If I use MM2, under what circumstances will the message offsets in the source and target clusters start differing. It seems that offsets are basically sequence numbers, and since the ordering is maintained, the offsets should be the same in both, no? Please help me understand why there is any need for offset translation?
The destination cluster isn’t write protected, so clients could still produce there, and the source cluster isn’t guaranteed to start mirroring from offset zero. Both cases will cause a difference