Hi have a use case where the topology looks something likes this (source topic) → Enrich data from materialized view, MAT1 → Rekey → Filter → REKEY → Create Materialized view MAT1.
Note the in rekey operation, I am changing the key and I am guaranteeing the old and new key will fall into the same partition.
As per my understanding, due to the rekey operations, the new changelog topic is created. Hence the values read at access and update of the materialized view won’t be consistent. Is this understanding correct? Or the operations at the same stream processor instance are handled atomically?