If I have a couple of clusters, both connected to each other as remote clusters.
If I perform a search against a single cluster, I get the benefit of caching of search results within a refresh interval window, if I I search the remote cluster again I appear to benefit from caching.
As soon as i prefix a search with *:, or concatenate indices, with a comma, i see +5 to 10seconds of latency. Whereas hitting individual clusters I see sub 50ms response times.
Any suggestions on configuration settings on the coordinator nodes or other areas
(To note the latency between clusters is sub 5ms)