Recently i tried to upgrade my cassandra cluster (3 + 3) from cassandra 3.0.15 to 4.1.0 in rolling upgrade manner, and i figured out that when a node with older version of cassandra act as coordinator, it cant read data from newer version of cassandra and vice versa. so there is a huge inconsistency happening in the middle of the upgradation process. and i assume its happening because of schema mismatch between C3 and C4.

So to resolve it can i use nodetool resetlocalschema, in the middle of cassandra node upgradation (3.0.15 to 4.1) to resolve the schema mismatch between 3.x and 4.x.

Suggest a solution to nullify the down time.