Add May 21 Elasticsearch serverless release notes

release-notes/elastic-cloud-serverless/_snippets/2026-05-21/breaking-changes.md

@@ -0,0 +1,2 @@
+## May 21, 2026 [elastic-2026-05-21-breaking-changes]
+_There are no breaking changes associated with this release._

release-notes/elastic-cloud-serverless/_snippets/2026-05-21/deprecations.md

@@ -0,0 +1,2 @@
+## May 21, 2026 [elastic-2026-05-21-deprecations]
+_There are no deprecations associated with this release._

release-notes/elastic-cloud-serverless/_snippets/2026-05-21/index.md

@@ -0,0 +1,66 @@
+## May 21, 2026 [elastic-release-notes-2026-05-21]
+### Features and enhancements [elastic-2026-05-21-features-enhancements]
+* Add per-tier `deletion_costs` to autoscaling metrics API.
+* Reserve circuit breaker memory for automaton construction. [#148722](https://github.com/elastic/elasticsearch/pull/148722) [#147428](https://github.com/elastic/elasticsearch/issues/147428)
+* Improve ES|QL field extraction on flattened fields. [#148891](https://github.com/elastic/elasticsearch/pull/148891)
+* Add query approximation support for ES|QL queries containing `FORK` command. [#148412](https://github.com/elastic/elasticsearch/pull/148412)
+* Implement bucket metadata. [#146077](https://github.com/elastic/elasticsearch/pull/146077)
+* Add default schema resolution to `UNION_BY_NAME`. [#149176](https://github.com/elastic/elasticsearch/pull/149176)
+* Speed up `REPLACE` on constant regex. [#149033](https://github.com/elastic/elasticsearch/pull/149033)
+* Cut S3 download + Snappy copies on hot path. [#149248](https://github.com/elastic/elasticsearch/pull/149248)
+* Improve ES|QL Parquet footer cache reliability. [#149045](https://github.com/elastic/elasticsearch/pull/149045)
+* Add logging tracking to the `_xpack/usage` endpoint. [#148087](https://github.com/elastic/elasticsearch/pull/148087)
+* Improve ES|QL packed-values grouping performance. [#148681](https://github.com/elastic/elasticsearch/pull/148681)
+* Improve S3 receive buffer pooling for ES|QL external sources. [#149023](https://github.com/elastic/elasticsearch/pull/149023)
+* Improve ES|QL TopN field extraction for external sources. [#149185](https://github.com/elastic/elasticsearch/pull/149185)
+* Improve ES|QL Parquet and ORC footer caching. [#149018](https://github.com/elastic/elasticsearch/pull/149018)
+* Improve performance of the `REPLACE` fast-path in ES|QL. [#149167](https://github.com/elastic/elasticsearch/pull/149167)
+* Improve ES|QL TopN UTF-8 string encoding performance. [#149022](https://github.com/elastic/elasticsearch/pull/149022)
+* Improve performance of Parquet datasource with nested-And short-circuit and clean up filter evaluation. [#148935](https://github.com/elastic/elasticsearch/pull/148935)
+* Improve reporting of external source rows in ES|QL metrics. [#148410](https://github.com/elastic/elasticsearch/pull/148410)
+* Improve performance of pushed-down filter evaluation in the ES|QL Parquet datasource. [#148918](https://github.com/elastic/elasticsearch/pull/148918)
+* Add SIMD `contains` for literal `LIKE` patterns in ES|QL. [#149026](https://github.com/elastic/elasticsearch/pull/149026)
+* Add an ES|QL optimizer rule that rewrites `SUM(X+c)` to `SUM(X) + c*COUNT(X)`. [#145510](https://github.com/elastic/elasticsearch/pull/145510) [#140470](https://github.com/elastic/elasticsearch/issues/140470)
+* Add generic byte matchers for pushdown predicates in ES|QL. [#149030](https://github.com/elastic/elasticsearch/pull/149030)
+* Reduce memory used by translog blob replication. [#148881](https://github.com/elastic/elasticsearch/pull/148881)
+* Reduce translog memory when active shards have zero operations. [#148992](https://github.com/elastic/elasticsearch/pull/148992)
+* Add S3 storage class tracking in object storage tests. [#149069](https://github.com/elastic/elasticsearch/pull/149069)
+* Add `TS_COLLAPSE` command. [#148690](https://github.com/elastic/elasticsearch/pull/148690)
+* Add document index and ID to reindex parse exception messages. [#145146](https://github.com/elastic/elasticsearch/pull/145146) [#48432](https://github.com/elastic/elasticsearch/issues/48432)
+* Enable stateless snapshots and allow relocation.
+* Reject reindexes with wrong parameters. [#145290](https://github.com/elastic/elasticsearch/pull/145290) [#37855](https://github.com/elastic/elasticsearch/issues/37855)
+* Decrease TSDB `look_ahead_time` to 9 minutes. [#145552](https://github.com/elastic/elasticsearch/pull/145552) [#142602](https://github.com/elastic/elasticsearch/issues/142602)
+* Support `jina-embeddings-v5-omni` input format for Jina integration. [#148551](https://github.com/elastic/elasticsearch/pull/148551)
+* Add product use case and origin to APM telemetry. [#148140](https://github.com/elastic/elasticsearch/pull/148140)
+* Add AI bots to user-agent regex. [#148665](https://github.com/elastic/elasticsearch/pull/148665)
+* Improve field metadata caching per shard for mapping. [#148865](https://github.com/elastic/elasticsearch/pull/148865)
+* Assign downsampling to least loaded eligible node. [#148220](https://github.com/elastic/elasticsearch/pull/148220)
+
+### Fixes [elastic-2026-05-21-fixes]
+* Fix missing deprecation warnings in transform create and preview APIs. [#147934](https://github.com/elastic/elasticsearch/pull/147934) [#82936](https://github.com/elastic/elasticsearch/issues/82936) [#82935](https://github.com/elastic/elasticsearch/issues/82935)
+* Fix centroid grouping failures in ES|QL aggregations. [#148815](https://github.com/elastic/elasticsearch/pull/148815) [#141318](https://github.com/elastic/elasticsearch/issues/141318)
+* Fix wrong warning in ES|QL `STATS` for unrolled multivalued fields. [#145968](https://github.com/elastic/elasticsearch/pull/145968)
+* Fix ES|QL "optimized incorrectly" error from project reorder. [#149053](https://github.com/elastic/elasticsearch/pull/149053) [#148612](https://github.com/elastic/elasticsearch/issues/148612)
+* Fix ES|QL Parquet pushdown for non-DOUBLE physical columns. [#149100](https://github.com/elastic/elasticsearch/pull/149100)
+* Fix stalled ES|QL reads from CSV and TSV sources. [#149189](https://github.com/elastic/elasticsearch/pull/149189)
+* Fix ES|QL histogram errors when filtering null values. [#148983](https://github.com/elastic/elasticsearch/pull/148983) [#147854](https://github.com/elastic/elasticsearch/issues/147854)
+* Fix `UNION_BY_NAME` COUNT/MIN/MAX pushdown for absent columns. [#149279](https://github.com/elastic/elasticsearch/pull/149279)
+* Fix unclear ES|QL error for `SPARKLINE` after `TS` command. [#149015](https://github.com/elastic/elasticsearch/pull/149015)
+* Fix ES|QL type errors on multi-file CSV and NDJSON data sources. [#148756](https://github.com/elastic/elasticsearch/pull/148756)
+* Fix ES|QL coordination thread pool usage. [#149034](https://github.com/elastic/elasticsearch/pull/149034)
+* Fix premature release of ES|QL search contexts on failures. [#145960](https://github.com/elastic/elasticsearch/pull/145960) [#145509](https://github.com/elastic/elasticsearch/issues/145509)
+* Fix ES|QL hangs on multi-file compressed external data sources. [#148797](https://github.com/elastic/elasticsearch/pull/148797)
+* Fix ES|QL Parquet NaN handling in column-index statistics. [#149094](https://github.com/elastic/elasticsearch/pull/149094)
+* Fix ES|QL null handling after JVM workaround. [#149050](https://github.com/elastic/elasticsearch/pull/149050)
+* Fix ES|QL sorting on `histogram` and `_tsid` fields. [#149085](https://github.com/elastic/elasticsearch/pull/149085)
+* Fix `?` wildcard matching in `constant_keyword` wildcard queries. [#148585](https://github.com/elastic/elasticsearch/pull/148585) [#141785](https://github.com/elastic/elasticsearch/issues/141785)
+* Fix reindex task status when a child task is banned. [#148199](https://github.com/elastic/elasticsearch/pull/148199) [#146314](https://github.com/elastic/elasticsearch/issues/146314)
+* Fix inference errors when rerank context is missing. [#148921](https://github.com/elastic/elasticsearch/pull/148921)
+* Fix rerank validation for updated inference code path. [#149202](https://github.com/elastic/elasticsearch/pull/149202)
+* Fix chunked inference validation for multiple grouped inputs. [#148660](https://github.com/elastic/elasticsearch/pull/148660)
+* Fix analyzer reload for search-as-you-type and wrapped analyzers. [#147695](https://github.com/elastic/elasticsearch/pull/147695) [#146914](https://github.com/elastic/elasticsearch/issues/146914)
+* Fix `_timeseries` JSON format in ES|QL stored source. [#148716](https://github.com/elastic/elasticsearch/pull/148716) [#148414](https://github.com/elastic/elasticsearch/issues/148414)
+* Align workflows system data stream mappings with Kibana schema. [#148886](https://github.com/elastic/elasticsearch/pull/148886) [#148871](https://github.com/elastic/elasticsearch/issues/148871)
+* Add optional `num_candidates` parameter to kNN retriever. [#148672](https://github.com/elastic/elasticsearch/pull/148672)
+* Fix inner hits overwrite when collapse and nested query both set inner hits. [#148915](https://github.com/elastic/elasticsearch/pull/148915) [#148905](https://github.com/elastic/elasticsearch/issues/148905)
+* Fix `ArrayIndexOutOfBoundsException` in `date_histogram` with `hard_bounds` outside data. [#148765](https://github.com/elastic/elasticsearch/pull/148765) [#148763](https://github.com/elastic/elasticsearch/issues/148763)

release-notes/elastic-cloud-serverless/_snippets/2026-05-21/known-issues.md

@@ -0,0 +1,2 @@
+## May 21, 2026 [elastic-2026-05-21-known-issues]
+_There are no known issues associated with this release._

release-notes/elastic-cloud-serverless/index.md

@@ -12,6 +12,9 @@ Review the changes, fixes, and more to {{serverless-full}}.
 <!-- :::{changelog} /releases
 ::: -->
 
+:::{include} _snippets/2026-05-21/index.md
+:::
+
 :::{include} _snippets/2026-05-19/index.md
 :::