Background we had a situation were loading shards caused the replicas could respond late, for example if we have 3 nodes and we tried to query with CL=ONE , the response was taking so much time.
Expected test behaviour
three nodes cluster
node 1 :has the shard
node 2: has the shard but not ready [find a way to force the status , maybe update it]
node 3: doesn't have the shard
if we query node 3, it should trigger search via replica and the response performance shouldn't be affected by node2 which has the shard still loading