Version 2.4
System Linux Centos
————-
hello everyone, I have a great cluster of opensearch servers, there are 7 machines with very high processing capacity, today we already deal with 700 million records. my problem is in Dashboards, a few days ago I had a problem on my server that hosts it and despite the cluster being available, my users were left without access. How could I set up a cluster, at least two servers, to host the Dashboard? can someone suggest me something? I think the biggest challenge is to keep the changes that are made in one, be reflected in the other.
If I understand you right, here is my experience:
Dashboards metadata are stored on OpenSearch itself so any of the instances of Dashboards that are accessed, will pull the data and metadata (definitions of visualizations, dashboards, …) from your OpenSearch cluster; all instances of Dashboards will be pulling the same data and metadata. As a result, the users will see no difference when they hit any of the Dashboards instances.
One standard setup is to have Dashboards running on every instance that is running OpenSearch and have it configured to talk locally to that instance of OpenSearch only. Then have a load balancer that fronts all of those instances of Dashboards.
1 Like