Lagging replicas in distributed queries are now excluded by default if they are behind by more than 5 minutes.
Why it matters
This feature prevents distributed queries from using outdated data from lagging replicas, ensuring more consistent and up-to-date query results by excluding replicas that are delayed beyond a configurable threshold. It improves data freshness and reliability for users querying distributed tables.How to use it
By default, replicas lagging more than 5 minutes are excluded automatically in distributed queries. Users can rely on this default behavior without additional configuration. To customize the lag threshold or behavior, relevant settings can be adjusted (details should be checked in the official docs or pull request).