Currently, the create-index API results in cluster state updates which are submitted unbatched. This can potentially lead to too many cluster state update tasks in cases where there are many concurrent create-index requests, and could be improved by using batching. Note that, we might need to carefully pick a reasonable size to batch these requests since this operation can be potentially expensive and involve mapping creation, etc.
Currently, the create-index API results in cluster state updates which are submitted unbatched. This can potentially lead to too many cluster state update tasks in cases where there are many concurrent create-index requests, and could be improved by using batching. Note that, we might need to carefully pick a reasonable size to batch these requests since this operation can be potentially expensive and involve mapping creation, etc.