Elasticsearch 8.4 发布

Aggregations

• Fix multi-value handling in composite agg #88638
• Fix: extract matrix stats using bucket_selector buckets_path #88271 (issue: #87454)
• Make the metric in the buckets_path parameter optional #87220 (issue: #72983)
• Propagate alias filters to significance aggs filters #88221 (issue: #81585)

Allocation

• Clamp auto-expand replicas to the closest value #87505 (issue: #84788)
• Prevent re-balancing using outdated node weights in some cases #88385 (issue: #88384)
• Remove any existing read_only_allow_delete index blocks when cluster.routing.allocation.disk.threshold_enabled is set to false #87841 (issue: #86383)
• Replace health request with a state observer #88641

Authentication

• Fix unique realm name check to cover default realms #87999

Authorization

• Add rollover permissions for remote_monitoring_agent #87717 (issue: #84161)

Autoscaling

• Autoscaling during shrink #88292 (issue: #85480)
• Do not include desired nodes in snapshots #87695

Cluster Coordination

• Improve rejection of ambiguous voting config name #89239

Data streams

• Fix renaming data streams with CCR replication #88875 (issue: #81751)

Distributed

• Fixed NullPointerException on bulk request #88385

EQL

• Avoid attempting PIT close on PIT open failure #87498
• Improve EQL Sequence circuit breaker precision #88538 (issue: #88300)

Geo

• Geo_line aggregation returns a geojson point when the resulting line has only one point #89199 (issue: #85748)
• Sort ranges in geo_distance aggregation #89154 (issue: #89147)

Health

• Fix NPE when checking if the last snapshot was success #88811
• Fixing a version check for master stability functionality #89322
• Fixing internal action names #89182
• Using the correct connection to fetch remote master history #87299

Highlighting

• Handle ordering in plain highlighter for multiple inputs #87414 (issue: #87210)

ILM+SLM

• Batch ILM move to retry step task update #86759

Infra/CLI

• Quote paths with whitespace in Windows service CLIs #89072 (issue: #89043)

Infra/Core

• Always close directory streams #88560
• Delete invalid settings for system indices #88903 (issue: #88324)
• Disallow three-digit minor and revision versions #87338
• Handle snapshot restore in file settings #89321 (issue: #89183)
• System indices ignore all user templates #87260 (issues: #42508, #74271)

Infra/Node Lifecycle

• Fix message for stalled shutdown #89254

Infra/Plugins

• Disable URL connection caching in SPIClassIterator #88586 (issue: #88275)

Infra/Scripting

• Script: UpdateByQuery can read doc version if requested #88740

Machine Learning

• Address potential bug where trained models get stuck in starting after being allocated to node #88945
• Fix BERT and MPNet tokenization bug when handling unicode accents #88907 (issue: #88900)
• Fix NLP question_answering task when best answer is only one token #88347
• Include start params in _stats for non-started model deployments #89091
• fix minor tokenization bug when using fill_mask task with roberta tokenizer #88825

Mapping

• Assign the right path to objects merged when parsing mappings #89389 (issue: #88573)
• Don’t modify source map when parsing composite runtime field #89114

Network

• Ensure that the extended socket options TCP_KEEPXXX are available #88935 (issue: #88897)

SQL

• Fix SqlSearchIT testAllTypesWithRequestToOldNodes #88883 (issue: #88866)
• Fix date range checks #87151 (issue: #77179)
• fix object equals #87887

Search

• Fix: use status code 500 for aggregation reduce phase errors if no shard failed #88551 (issue: #20004)
• Override bulk visit methods of exitable point visitor #82120

Security

• Ensure secureString remain open when reloading secure settings #88922

Snapshot/Restore

• Fix queued snapshot assignments after partial snapshot fails due to delete #88470 (issue: #86724)

Transform

• Handle update error correctly #88619

Vector Search

• Deprecate the _knn_search endpoint #88828

Aggregations

• Adding cardinality support for random_sampler agg #86838
• Minor RangeAgg optimization #86935 (issue: #84262)
• Speed counting filters/range/date_histogram aggs #81322
• Update bucket metric pipeline agg paths to allow intermediate single bucket and bucket qualified multi-bucket aggs #85729

Allocation

• Add debug information to ReactiveReason about assigned and unassigned shards #86132 (issue: #85243)
• Optimize log cluster health performance. #87723
• Use desired nodes during data tier allocation decisions #87735

Audit

• Audit API key ID when create or grant API keys #88456
• Include API key metadata in audit log when an API key is created, granted, or updated #88642
• Updatable API keys – logging audit trail event #88276
• User Profile – audit support for security domain #87097

Authentication

• If signature validation fails, reload JWKs and retry if new JWKs are found #88023

Authorization

• App permissions with action patterns do not retrieve privileges #85455
• Cancellable Profile Has Privilege check #87224
• Return action denied error when user with insufficient privileges (manage_own_api_key) attempts a grant API key request #87461 (issue: #87438)
• Update indices permissions to Enterprise Search service account #88703

Autoscaling

• Add processors to autoscaling capacity response #87895
• Keep track of desired nodes status in cluster state #87474

Cluster Coordination

• Deduplicate mappings in persisted cluster state #88479
• Expose segment details in PCSS debug log #87412
• Periodic warning for 1-node cluster w/ seed hosts #88013 (issue: #85222)
• Report overall mapping size in cluster stats #87556

Data streams

• Give doc-value-only mappings to numeric fields on metrics templates #87100

Distributed

• Adding the ability to register a PeerFinderListener to Coordinator #88626
• Make Desired Nodes API operator-only #87778 (issue: #87777)
• Support “dry run” mode for updating Desired Nodes #88305

FIPS

• Log warning when hash function used by cache is not recommended in FIPS mode #86740
• Log warning when hashers for stored API keys or service tokens are not compliant with FIPS #87363

Geo

• Optimize geogrid aggregations for singleton points #87439
• Support cartesian shape with doc values #88487
• Use a faster but less accurate log algorithm for computing Geotile Y coordinate #87515
• Use faster maths to project WGS84 to mercator #88231

Health

• Add health user action for unhealthy SLM policy failure counts #88523
• Adding a transport action to get cluster formation info #87306
• Adding additional capability to the master_is_stable health indicator service #87482
• Creating a transport action for the CoordinationDiagnosticsService #87984
• Move the master stability logic into its own service separate from the HealthIndicatorService #87672
• Polling cluster formation state for master-is-stable health indicator #88397
• Remove cluster block preflight check from health api #87520 (issue: #87464)

ILM+SLM

• Add min_* conditions to rollover #83345
• Track the count of failed invocations since last successful policy snapshot #88398

Infra/Core

• Improve console exception messages #87942
• Print full exception when console is non-interactive #88297
• Stop making index read-only when executing force merge index lifecycle management action #81162 (issue: #81162)
• Stream input and output support for optional collections #88127
• Update version of internal http client #87491

Infra/Logging

• Catch an exception when formatting a string fails #87132

Infra/Scripting

• Script: Add Metadata to ingest context #87309
• Script: Metadata for update context #88333

Infra/Settings

• Convert disk watermarks to RelativeByteSizeValues #88719

Ingest

• Allow pipeline processor to ignore missing pipelines #87354
• Move the ingest attachment processor to the default distribution #87989
• Only perform ensureNoSelfReferences check during ingest when needed #87352 (issue: #87335)
• Removing BouncyCastle dependencies from ingest-attachment plugin #88031

Machine Learning

• Add authorization info to ML config listings #87884
• Add deployed native models to inference_stats in trained model stats response #88187
• Add inference cache hit count to inference node stats #88807
• Add new cache_size parameter to trained_model deployments API #88450
• Expand allowed NER labels to be any I-O-B tagged labels #87091
• Improve scalability of NLP models #87366
• Indicate overall deployment failure if all node routes are failed #88378
• New frequent_items aggregation #83055

Mapping

• Enable synthetic source support on constant keyword fields #88603
• Speed up NumberFieldMapper #85688

Monitoring

• JvmService use SingleObjectCache #87236

Network

• Allow start cluster with unreachable remote clusters #87298
• Increase http.max_header_size default to 16kb #88725 (issue: #88501)

Query Languages

• Add support for VERSION field type in SQL and EQL #87590 (issue: #83375)

Rollup

• [TSDB] Add Kahan support to downsampling summation #87554

SQL

• Implement support for partial search results in SQL CLI #86982 (issue: #86082)
• Update Tableau connector to use connection dialog v2 #88462

Search

• Add mapping stats for indexed dense_vectors #86859
• Improve error when sorting on incompatible types #88399 (issue: #73146)
• Support kNN vectors in disk usage action #88785 (issue: #84801)

Security

• Add setting for tcp_keepalive for oidc back-channel #87868
• Support run_as another user when granting API keys #88335
• Support exists query for API key query #87229
• Updatable API keys – REST API spec and tests #88270
• Updatable API keys – noop check #88346

Snapshot/Restore

• INFO logging of snapshot restore and completion #88257 (issue: #86610)
• Make snapshot deletes not block the repository during data blob deletes #86514
• Retry after all S3 get failures that made progress #88015 (issue: #87243)
• Speed up creating new IndexMetaDataGenerations without removed snapshots #88344
• Update HDFS Repository to HDFS 3.3.3 #88039

Stats

• Sort ingest pipeline stats by use #88035

TLS

• Add issuer to GET _ssl/certificates #88445

Transform

• Add authorization info to transform config listings #87570
• Implement per-transform num_failure_retries setting #87361

Authentication

• Support updates of API key attributes (single operation route) #88186

Health

• Master stability health indicator part 1 (when a master has been seen recently) #86524
• Remove help_url,rename summary to symptom, and user_actions to diagnosis #88553 (issue: #88474)

Infra/Core

• File Settings Service #88329

Infra/Logging

• Stable logging API – the basic use case #86612

Machine Learning

• Make composite aggs in datafeeds Generally Available #88589

Search

• Add mode option to _source field mapper #88211

TSDB

• TSDB: Implement downsampling ILM Action for time-series indices #87269 (issue: #68609)

Vector Search

• Integrate ANN into _search endpoint #88694 (issue: #87625)

Infra/Core

• Upgrade to Log4J 2.18.0 #88237

Network

• Upgrade to Netty 4.1.77 #86630

转自 Elasticsearch version 8.4.0 | Elasticsearch Guide [8.4] | Elastic