I'm confused about the property
action.search.shard_count.limit. I assumed it was a node property, but we have started encountering it when querying time series data spread across multiple nodes. The total goes over the default of 1000, but none of the nodes hold 1000 shards on their own.
This line seems to suggest that it was intended to be a limit per node (Property.NodeScope).
But the following compares the property value to the total number of shards on local and remote:
I'm thinking of configuring that property by multiplying it by the number of nodes in the cluster. Does that make sense or is there another recommended approach?