I have been running a small, 3-node, ES 1.4.5 cluster. I put a load balancer(HAproxy 1.4.24) in front of it to balance write requests from an Apache Storm topology.
I realised I was loosing some data, and having a look at the Storm logs I could see instances of this error:
2016-01-01T13:02:56.467+0000 STDIO [INFO] ERROR: None of the configured nodes were available: [[ES1][SzFmCc3-T16H0Q7Z9En9Nw][es1][inet[/10.0.0.28:9300]]]
10.0.0.28 is the IP of the load balancer.
I tried removing the load balancer and just add all the node addresses to a java TransportClient. So far, removing the LB layer seems to have fixed the issue.
Does any one have any clue on why this was happening? I'm trying to figure out if this is really a good solution for the issue.