Elastic doesn't use maximum ram allocated

node 3

node 2

node 1 - is the master node

even I do a search that lasts for 3+ minutes, ram stays constant. Can you please help me?

Cluster info:
3 node cluster
4 cpu 15 gb ram per node
7gb allocated to heap

version:
elasticsearch 6.3.0

Each node will use as much heap as needed, so I do not see these pattern indicating any problem apart from possibly an uneven load.

Are all nodes master-eligible?

Do all nodes hold data?

Do you distribute requests across all nodes in the cluster?