# OpenSearch: what changed from 2 to 3

- Product: OpenSearch (https://whatsnew.fyi/product/opensearch)
- Vendor: OpenSearch Software Foundation
- Range: changelog entries numbered after 2.19.6 up to and including 3.8.0, stable releases only
- Entries below: 7 releases (newest first)
- Resolved: 2 is 2.19.6 and 3 is 3.8.0, the newest stable release of each major we track
- Carrying security changes: 0 · CVEs mentioned: 3 · Mentioning breaking changes: 0 · Removing or deprecating something: 1
- Page: https://whatsnew.fyi/product/opensearch/compare/2...3

What's New is an index, not a publisher: every entry below links to the vendor's own release notes, which are the authoritative source. Entries are labelled where they are hand-curated sample data, pre-releases, or drawn from a secondary source such as a developer blog.

Reuse: the summaries, labels and curation here are © What's New. Quote freely with attribution and a link back; wholesale republication of the corpus is not permitted — terms: https://whatsnew.fyi/terms. The vendors' own release notes remain their publishers'.

## What changed (126 changes, grouped by kind)

### Added

#### 3.8.0 (2026-08-05)

- Add API to modify a data stream's backing indices via POST /_data_stream/_modify
- Add multivalue_doc_count aggregation that returns the number of documents with two or more values
- Add pull-based ingestion from Hive tables via new ingestion-hive plugin
- Add support for HTTP/3 on the client side via JDK HttpClient
- Add gRPC support for Point-in-Time (PIT) create and delete operations
- Add gRPC API support for extra_field_values in bulk requests
- Add cross-cluster streaming support to the remote cluster client via Arrow Flight
- Add coordinator-side index-level field domain pruning before can_match for faster time-range queries
- Add bulkscorer script processing interface in ScriptScore query for bulk scoring optimizations
- Add process-wide ObjectInputFilter to reject Java deserialization by default, gated behind bootstrap.serial_filter setting
- Add cluster-level default setting for delayed shard allocation timeout cluster.routing.allocation.unassigned.node_left.delayed_timeout

#### 3.7.0 (2026-06-09)

- Dynamic properties support for pattern-based field definitions without cluster state mapping updates
- Pluggable data format engine with DataFormatAwareEngine for multi-format indexing
- Lucene engine implementation for pluggable data formats
- Merge support for Parquet data format plugin via streaming k-way merge sort
- Directory and IndexInput layers for WritableWarm tiered storage
- Server-side implementation for tiering status APIs (GetTieringStatus and ListTieringStatus)
- Server-side implementation for HotToWarm, WarmToHot, and CancelTiering APIs
- Prefetch settings and stored fields prefetch for WritableWarm tiered storage
- Slow logs, per-query metrics, and migration metrics for WritableWarm tiered storage
- Module wiring and integration tests for WritableWarm tiered storage
- Tiered object storage crate for warm node file routing
- Event-driven scheduler and stage execution for analytics engine
- Coordinator-side DataFusion reduce with streaming Arrow batches
- Distributed aggregation with partial/final mode for analytics engine
- Distributed join planning and execution for analytics engine
- PPL append command support with multi-child stage runtime for Union
- PPL dedup command support via ROW_NUMBER window function
- PPL eventstats and streamstats window function support
- PPL top and rare command support via window functions
- PPL parse command with regex mode via Rust UDFs

#### 3.6.0 (2026-04-07)

- Add bitmap64 query support
- Add warmup phase for pull-based ingestion to prevent serving stale data before catching up with the streaming source
- Implement field_mapping ingestion message mapper for pull-based ingestion
- Add mapper_settings support and field_mapping mapper type for pull-based ingestion
- Add index warmer support for replica shards using segment replication
- Add indices to search request slow log for easier request identification
- Add node-level JVM and CPU runtime metrics following OpenTelemetry semantic conventions
- Add new Sensitive setting property for tiering dynamic settings authorization
- Add adaptive shard selection for bulk writes on append-only indices
- Add support for expected remote cluster name validation in CCS sniff mode
- Add search_settings support to WLM workload groups with initial timeout setting
- Add scroll API support for workload management rule-based autotagging
- Add stream request flag to SearchRequestContext for plugin consumption
- Add intra-segment support for single-value metric aggregations (sum, min, max, avg, stats, cardinality, value_count)
- Add ref_path support for package-based Hunspell dictionary loading with multi-tenant isolation
- Add indexer interface to decouple IndexShard from Engine for pluggable engine architectures

#### 3.5.0 (2026-02-12)

- Support for fields containing dots in their name as literals
- Support for forward translog reading
- Public getter method in SourceFieldMapper to return excluded field
- Relax jar hell check when extended plugins share transitive dependencies
- Public getter method in SourceFieldMapper to return included field
- Support for HTTP/3 server side
- Circuit breaker support for gRPC transport to prevent out-of-memory errors
- Index-level-encryption support for snapshots and remote-store
- Support for hll field mapper to support cardinality rollups
- Tracing support for StreamingRestChannel
- New libs/netty4 module to share common implementation between netty-based plugins and modules
- Validation to make crypto store settings immutable
- Concurrent translog recovery to accelerate segment replication primary promotion
- X-Request-Id to uniquely identify a search request
- TopN selection logic for streaming terms aggregations
- Support for Intra Segment Search
- AdditionalCodecs and EnginePlugin::getAdditionalCodecs hook to allow additional Codec registration

#### 3.4.0 (2025-12-17)

- Allow setting index.creation_date on index creation and restore for plugin compatibility and migrations
- Add support for a ForkJoinPool type
- Add separate shard limit validation for local and remote indices
- Use Lucene pack method for half_float and unsigned_long when using ApproximatePointRangeQuery
- New cluster setting search.query.max_query_string_length_monitor_only
- Add a mapper for context aware segments grouping criteria
- Return full error for GRPC error response
- Add support for repository with Server side encryption enabled and client side encryption as well based on a flag
- Add pluggable gRPC interceptors with explicit ordering
- Add BindableServices extension point to transport-grpc-spi
- Add metrics for the merged segment warmer feature
- Handle deleted documents for filter rewrite sub-aggregation optimization
- Add bulk collect API for filter rewrite sub-aggregation optimization
- Allow collectors take advantage of preaggregated data using collectRange API
- Bulk collection logic for metrics and cardinality aggregations
- Add pointer based lag metric in pull-based ingestion
- Introduce internal API for retrieving metadata about requested indices from transport actions
- Add cluster defaults for merge autoThrottle, maxMergeThreads, and maxMergeCount; add segment size filter to the merged segment warmer
- Add build-tooling to run in FIPS environment
- Add SMILE/CBOR/YAML document format support to Bulk GRPC endpoint

### Changed

#### 3.8.0 (2026-08-05)

- Add HTTP request body decompression (gzip/deflate) to reactor-netty4 transport
- Add cross-setting validator to prevent cluster-manager crash loop when both balance factors are set to zero
- Add support for double, int, and long types in ExtraFieldValue for gRPC bulk indexing
- Add timing logs for repository create, update, and delete operations
- Disable Mustache partial template resolution in search templates
- Use binary serialization for resource usage headers to reduce CPU overhead on the search path

#### 3.6.0 (2026-04-07)

- Remove experimental tag for pull-based ingestion, marking it as public API
- Support Docker distribution builds for ppc64le, arm64, and s390x architectures
- Fallback to Netty HTTP client when AWS CRT client is unavailable on the target platform
- Expose wrapped scorer in ProfileScorer for plugin access to custom scorer methods
- Remove X-Request-Id format restrictions and make maximum length configurable
- Make telemetry Tags immutable with allocation-efficient factories and content-based equality
- Prevent criteria field updates for context-aware indices to simplify version management
- Use ReadAdviseByContext for MMapDirectory instead of Lucene default read advise

#### 3.5.0 (2026-02-12)

- Handle custom metadata files in subdirectory-store
- Add support for missing proto fields in GRPC FunctionScore and Highlight
- Use compact object headers with JDK25+

#### 3.4.0 (2025-12-17)

- Combining filter rewrite and skip list to optimize sub aggregation

### Fixed

#### 3.8.0 (2026-08-05)

- Fix OpenSearchTimeoutException to return HTTP 504 instead of 500
- Fix ClassCastException on malformed mappings in create index to return 400 instead of 500

#### 3.6.0 (2026-04-07)

- Fix copy_to functionality for geo_point fields with object and array values
- Fix field_caps returning empty results for disable_objects mappings and field name corruption
- Fix terms lookup subquery to use cluster max_clause_count setting instead of hardcoded fallback
- Fix terms aggregation performance regression on high-cardinality fields by adding a max cardinality setting
- Fix terms aggregation performance regression using segment-to-global ordinals mapping
- Fix array_index_out_of_bounds_exception with wildcard and aggregations under concurrent access
- Add range validations in query builder and field mapper
- Fix synonym_graph filter failure with word_delimiter_graph by handling analyzer dependency ordering
- Fix index template pattern collision false positives for multi-wildcard patterns
- Fix SecurityException when using opensearch.cgroups.hierarchy.override setting
- Fix SLF4J component error caused by slf4j-api version mismatch with log4j binding
- Fix JSON escaping in task details log metadata
- Fix ProfileScorer.getChildren() to expose wrapped scorer in the scorer tree hierarchy

#### 3.5.0 (2026-02-12)

- Fix Snapshot rename replacement unbounded length rename

#### 3.3.2 (2025-10-30)

- Fix sub-aggregator casting for search with profile=true in Star Tree
- Allow JRT protocol URLs in protection domain extraction for Java Agent
- Fix bwc @timestamp upgrade issue by adding a version check on skip_list param (also in 3.3.1)
- Fix issue with updating core with a patch number other than 0 (also in 3.3.1)
- Fix IndexOutOfBoundsException when running include/exclude on non-existent prefix in terms aggregations
- Add S3Repository.LEGACY_MD5_CHECKSUM_CALCULATION to list of repository-s3 settings

#### 3.3.1 (2025-10-22)

- Fix sub-aggregator casting for search with profile=true

### Deprecated

#### 3.8.0 (2026-08-05)

- Deprecate RestClient in favor of the internal HTTP client

## Release notes

### 3.8.0

- Date: 2026-08-05
- Version: 3.8.0
- Original notes: https://github.com/opensearch-project/OpenSearch/releases/tag/3.8.0
- Permalink: https://whatsnew.fyi/product/opensearch/releases/3.8.0

- **added** — Add API to modify a data stream's backing indices via POST /_data_stream/_modify
- **added** — Add multivalue_doc_count aggregation that returns the number of documents with two or more values
- **added** — Add pull-based ingestion from Hive tables via new ingestion-hive plugin
- **added** — Add support for HTTP/3 on the client side via JDK HttpClient
- **added** — Add gRPC support for Point-in-Time (PIT) create and delete operations
- **added** — Add gRPC API support for extra_field_values in bulk requests
- **added** — Add cross-cluster streaming support to the remote cluster client via Arrow Flight
- **added** — Add coordinator-side index-level field domain pruning before can_match for faster time-range queries
- **added** — Add bulkscorer script processing interface in ScriptScore query for bulk scoring optimizations
- **added** — Add process-wide ObjectInputFilter to reject Java deserialization by default, gated behind bootstrap.serial_filter setting
- **added** — Add cluster-level default setting for delayed shard allocation timeout cluster.routing.allocation.unassigned.node_left.delayed_timeout
- **changed** — Add HTTP request body decompression (gzip/deflate) to reactor-netty4 transport
- **changed** — Add cross-setting validator to prevent cluster-manager crash loop when both balance factors are set to zero
- **changed** — Add support for double, int, and long types in ExtraFieldValue for gRPC bulk indexing
- **changed** — Add timing logs for repository create, update, and delete operations
- **deprecated** — Deprecate RestClient in favor of the internal HTTP client
- **changed** — Disable Mustache partial template resolution in search templates
- **changed** — Use binary serialization for resource usage headers to reduce CPU overhead on the search path
- **fixed** — Fix OpenSearchTimeoutException to return HTTP 504 instead of 500
- **fixed** — Fix ClassCastException on malformed mappings in create index to return 400 instead of 500

##### Version 3.8.0 Release Notes

Compatible with OpenSearch and OpenSearch Dashboards version 3.8.0

###### Features

* Add API to modify a data stream's backing indices (`POST /_data_stream/_modify`) ([#22487](https://github.com/opensearch-project/OpenSearch/pull/22487))
* Add `multivalue_doc_count` aggregation that returns the number of documents with two or more values ([#20472](https://github.com/opensearch-project/OpenSearch/pull/20472))
* Add pull-based ingestion from Hive tables via new `ingestion-hive` plugin ([#21700](https://github.com/opensearch-project/OpenSearch/pull/21700))
* Add support for HTTP/3 on the client side via JDK HttpClient ([#21718](https://github.com/opensearch-project/OpenSearch/pull/21718))
* Add gRPC support for Point-in-Time (PIT) create and delete operations ([#22105](https://github.com/opensearch-project/OpenSearch/pull/22105))
* Add gRPC API support for `extra_field_values` in bulk requests ([#21907](https://github.com/opensearch-project/OpenSearch/pull/21907))
* Add cross-cluster streaming support to the remote cluster client via Arrow Flight ([#22359](https://github.com/opensearch-project/OpenSearch/pull/22359))
* Add coordinator-side index-level field domain pruning before `can_match` for faster time-range queries ([#21865](https://github.com/opensearch-project/OpenSearch/pull/21865))
* Add bulkscorer script processing interface in ScriptScore query for bulk scoring optimizations ([#22423](https://github.com/opensearch-project/OpenSearch/pull/22423))
* Add process-wide ObjectInputFilter to reject Java deserialization by default, gated behind `bootstrap.serial_filter` setting ([#22073](https://github.com/opensearch-project/OpenSearch/pull/22073))
* Add cluster-level default setting for delayed shard allocation timeout (`cluster.routing.allocation.unassigned.node_left.delayed_timeout`) ([#22379](https://github.com/opensearch-project/OpenSearch/pull/22379))

###### Enhancements

* Add HTTP request body decompression (gzip/deflate) to reactor-netty4 transport ([#22382](https://github.com/opensearch-project/OpenSearch/pull/22382))
* Add cross-setting validator to prevent cluster-manager crash loop when both balance factors are set to zero ([#22391](https://github.com/opensearch-project/OpenSearch/pull/22391))
* Add support for double, int, and long types in ExtraFieldValue for gRPC bulk indexing ([#22058](https://github.com/opensearch-project/OpenSearch/pull/22058))
* Add timing logs for repository create, update, and delete operations ([#22489](https://github.com/opensearch-project/OpenSearch/pull/22489))
* Add parent action name to `ProcessorGenerationContext` for conditional system-generated processor logic ([#22195](https://github.com/opensearch-project/OpenSearch/pull/22195))
* Add getter for remote translog transfer tracker to enable plugin access to tracking metrics ([#22453](https://github.com/opensearch-project/OpenSearch/pull/22453))
* Deprecate `RestClient` in favor of the internal HTTP client ([#22116](https://github.com/opensearch-project/OpenSearch/pull/22116))
* Disable Mustache partial template resolution in search templates ([#22438](https://github.com/opensearch-project/OpenSearch/pull/22438))
* Use binary serialization for resource usage headers to reduce CPU overhead on the search path ([#21230](https://github.com/opensearch-project/OpenSearch/pull/21230))
* Preserve resolved search pipeline id and inline pipeline source through query rewrite ([#22501](https://github.com/opensearch-project/OpenSearch/pull/22501))
* Fork `GET _aliases` serialization to MANAGEMENT thread pool to avoid blocking transport threads ([#21954](https://github.com/opensearch-project/OpenSearch/pull/21954))
* Strengthen `scroll_id` validation to prevent OOM from crafted scroll IDs ([#22396](https://github.com/opensearch-project/OpenSearch/pull/22396))
* Add depth tracking to recursive deserialization to prevent StackOverflowError ([#22404](https://github.com/opensearch-project/OpenSearch/pull/22404))

_[Truncated at 4000 characters — full notes: https://github.com/opensearch-project/OpenSearch/releases/tag/3.8.0]_

### 3.7.0

- Date: 2026-06-09
- Version: 3.7.0
- Original notes: https://github.com/opensearch-project/OpenSearch/releases/tag/3.7.0
- Permalink: https://whatsnew.fyi/product/opensearch/releases/3.7.0

- **added** — Dynamic properties support for pattern-based field definitions without cluster state mapping updates
- **added** — Pluggable data format engine with DataFormatAwareEngine for multi-format indexing
- **added** — Lucene engine implementation for pluggable data formats
- **added** — Merge support for Parquet data format plugin via streaming k-way merge sort
- **added** — Directory and IndexInput layers for WritableWarm tiered storage
- **added** — Server-side implementation for tiering status APIs (GetTieringStatus and ListTieringStatus)
- **added** — Server-side implementation for HotToWarm, WarmToHot, and CancelTiering APIs
- **added** — Prefetch settings and stored fields prefetch for WritableWarm tiered storage
- **added** — Slow logs, per-query metrics, and migration metrics for WritableWarm tiered storage
- **added** — Module wiring and integration tests for WritableWarm tiered storage
- **added** — Tiered object storage crate for warm node file routing
- **added** — Event-driven scheduler and stage execution for analytics engine
- **added** — Coordinator-side DataFusion reduce with streaming Arrow batches
- **added** — Distributed aggregation with partial/final mode for analytics engine
- **added** — Distributed join planning and execution for analytics engine
- **added** — PPL append command support with multi-child stage runtime for Union
- **added** — PPL dedup command support via ROW_NUMBER window function
- **added** — PPL eventstats and streamstats window function support
- **added** — PPL top and rare command support via window functions
- **added** — PPL parse command with regex mode via Rust UDFs

##### Version 3.7.0 Release Notes

Compatible with OpenSearch and OpenSearch Dashboards version 3.7.0

###### Features

* Add dynamic properties support for pattern-based field definitions without cluster state mapping updates ([#20816](https://github.com/opensearch-project/OpenSearch/pull/20816))
* Add pluggable data format engine with DataFormatAwareEngine for multi-format indexing ([#21181](https://github.com/opensearch-project/OpenSearch/pull/21181))
* Add Lucene engine implementation for pluggable data formats ([#21299](https://github.com/opensearch-project/OpenSearch/pull/21299))
* Add merge support for Parquet data format plugin via streaming k-way merge sort ([#21079](https://github.com/opensearch-project/OpenSearch/pull/21079))
* Add directory and IndexInput layers for WritableWarm tiered storage ([#21178](https://github.com/opensearch-project/OpenSearch/pull/21178))
* Add server-side implementation for tiering status APIs (GetTieringStatus and ListTieringStatus) ([#21220](https://github.com/opensearch-project/OpenSearch/pull/21220))
* Add server-side implementation for HotToWarm, WarmToHot, and CancelTiering APIs ([#21295](https://github.com/opensearch-project/OpenSearch/pull/21295))
* Add prefetch settings and stored fields prefetch for WritableWarm tiered storage ([#21285](https://github.com/opensearch-project/OpenSearch/pull/21285))
* Add slow logs, per-query metrics, and migration metrics for WritableWarm tiered storage ([#21332](https://github.com/opensearch-project/OpenSearch/pull/21332))
* Add module wiring and integration tests for WritableWarm tiered storage ([#21427](https://github.com/opensearch-project/OpenSearch/pull/21427))
* Add tiered object storage crate for warm node file routing ([#21204](https://github.com/opensearch-project/OpenSearch/pull/21204))
* Add event-driven scheduler and stage execution for analytics engine ([#21242](https://github.com/opensearch-project/OpenSearch/pull/21242))
* Add coordinator-side DataFusion reduce with streaming Arrow batches ([#21356](https://github.com/opensearch-project/OpenSearch/pull/21356))
* Add distributed aggregation with partial/final mode for analytics engine ([#21457](https://github.com/opensearch-project/OpenSearch/pull/21457))
* Add distributed join planning and execution for analytics engine ([#21639](https://github.com/opensearch-project/OpenSearch/pull/21639))
* Add PPL `append` command support with multi-child stage runtime for Union ([#21474](https://github.com/opensearch-project/OpenSearch/pull/21474))
* Add PPL `dedup` command support via ROW_NUMBER window function ([#21622](https://github.com/opensearch-project/OpenSearch/pull/21622))
* Add PPL `eventstats` and `streamstats` window function support ([#21734](https://github.com/opensearch-project/OpenSearch/pull/21734))
* Add PPL `top` and `rare` command support via window functions ([#21593](https://github.com/opensearch-project/OpenSearch/pull/21593))
* Add PPL `parse` command with regex mode via Rust UDFs ([#21573](https://github.com/opensearch-project/OpenSearch/pull/21573))
* Add PPL `rex` command with sed and extract modes ([#21550](https://github.com/opensearch-project/OpenSearch/pull/21550))
* Add PPL `spath` command with auto-extract mode via json_extract_all UDF ([#21664](https://github.com/opensearch-project/OpenSearch/pull/21664))
* Add 7 PPL JSON scalar functions to analytics engine route ([#21513](https://github.com/opensearch-project/OpenSearch/pull/21513))
* Add 23 PPL datetime scalar functions to analytics engine route ([#21556](https://github.com/opensearch-project/OpenSearch/pull/21556))
* Add 14 additional PPL datetime functions (Wave A) including strftime, date_format, maketime ([#21582](https://github.com/opensearch-project/OpenSearch/pull/21582))
* Add 30+ PPL math scalar functions to analytics engine ([#21520](https://github.com/opensearch-project/OpenSearch/pull/21520))
* Add PPL string scalar functions to analytics engine (18 functions) ([

_[Truncated at 4000 characters — full notes: https://github.com/opensearch-project/OpenSearch/releases/tag/3.7.0]_

### 3.6.0

- Date: 2026-04-07
- Version: 3.6.0
- Original notes: https://github.com/opensearch-project/OpenSearch/releases/tag/3.6.0
- Permalink: https://whatsnew.fyi/product/opensearch/releases/3.6.0

- **added** — Add bitmap64 query support
- **added** — Add warmup phase for pull-based ingestion to prevent serving stale data before catching up with the streaming source
- **added** — Implement field_mapping ingestion message mapper for pull-based ingestion
- **added** — Add mapper_settings support and field_mapping mapper type for pull-based ingestion
- **added** — Add index warmer support for replica shards using segment replication
- **added** — Add indices to search request slow log for easier request identification
- **added** — Add node-level JVM and CPU runtime metrics following OpenTelemetry semantic conventions
- **added** — Add new Sensitive setting property for tiering dynamic settings authorization
- **added** — Add adaptive shard selection for bulk writes on append-only indices
- **added** — Add support for expected remote cluster name validation in CCS sniff mode
- **added** — Add search_settings support to WLM workload groups with initial timeout setting
- **added** — Add scroll API support for workload management rule-based autotagging
- **added** — Add stream request flag to SearchRequestContext for plugin consumption
- **added** — Add intra-segment support for single-value metric aggregations (sum, min, max, avg, stats, cardinality, value_count)
- **added** — Add ref_path support for package-based Hunspell dictionary loading with multi-tenant isolation
- **added** — Add indexer interface to decouple IndexShard from Engine for pluggable engine architectures
- **changed** — Remove experimental tag for pull-based ingestion, marking it as public API
- **changed** — Support Docker distribution builds for ppc64le, arm64, and s390x architectures
- **changed** — Fallback to Netty HTTP client when AWS CRT client is unavailable on the target platform
- **changed** — Expose wrapped scorer in ProfileScorer for plugin access to custom scorer methods
- **changed** — Remove X-Request-Id format restrictions and make maximum length configurable
- **changed** — Make telemetry Tags immutable with allocation-efficient factories and content-based equality
- **changed** — Prevent criteria field updates for context-aware indices to simplify version management
- **changed** — Use ReadAdviseByContext for MMapDirectory instead of Lucene default read advise
- **fixed** — Fix copy_to functionality for geo_point fields with object and array values
- **fixed** — Fix field_caps returning empty results for disable_objects mappings and field name corruption
- **fixed** — Fix terms lookup subquery to use cluster max_clause_count setting instead of hardcoded fallback
- **fixed** — Fix terms aggregation performance regression on high-cardinality fields by adding a max cardinality setting
- **fixed** — Fix terms aggregation performance regression using segment-to-global ordinals mapping
- **fixed** — Fix array_index_out_of_bounds_exception with wildcard and aggregations under concurrent access
- **fixed** — Add range validations in query builder and field mapper
- **fixed** — Fix synonym_graph filter failure with word_delimiter_graph by handling analyzer dependency ordering
- **fixed** — Fix index template pattern collision false positives for multi-wildcard patterns
- **fixed** — Fix SecurityException when using opensearch.cgroups.hierarchy.override setting
- **fixed** — Fix SLF4J component error caused by slf4j-api version mismatch with log4j binding
- **fixed** — Fix JSON escaping in task details log metadata
- **fixed** — Fix ProfileScorer.getChildren() to expose wrapped scorer in the scorer tree hierarchy

##### Version 3.6.0 Release Notes

Compatible with OpenSearch and OpenSearch Dashboards version 3.6.0

###### Features

* Add bitmap64 query support ([#20606](https://github.com/opensearch-project/OpenSearch/pull/20606))
* Add warmup phase for pull-based ingestion to prevent serving stale data before catching up with the streaming source ([#20526](https://github.com/opensearch-project/OpenSearch/pull/20526))
* Implement field_mapping ingestion message mapper for pull-based ingestion ([#20729](https://github.com/opensearch-project/OpenSearch/pull/20729))
* Add mapper_settings support and field_mapping mapper type for pull-based ingestion ([#20722](https://github.com/opensearch-project/OpenSearch/pull/20722))
* Remove experimental tag for pull-based ingestion, marking it as public API ([#20704](https://github.com/opensearch-project/OpenSearch/pull/20704))

###### Enhancements

* Add index warmer support for replica shards using segment replication ([#20650](https://github.com/opensearch-project/OpenSearch/pull/20650))
* Add indices to search request slow log for easier request identification ([#20588](https://github.com/opensearch-project/OpenSearch/pull/20588))
* Add node-level JVM and CPU runtime metrics following OpenTelemetry semantic conventions ([#20844](https://github.com/opensearch-project/OpenSearch/pull/20844))
* Add new `Sensitive` setting property for tiering dynamic settings authorization ([#20901](https://github.com/opensearch-project/OpenSearch/pull/20901))
* Add adaptive shard selection for bulk writes on append-only indices ([#20065](https://github.com/opensearch-project/OpenSearch/pull/20065))
* Add support for expected remote cluster name validation in CCS sniff mode ([#20532](https://github.com/opensearch-project/OpenSearch/pull/20532))
* Add `search_settings` support to WLM workload groups with initial `timeout` setting ([#20536](https://github.com/opensearch-project/OpenSearch/pull/20536))
* Add scroll API support for workload management rule-based autotagging ([#20151](https://github.com/opensearch-project/OpenSearch/pull/20151))
* Add stream request flag to SearchRequestContext for plugin consumption ([#20530](https://github.com/opensearch-project/OpenSearch/pull/20530))
* Support Docker distribution builds for ppc64le, arm64, and s390x architectures ([#20678](https://github.com/opensearch-project/OpenSearch/pull/20678))
* Fallback to Netty HTTP client when AWS CRT client is unavailable on the target platform ([#20698](https://github.com/opensearch-project/OpenSearch/pull/20698))
* Add intra-segment support for single-value metric aggregations (sum, min, max, avg, stats, cardinality, value_count) ([#20503](https://github.com/opensearch-project/OpenSearch/pull/20503))
* Expose wrapped scorer in ProfileScorer for plugin access to custom scorer methods ([#20549](https://github.com/opensearch-project/OpenSearch/pull/20549))
* Fix ProfileScorer.getChildren() to expose wrapped scorer in the scorer tree hierarchy ([#20607](https://github.com/opensearch-project/OpenSearch/pull/20607))
* Remove X-Request-Id format restrictions and make maximum length configurable ([#21048](https://github.com/opensearch-project/OpenSearch/pull/21048))
* Make telemetry Tags immutable with allocation-efficient factories and content-based equality ([#20788](https://github.com/opensearch-project/OpenSearch/pull/20788))
* Add `ref_path` support for package-based Hunspell dictionary loading with multi-tenant isolation ([#20840](https://github.com/opensearch-project/OpenSearch/pull/20840))
* Prevent criteria field updates for context-aware indices to simplify version management ([#20250](https://github.com/opensearch-project/OpenSearch/pull/20250))
* Add indexer interface to decouple IndexShard from Engine for pluggable engine architectures ([#20675](https://github.com/opensearch-project/OpenSearch/pull/20675))
* Use `ReadAdviseByContext` for MMapDirectory instead of Lucene default read advise ([#21062](https://github.com/opensearch-proje

_[Truncated at 4000 characters — full notes: https://github.com/opensearch-project/OpenSearch/releases/tag/3.6.0]_

### 3.5.0

- Date: 2026-02-12
- Version: 3.5.0
- Original notes: https://github.com/opensearch-project/OpenSearch/releases/tag/3.5.0
- Permalink: https://whatsnew.fyi/product/opensearch/releases/3.5.0

- **added** — Support for fields containing dots in their name as literals
- **added** — Support for forward translog reading
- **added** — Public getter method in SourceFieldMapper to return excluded field
- **added** — Relax jar hell check when extended plugins share transitive dependencies
- **added** — Public getter method in SourceFieldMapper to return included field
- **added** — Support for HTTP/3 server side
- **added** — Circuit breaker support for gRPC transport to prevent out-of-memory errors
- **added** — Index-level-encryption support for snapshots and remote-store
- **added** — Support for hll field mapper to support cardinality rollups
- **added** — Tracing support for StreamingRestChannel
- **added** — New libs/netty4 module to share common implementation between netty-based plugins and modules
- **added** — Validation to make crypto store settings immutable
- **added** — Concurrent translog recovery to accelerate segment replication primary promotion
- **added** — X-Request-Id to uniquely identify a search request
- **added** — TopN selection logic for streaming terms aggregations
- **added** — Support for Intra Segment Search
- **added** — AdditionalCodecs and EnginePlugin::getAdditionalCodecs hook to allow additional Codec registration
- **changed** — Handle custom metadata files in subdirectory-store
- **changed** — Add support for missing proto fields in GRPC FunctionScore and Highlight
- **changed** — Use compact object headers with JDK25+
- **fixed** — Fix Snapshot rename replacement unbounded length rename

##### Version 3.5.0 Release Notes

Compatible with OpenSearch and OpenSearch Dashboards version 3.5.0

###### Added
- Add support for fields containing dots in their name as literals ([#19958](https://github.com/opensearch-project/OpenSearch/pull/19958))
- Add support for forward translog reading ([#20163](https://github.com/opensearch-project/OpenSearch/pull/20163))
- Added public getter method in `SourceFieldMapper` to return excluded field ([#20205](https://github.com/opensearch-project/OpenSearch/pull/20205))
- Add integ test for simulating node join left event when data node cluster state publication lag because the cluster applier thread being busy ([#19907](https://github.com/opensearch-project/OpenSearch/pull/19907)).
- Relax jar hell check when extended plugins share transitive dependencies ([#20103](https://github.com/opensearch-project/OpenSearch/pull/20103))
- Added public getter method in `SourceFieldMapper` to return included field ([#20290](https://github.com/opensearch-project/OpenSearch/pull/20290))
- Support for HTTP/3 (server side) ([#20017](https://github.com/opensearch-project/OpenSearch/pull/20017))
- Add circuit breaker support for gRPC transport to prevent out-of-memory errors ([#20203](https://github.com/opensearch-project/OpenSearch/pull/20203))
- Add index-level-encryption support for snapshots and remote-store ([#20095](https://github.com/opensearch-project/OpenSearch/pull/20095))
- Adding BackWardCompatibility test for remote publication enabled cluster ([#20221](https://github.com/opensearch-project/OpenSearch/pull/20221))
- Support for hll field mapper to support cardinality rollups ([#20129](https://github.com/opensearch-project/OpenSearch/pull/20129))
- Add tracing support for StreamingRestChannel  ([#20361](https://github.com/opensearch-project/OpenSearch/pull/20361))
- Introduce new libs/netty4 module to share common implementation between netty-based plugins and modules (transport-netty4, transport-reactor-netty4) ([#20447](https://github.com/opensearch-project/OpenSearch/pull/20447))
- Add validation to make crypto store settings immutable ([#20123](https://github.com/opensearch-project/OpenSearch/pull/20123))
- Introduce concurrent translog recovery to accelerate segment replication primary promotion ([#20251](https://github.com/opensearch-project/OpenSearch/pull/20251))
- Update to `almalinux:10` ([#20482](https://github.com/opensearch-project/OpenSearch/pull/20482))
- Add X-Request-Id to uniquely identify a search request ([#19798](https://github.com/opensearch-project/OpenSearch/pull/19798))
- Added TopN selection logic for streaming terms aggregations ([#20481](https://github.com/opensearch-project/OpenSearch/pull/20481))
- Added support for Intra Segment Search ([#19704](https://github.com/opensearch-project/OpenSearch/pull/19704))
- Introduce AdditionalCodecs and EnginePlugin::getAdditionalCodecs hook to allow additional Codec registration ([#20411](https://github.com/opensearch-project/OpenSearch/pull/20411))

###### Changed
- Handle custom metadata files in subdirectory-store ([#20157](https://github.com/opensearch-project/OpenSearch/pull/20157))
- Add support for missing proto fields in GRPC FunctionScore and Highlight ([#20169](https://github.com/opensearch-project/OpenSearch/pull/20169))
- Ensure all modules are included in INTEG_TEST testcluster distribution ([#20241](https://github.com/opensearch-project/OpenSearch/pull/20241))
- Cleanup HttpServerTransport.Dispatcher in Netty tests ([#20160](https://github.com/opensearch-project/OpenSearch/pull/20160))
- Use compact object headers with JDK25+ ([#20392](https://github.com/opensearch-project/OpenSearch/pull/20392))
- Add `cluster.initial_cluster_manager_nodes` to testClusters OVERRIDABLE_SETTINGS ([#20348](https://github.com/opensearch-project/OpenSearch/pull/20348))
- Add BigInteger support for unsigned_long fields in gRPC transport ([#20346](https://github.com/opensearch-project/OpenSearch/pul

_[Truncated at 4000 characters — full notes: https://github.com/opensearch-project/OpenSearch/releases/tag/3.5.0]_

### 3.4.0

- Date: 2025-12-17
- Version: 3.4.0
- Original notes: https://github.com/opensearch-project/OpenSearch/releases/tag/3.4.0
- Permalink: https://whatsnew.fyi/product/opensearch/releases/3.4.0

- **added** — Allow setting index.creation_date on index creation and restore for plugin compatibility and migrations
- **added** — Add support for a ForkJoinPool type
- **added** — Add separate shard limit validation for local and remote indices
- **added** — Use Lucene pack method for half_float and unsigned_long when using ApproximatePointRangeQuery
- **added** — New cluster setting search.query.max_query_string_length_monitor_only
- **added** — Add a mapper for context aware segments grouping criteria
- **added** — Return full error for GRPC error response
- **added** — Add support for repository with Server side encryption enabled and client side encryption as well based on a flag
- **added** — Add pluggable gRPC interceptors with explicit ordering
- **added** — Add BindableServices extension point to transport-grpc-spi
- **added** — Add metrics for the merged segment warmer feature
- **added** — Handle deleted documents for filter rewrite sub-aggregation optimization
- **added** — Add bulk collect API for filter rewrite sub-aggregation optimization
- **added** — Allow collectors take advantage of preaggregated data using collectRange API
- **added** — Bulk collection logic for metrics and cardinality aggregations
- **added** — Add pointer based lag metric in pull-based ingestion
- **added** — Introduce internal API for retrieving metadata about requested indices from transport actions
- **added** — Add cluster defaults for merge autoThrottle, maxMergeThreads, and maxMergeCount; add segment size filter to the merged segment warmer
- **added** — Add build-tooling to run in FIPS environment
- **added** — Add SMILE/CBOR/YAML document format support to Bulk GRPC endpoint
- **changed** — Combining filter rewrite and skip list to optimize sub aggregation

##### Version 3.4.0 Release Notes

Compatible with OpenSearch and OpenSearch Dashboards version 3.4.0

###### Added
- Allow setting index.creation_date on index creation and restore for plugin compatibility and migrations ([#19931](https://github.com/opensearch-project/OpenSearch/pull/19931))
- Add support for a ForkJoinPool type ([#19008](https://github.com/opensearch-project/OpenSearch/pull/19008))
- Add seperate shard limit validation for local and remote indices ([#19532](https://github.com/opensearch-project/OpenSearch/pull/19532))
- Use Lucene `pack` method for `half_float` and `unsigned_long` when using `ApproximatePointRangeQuery` ([#19553](https://github.com/opensearch-project/OpenSearch/pull/19553))
- New cluster setting search.query.max_query_string_length_monitor_only ([#19539](https://github.com/opensearch-project/OpenSearch/pull/19539))
- Add a mapper for context aware segments grouping criteria ([#19233](https://github.com/opensearch-project/OpenSearch/pull/19233))
- Return full error for GRPC error response ([#19568](https://github.com/opensearch-project/OpenSearch/pull/19568))
- Add support for repository with Server side encryption enabled and client side encryption as well based on a flag ([#19630](https://github.com/opensearch-project/OpenSearch/pull/19630))
- Add pluggable gRPC interceptors with explicit ordering([#19005](https://github.com/opensearch-project/OpenSearch/pull/19005))
- Add BindableServices extension point to transport-grpc-spi ([#19304](https://github.com/opensearch-project/OpenSearch/pull/19304))
- Add metrics for the merged segment warmer feature ([#18929](https://github.com/opensearch-project/OpenSearch/pull/18929))
- Handle deleted documents for filter rewrite sub-aggregation optimization ([#19643](https://github.com/opensearch-project/OpenSearch/pull/19643))
- Add bulk collect API for filter rewrite sub-aggregation optimization ([#19933](https://github.com/opensearch-project/OpenSearch/pull/19933))
- Allow collectors take advantage of preaggregated data using collectRange API ([#20009](https://github.com/opensearch-project/OpenSearch/pull/20009))
- Bulk collection logic for metrics and cardinality aggregations ([#20067](https://github.com/opensearch-project/OpenSearch/pull/20067))
- Add pointer based lag metric in pull-based ingestion ([#19635](https://github.com/opensearch-project/OpenSearch/pull/19635))
- Introduced internal API for retrieving metadata about requested indices from transport actions ([#18523](https://github.com/opensearch-project/OpenSearch/pull/18523))
- Add cluster defaults for merge autoThrottle, maxMergeThreads, and maxMergeCount; Add segment size filter to the merged segment warmer ([#19629](https://github.com/opensearch-project/OpenSearch/pull/19629))
- Add build-tooling to run in FIPS environment ([#18921](https://github.com/opensearch-project/OpenSearch/pull/18921))
- Add SMILE/CBOR/YAML document format support to Bulk GRPC endpoint ([#19744](https://github.com/opensearch-project/OpenSearch/pull/19744))
- Make test-suite runnable under FIPS compliance support ([#18491](https://github.com/opensearch-project/OpenSearch/pull/18491))
- Implement GRPC Search params `Highlight`and `Sort` ([#19868](https://github.com/opensearch-project/OpenSearch/pull/19868))
- Implement GRPC ConstantScoreQuery, FuzzyQuery, MatchBoolPrefixQuery, MatchPhrasePrefix, PrefixQuery, MatchQuery ([#19854](https://github.com/opensearch-project/OpenSearch/pull/19854))
- Add async periodic flush task support for pull-based ingestion ([#19878](https://github.com/opensearch-project/OpenSearch/pull/19878))
- Add support for context aware segments ([#19098](https://github.com/opensearch-project/OpenSearch/pull/19098))
- Implement GRPC FunctionScoreQuery ([#19888](https://github.com/opensearch-project/OpenSearch/pull/19888))
- Implement error_trace parameter for bulk requests ([#19985](https://github.com/opensearch-project/OpenSearch/pull/19985))
- Allow the truncate filter in normalizers ([#19778](

_[Truncated at 4000 characters — full notes: https://github.com/opensearch-project/OpenSearch/releases/tag/3.4.0]_

### 3.3.2

- Date: 2025-10-30
- Version: 3.3.2
- Original notes: https://github.com/opensearch-project/OpenSearch/releases/tag/3.3.2
- Permalink: https://whatsnew.fyi/product/opensearch/releases/3.3.2

- **fixed** — Fix sub-aggregator casting for search with profile=true in Star Tree
- **fixed** — Allow JRT protocol URLs in protection domain extraction for Java Agent
- **fixed** — Fix bwc @timestamp upgrade issue by adding a version check on skip_list param
- **fixed** — Fix issue with updating core with a patch number other than 0
- **fixed** — Fix IndexOutOfBoundsException when running include/exclude on non-existent prefix in terms aggregations
- **fixed** — Add S3Repository.LEGACY_MD5_CHECKSUM_CALCULATION to list of repository-s3 settings

##### Version 3.3.2 Release Notes

Compatible with OpenSearch 3.3.2 and OpenSearch Dashboards 3.3.0

###### Fixed
* [Star Tree] Fix sub-aggregator casting for search with profile=true ([#19652](https://github.com/opensearch-project/OpenSearch/pull/19652))
* [Java Agent] Allow JRT protocol URLs in protection domain extraction ([#19683](https://github.com/opensearch-project/OpenSearch/pull/19683))
* Fix bwc @timestamp upgrade issue by adding a version check on skip_list param ([#19671](https://github.com/opensearch-project/OpenSearch/pull/19671))
* Fix issue with updating core with a patch number other than 0 ([#19377](https://github.com/opensearch-project/OpenSearch/pull/19377))
* Fix IndexOutOfBoundsException when running include/exclude on non-existent prefix in terms aggregations ([#19637](https://github.com/opensearch-project/OpenSearch/pull/19637))
* Add S3Repository.LEGACY_MD5_CHECKSUM_CALCULATION to list of repository-s3 settings ([#19789](https://github.com/opensearch-project/OpenSearch/pull/19789))

###### Dependencies
* Bump ch.qos.logback modules from 1.5.18 to 1.5.20 in HDFS test fixture ([#19764](https://github.com/opensearch-project/OpenSearch/pull/19764))
* Bump org.bouncycastle:bc-fips from 2.1.1 to 2.1.2 ([#19817](https://github.com/opensearch-project/OpenSearch/pull/19817))

### 3.3.1

- Date: 2025-10-22
- Version: 3.3.1
- Original notes: https://github.com/opensearch-project/OpenSearch/releases/tag/3.3.1
- Permalink: https://whatsnew.fyi/product/opensearch/releases/3.3.1

- **fixed** — Fix issue with updating core with a patch number other than 0
- **fixed** — Fix sub-aggregator casting for search with profile=true
- **fixed** — Fix bwc @timestamp upgrade issue by adding a version check on skip_list param

##### Version 3.3.1 Release Notes

OpenSearch 3.3.1 Only.

###### Fixed
* Fix issue with updating core with a patch number other than 0 ([#19377](https://github.com/opensearch-project/OpenSearch/pull/19377))
* [Star Tree] Fix sub-aggregator casting for search with profile=true ([#19652](https://github.com/opensearch-project/OpenSearch/pull/19652))
* Fix bwc @timestamp upgrade issue by adding a version check on skip_list param ([#19671](https://github.com/opensearch-project/OpenSearch/pull/19671))