hi, if i have a row in a table that contain large data (not necessary super wide row), say 10 G and a replication factor of 3.
During a repair, if the data of the row in each of the node is simply off by 1 byte, is cassandra smart enough to stream only partial of the data (maybe based on a range of the clustering key) ? or does it have to stream all 10G of data from other 2 nodes and compare and consolidate? If it is the second case, besides bandwith usage spike, is there any other negative impact on cassandra node? thanks
