We seem to have a problem with the response time on search-requests towards the ElasticSearch after going from a single node to setting up a clustered solution. Before we set up a cluster the response time was always less than a second. Now I have seen requests take up to two minutes. Seems that it can be related to the index or something? Yesterday I deleted the indexes and reloaded them.
We are running elastic version 6.2.3 on two windows server 2016 machines with together with Informatica configuration. On the first machine we have the primary master node, and on the second machine we have the two second master nodes.
According to the escluster logs I see that all the three nodes have status green and they seem to be connected to eachother.
Any tips on the slow response?
This article discusses an issue with too many requests, but the cluster has not had any more requests now than before. And even after a restart early in the morning the response time is up to two minutes.
I can also mention that we have a heapsize of 4gb, and probably about 50-200 requests per hour. Is it recommended to increase?