In order to increase the availability of Grafana monitoring is there a way to configure multiple ElasticSearch nodes in one datasource and let Grafana 'load balance' between them, or configure explicit primary and failover nodes?
I'm using a 3-node ElasticSearch cluster to store monitoring data, however during a rolling update (or if a node fails) Grafana will error if it can't connect. I have spread the config for different indexes across instances, however 1/3 of the dashboards will typically stop working if a node fails so its not ideal.
