ok, gents. I could really use some guidance on how to resolve issues we're seeing with 2.1 not creating and distributing shards in a balanced way across our cluster. Again the env is 4 node cluster. ~1.2->1.8TB/index per day. Currently we default to 8 shards per index and as of recently it looks like not all the nodes are creating two shards and the disk usage gets high on certain disks in the data path.
What can I do to stop babysitting this by relocating shards manually?