I have a high write / ingest volume and have configured dedicated ingest nodes for my cluster.
I have non-Elasticsearch workers in front of these ingest nodes which pre-process data and which are happily sending data into the cluster. Currently, there is an AWS NLB in between these nodes and the ingest nodes to provide some resilience in the event of an ingest node going away.
Is there a recommended method for handling the discovery of the ingest nodes rather than having the NLB? Hard-coding IP addresses or even DNS names into the configurations doesn't feel like a great solution.