Inquiry about setting up a Milvus replication cluster using Milvus-CDC

[ian-lss]

Hello,

I am trying to set up a replication milvus cluster using Milvus-CDC.
I have confirmed that incremental data replication works well as stated in the documentation.

As of now, Milvus-CDC is designed to synchronize only incremental data.

To set up a replication database similar to MySQL,
I need to back up the previous data using Milvus-backup, restore this backup data to the replica Milvus cluster, and then use the recorded position from the backup to register Milvus-CDC.
However, the latest version of Milvus-backup does not record this position.
Should I use the [flush segments] information in the milvus-backup log?

How can I obtain this position(checkpoint?)?

[xiaofan-luan]

@SimFG could help on this

[SimFG]

@ian-lss
Currently, the linkage between backup tool and cdc tool is under development, and it is expected to be completed in mid-July.

If you need to use it now, it cannot support the given position. The current usage mode: stop writing data to milvus, then use the backup tool to back up the existing data, restore this part of the data to another milvus, and then start cdc synchronization without transferring the point. After cdc works normally, milvus can be used normally.

In subsequent versions, the backup tool supports returning the position, and the cdc tool supports passing the position. At this time, the stop time of source milvus only needs to get the backup tool return position to work normally.

[ian-lss]

@SimFG @xiaofan-luan

I really appreciate both of your responses, and I will be looking forward to mid-July.

Thank you.