Return more information in RefreshStats #49363
Labels
:Distributed/Engine
Anything around managing Lucene and the Translog in an open shard.
>enhancement
:Search/Search
Search-related issues that do not fall into other categories
Team:Distributed
Meta label for distributed team
Team:Search
Meta label for search team
A common cause of performance issues in elasticsearch is long-lasting or too-frequent refreshes (eg indexing multiple individual docs with
refresh=true
rather than using bulk loads, or enablingeager_global_orginals
on high cardinality fields). Refresh stats currently give us the total number of refreshes and the total time they took, but this isn't particularly useful when trying to track down these refresh issues.One possible enhancement would be to keep track of the number and length of refreshes in the previous 60 mins, and return that; we could also track the tail percentile refresh lengths and warn if a refresh takes longer than a certain amount of time.
The text was updated successfully, but these errors were encountered: