going to kafka