Kibana Index Pattern field discovery concurrency configuration

Index Pattern Field Metadata Refresh

Kibana uses Index Patterns to map Elasticsearch indices to the user interface, utilizing a field discovery process to populate the available fields list. To minimize load on the Elasticsearch Cluster State API, the system employs an internal cache for this metadata.

In environments with a high cardinality of fields, triggering the "Refresh field list" action simultaneously across multiple sessions may lead to race conditions. There is uncertainty regarding how the Kibana server handles these concurrent requests when a slower, outdated metadata response arrives after a more recent update, potentially overwriting the cached field list.

What mechanism does Kibana use to ensure the atomicity of field discovery updates during concurrent refreshes? Is there a configuration to throttle these requests or enforce a versioning check to prevent outdated metadata from overwriting the current cache?