So, i found the https://www.elastic.co/guide/en/elasticsearch/reference/2.2/cat-nodes.html API for getting the cluster info. I'm trying to create a breakdown of the cluster memory, and so far what i have is:
Which as far as i can tell is accounting for approximately 35% of the node heap usage. The nodes are all running rear 80% heap, so there is 45% of the heap that can;t be accounted for by fielddata, query cache, sements, or translog. The remaining, unaccounted for data is very slowly rising and takingup more and more space, so i'm worried that eventually we we not have any heap left for normal traffic. Are there other fields that I should target in this query (or a related query) that can help me get a full idea of where that heap is being spent?
we're running 2.2, for the record