Repository navigation
[ES|QL] Return the interval in the column metadata #138146
Description
Activity
Pinging @elastic/es-analytical-engine (Team:Analytics)
Note:
- For
BUCKET(@timestamp, 50, start, end)we can include the start and end in the metadata, given these are constants. - For
BUCKET(@timestamp, 1 hour)we cannot include a reasonable start and end. These either have to be obtained from the data, e.g. by running anINLINE STATSto get the min/max - or they need to be picked up from the time picker in Kibana.
Maybe, ESQL could also look at the query that the time picker generates and try to infer the min/max from that; it won't work in all cases, though and would increase complexity a good bit.
Reacted by Stratou- For
- added a parent issue
on Mar 30, 2026 - removed a parent issue
on Apr 20, 2026
Description
When the users are using the BUCKET function (
BUCKET(@timestamp, 50, start, end)), ES is doing an autobucketing and returns the results. The interval in some cases is not easy to get computed. For example when the date histogram returns only one data point.In the case of
BUCKET(@timestamp, 1 hour)we can find the interval by parsing the query but this is also not super convenient.We would like in order to fix bugs like this elastic/kibana#215314 to have the interval as metadata information on the response. Kibana then can use this interval to draw the charts correctly.