ES|QL FIRST function
field- The search field
sortField-
The sort field
This function calculates the earliest occurrence of the search field (the first parameter), where sorting order is determined by the sort field (the second parameter). This sorting order is always ascending and null values always sort last. Both fields support null, single-valued, and multi-valued input. If the earliest sort field value appears in multiple documents, this function is allowed to return any corresponding search field value.
| field | sortField | result |
|---|---|---|
| boolean | date | boolean |
| boolean | date_nanos | boolean |
| boolean | long | boolean |
| date | date | date |
| date | date_nanos | date |
| date | long | date |
| date_nanos | date | date_nanos |
| date_nanos | date_nanos | date_nanos |
| date_nanos | long | date_nanos |
| double | date | double |
| double | date_nanos | double |
| double | long | double |
| integer | date | integer |
| integer | date_nanos | integer |
| integer | long | integer |
| ip | date | ip |
| ip | date_nanos | ip |
| ip | long | ip |
| keyword | date | keyword |
| keyword | date_nanos | keyword |
| keyword | long | keyword |
| long | date | long |
| long | date_nanos | long |
| long | long | long |
| text | date | keyword |
| text | date_nanos | keyword |
| text | long | keyword |
@timestamp | name | number
"2025-11-25T00:00:00.000Z | alpha | 1"
"2025-11-25T00:00:01.000Z | alpha | 2"
"2025-11-25T00:00:02.000Z | bravo | null"
"2025-11-25T00:00:03.000Z | alpha | 4"
"2025-11-25T00:00:04.000Z | bravo | 5"
"2025-11-25T00:00:05.000Z | charlie | [6, 7, 8]"
"2025-11-25T00:00:06.000Z | delta | null"
From dataset
| STATS first_val = FIRST(number, @timestamp)
| first_val:long |
|---|
| 1 |
This can use a significant amount of memory and ES|QL doesn’t yet grow aggregations beyond the memory available. This function will continue to work until it is used to collect more values than can fit into memory, in which case it will fail the query with a Circuit Breaker Error. This is especially the case when grouping on a field with a large number of unique values, and even more so if the search field has multi-values of high cardinality.