It would be helpful for clients to get cluster slots/shards information during a node failover and is loading data.