I'm trying to work myself into a situation where I can have a snapshot taken by Debezium without causing too much load of my normal consumers.
My idea was to have a second instance of the mysql-connector create a snapshot pushing into a different set of topics than the normal connector, so I have two sets of consumers processing data at the same time. This way changes are processed by the normal consumers without the delay caused by the data from the snapshot.
Fox context, processing a full snapshot takes our system nearly 3 days (we have a lot of data...)
So far I can get the second instance of Debezium to run within my Strimzi connect cluster, and I also got it to run in a different cluster. But with either solution Im finding that Debezium copies the binlog offsets from the instance I already had running, and doesn't start a new snapshot.
Id like suggestions on how to proceed, how can I, in parallel to my normal processes, have a full snapshot be taken by debezium?