[role="xpack"]
[[sql-search-api]]
=== SQL search API
++++
<titleabbrev>SQL search</titleabbrev>
++++

Returns results for an <<sql-rest-overview,SQL search>>.

[source,console]
----
POST _sql?format=txt
{
  "query": "SELECT * FROM library ORDER BY page_count DESC LIMIT 5"
}
----
// TEST[setup:library]

[[sql-search-api-request]]
==== {api-request-title}

`GET _sql`

`POST _sql`

[[sql-search-api-prereqs]]
==== {api-prereq-title}

* If the {es} {security-features} are enabled, you must have the `read`
<<privileges-list-indices,index privilege>> for the data stream, index,
or alias you search.

[[sql-search-api-limitations]]
===== Limitations

See <<sql-limitations>>.

[[search-api-query-params]]
==== {api-query-parms-title}

`delimiter`::
(Optional, string) Separator for CSV results. Defaults to `,`. The API only
supports this parameter for CSV responses.

`format`::
(Optional, string) Format for the response. For valid values, see
<<sql-rest-format>>.
+
You can also specify a format using the `Accept` HTTP header. If you specify
both this parameter and the `Accept` HTTP header, this parameter takes
precedence.

[role="child_attributes"]
[[sql-search-api-request-body]]
==== {api-request-body-title}

`allow_partial_search_results`::
(Optional, Boolean) If `true`, returns partial results if there are shard
request timeouts or <<shard-failures,shard failures>>. If `false`, returns an error with no
partial results. Defaults to `false`.

`catalog`::
(Optional, string) Default catalog (cluster) for queries. If unspecified, the
queries execute on the data in the local cluster only.
+
experimental:[] See <<modules-cross-cluster-search,{ccs}>>.

`columnar`::
(Optional, Boolean) If `true`, returns results in a columnar format. Defaults to
`false`. The API only supports this parameter for CBOR, JSON, SMILE, and YAML
responses. See <<sql-rest-columnar>>.

`cursor`::
(Optional, string) <<sql-pagination,Cursor>> used to retrieve a set of paginated
results. If you specify a `cursor`, the API only uses the `columnar` and
`time_zone` request body parameters. It ignores other request body parameters.

[[sql-search-api-fetch-size]]
`fetch_size`::
(Optional, integer) Maximum number of rows to return in the response. Defaults
to `1000`.

[[sql-search-api-field-multi-value-leniency]]
`field_multi_value_leniency`::
(Optional, Boolean) If `false`, the API returns an error for fields containing
<<array,array values>>. If `true`, the API returns the first value from the
array with no guarantee of consistent results. Defaults to `false`.

`filter`::
(Optional, object) <<query-dsl,Query DSL>> used to filter documents for the SQL
search. See <<sql-rest-filtering>>.

`index_include_frozen`::
(Optional, Boolean) If `true`, the search can run on frozen indices. Defaults to
`false`.

`keep_alive`::
(Optional, <<time-units,time value>>) Retention period for an
<<sql-async,async>> or <<sql-store-searches,saved synchronous search>>. Defaults
to `5d` (five days).

`keep_on_completion`::
(Optional, Boolean) If `true`, {es} <<sql-store-searches,stores synchronous
searches>> if you also specify the `wait_for_completion_timeout` parameter. If
`false`, {es} only stores <<sql-async,async searches>> that don't finish before
the `wait_for_completion_timeout`. Defaults to `false`.

`page_timeout`::
(Optional, <<time-units,time value>>) Minimum retention period for the scroll
cursor. After this time period, a <<sql-pagination,pagination request>> might
fail because the scroll cursor is no longer available. Subsequent scroll requests
prolong the lifetime of the scroll cursor by the duration of `page_timeout` in
the scroll request. Defaults to `45s` (45 seconds).

`params`::
(Optional, array) Values for parameters in the `query`. For syntax, see
<<sql-rest-params>>.

`query`::
(Required, object) SQL query to run. For syntax, see <<sql-spec>>.

`request_timeout`::
(Optional, <<time-units,time value>>) Timeout before the request fails. Defaults
to `90s` (90 seconds).

include::{es-repo-dir}/search/search.asciidoc[tag=runtime-mappings-def]

[[sql-search-api-time-zone]]
`time_zone`::
(Optional, string) ISO-8601 time zone ID for the search. Several
<<sql-functions-datetime,SQL date/time functions>> use this time zone. Defaults
to `Z` (UTC).

`wait_for_completion_timeout`::
(Optional, <<time-units,time value>>) Period to wait for complete results.
Defaults to no timeout, meaning the request waits for complete search results.
If the search doesn’t finish within this period, the search becomes
<<sql-async,async>>.
+
To <<sql-store-searches,save a synchronous search>>, you must specify this
parameter and the `keep_on_completion` parameter.

[role="child_attributes"]
[[sql-search-api-response-body]]
==== {api-response-body-title}

The SQL search API supports <<sql-rest-format,multiple response formats>>. Most
response formats use a tabular layout. JSON responses contain the following
properties:

`id`::
(string) Identifier for the search. This value is only returned for
<<sql-async,async>> and <<sql-store-searches,saved synchronous searches>>. For
CSV, TSV, and TXT responses, this value is returned in the `Async-ID` HTTP
header.

`is_running`::
(Boolean) If `true`, the search is still running. If `false`, the search has
finished. This value is only returned for <<sql-async,async>> and
<<sql-store-searches,saved synchronous searches>>. For CSV, TSV, and TXT
responses, this value is returned in the `Async-partial` HTTP header.

`is_partial`::
(Boolean) If `true`, the response does not contain complete search results. If
`is_partial` is `true` and `is_running` is `true`, the search is still running.
If `is_partial` is `true` but `is_running` is `false`, the results are partial
due to a failure or timeout.
+
This value is only returned for <<sql-async,async>> and
<<sql-store-searches,saved synchronous searches>>. For CSV, TSV, and TXT
responses, this value is returned in the `Async-partial` HTTP header.

`rows`::
(array of arrays)
Values for the search results.

`columns`::
(array of objects)
Column headings for the search results. Each object is a column.
+
.Properties of `columns` objects
[%collapsible%open]
====
`name`::
(string) Name of the column.

`type`::
(string) Data type for the column.
====

`cursor`::
(string) <<sql-pagination,Cursor>> for the next set of paginated results. For
CSV, TSV, and TXT responses, this value is returned in the `Cursor` HTTP header.