One of the problems that we often hear with the elasticsearch community is huge cluster_state. From my prior experience, I have noticed that in majority of the cases too many mapping field definitions is the cause of the problem. For one of our use case, we have plethora of different fields present in our JSON document.
Obviously we don't want each of them to be indexed, but non-indexed fields are also present in cluster_state. Is there a way to mask out these non-indexed fields from cluster_state?