We have Elasticsearch ES cluster in amazon aws with "Load balancer". These ES nodes are accessed using java code deployed in tomcat using ES Java API. The system is failing randomly.
Transport client was initiated with aws load balancer url, and later after these random disconnections between client and nodes, we have replaced with actual IP of the nodes. After this one change the system is looking stable till now.
Also, if anyone could suggest ideal tcp keep alive time, that would be awesome.
ES: 1.7 version
Client OS: Cent OS, EC2 Instances
Client JDK: 1.8.0_25
Client ES: 1.7
Did anyone face this issue with LB as input for transport client param??