This is a similar issue as
With a key-value table on a 2-node cluster with RF2 and there is a MV that put `value` as partition key.
Insert with ConsistencyLevel.ONE twice with different data, assuming there is a network partition between 2 nodes:
- Node 1 received: pk="a" -> value=1 @ts1
- Node 2 received: pk="a" -> value=2 @ts2 where ts2 > ts1
When querying `value`=1 on MV with ConsistencyLevel.ALL, we got stale row: "a" -> 1.