IMPORTANT: No additional bug fixes or documentation updates
will be released for this version. For the latest information, see the
current release documentation.
Elasticsearch version 8.1.0edit
Also see Breaking changes in 8.1.
Known issuesedit
-
Parsing a request when the last element in an array is filtered out
(for instance using
_source_includes) fails. This is due to a bug in Jackson parser. Fixed in Elasticsearch 8.6.1 (#91456)
Breaking changesedit
Bug fixesedit
- Aggregations
- Allocation
- Authorization
- Autoscaling
- Client
- Distributed
- Engine
- Geo
- ILM+SLM
-
-
Fix
PolicyStepsRegistry'scachedStepsnull handling #84588
-
Fix
- Indices APIs
- Infra/Core
-
- Always re-run Feature migrations which have encountered errors #83918 (issue: #83917)
-
Copy
trace.idin threadcontext stash #83218 -
Preserve context in
ResultDeduplicator#84038 (issue: #84036) -
Registration of
SystemIndexMigrationTasknamed xcontent objects #84192 (issue: #84115) -
Update system index mappings if
_metais null #83896 (issue: #83890)
- Infra/REST API
-
- Do not allow safelisted media types on Content-Type #83448
- Infra/Scripting
- Ingest
- Machine Learning
-
- Allow autoscaling to work when vertical scaling is possible #84242 (issue: #84198)
-
Correctly capture min stats for
inference.ingest_processorsin ML usage #82352 - Fail queued inference requests with cause if the process crashes #81584
-
Fix NLP tokenization
never_splithandling around punctuation #82982 -
Fix
ZeroShotClassificationConfigupdate mixing fields #82848 - Fix bug where initial scale from 0→1 could scale too high #84244
- Fix submit after shutdown in process worker service #83645 (issue: #83633)
-
Fixes
categorize_textparameter validation to be parse order independent #82628 (issue: #82629) - Record node shutdown start time for each node #84355
- Register the named X-content parser for snapshot upgrade params #84420 (issue: #84419)
- Retry anomaly detection job recovery during relocation #83456
-
Return
zxxforlang_ident_model_1if no valid text is found for language identification #82746 (issue: #81933) - Text structure finder caps exclude lines pattern at 1000 characters #84236 (issue: #83434)
- Validate vocabulary on model deployment #81548 (issue: #81470)
- Wait for model process to stop in stop deployment #83644
- Fix a bug in the tuning of the hyperparameters when training regression classification models #2128
- Improve training stability for regression and classification models #2144, #2147, #2150
- Avoid edge cases in the classification weights calculation to maximize minimum recall which could lead to only a single class being predicted #2194
- Address cause of "[CStatisticalTests.cc@102] Test statistic is nan" log errors #2196
- Address possible causes of "x = NaN, distribution = N5boost4math23students_t_distribution" log errors #2197
- Fix bug restoring data gatherer state for time of day and week anomaly detection functions. This could lead to "No queue item for time" and "Time is out of range. Returning earliest bucket index" log errors #2213
- Mapping
- Network
- Packaging
- Recovery
- SQL
-
- Fix txt format for empty result sets #83376
- Search
- Security
- Snapshot/Restore
-
-
Adjust
LinuxFileSystemNatives.allocatedSizeInBytesfor aarch64 architectures #81376 (issues: #80437, #81362) - Distinguish "missing repository" from "missing repository plugin" #82457 (issue: #81758)
-
Fix
DirectBlobContainerIndexInputcloning method #84341 (issue: #84238) - Move get snapshots serialization to management pool #83215
-
Preserve context in
snapshotDeletionListeners#84089 (issue: #84036)
-
Adjust
- TSDB
-
- Fix time series timestamp meta missing #80695
- Transform
- Watcher
Deprecationsedit
- CRUD
- Cluster Coordination
-
- Remove last few mentions of Zen discovery #80410
- Search
- SQL
Enhancementsedit
- Aggregations
- Allocation
- Analysis
-
- Expose Japanese completion filter to kuromoji analysis plugin #81858
- Authentication
- Authorization
- Cluster Coordination
-
- Add detail to slow cluster state warning message #83221
- Batch Index Settings Update Requests #82896 (issue: #79866)
- Improve node-join task descriptions #80090
-
Make
PeerFinderlog messages happier #83222 - More compact serialization of metadata #82608 (issue: #77466)
- Paginate persisted cluster state #78875
- Reduce verbosity-increase timeout to 3 minutes #81118
- Use network recycler for publications #80650 (issue: #80111)
- Data streams
- ILM+SLM
- Indices APIs
- Infra/Circuit Breakers
- Infra/Core
- Infra/REST API
- Infra/Scripting
-
- Add $ syntax as a shortcut for field in Painless #80518
-
Add
BinaryDocValuesFieldto replaceBytesRef(ScriptDocValues)#79760 - Add a geo point field for the scripting fields api #81395
- Add date fields to the scripting fields api #81272
- Add half float mapping to the scripting fields API #82294
- Add scaled float to the scripting fields API #82275
-
Add support for
GeoShapeto the scripting fields API #81617 - Fields API for IP mapped type #81396
- Fields API for byte, double, float, integer, long, short #81126 (issue: #79105)
- Fields API for flattened mapped type #82590
-
Fields API for x-pack
constant_keyword#82292 - Fields API for x-pack version, doc version, seq no, mumur3 #81476
- Improve support for joda datetime to java datetime in Painless #83099
- Keyword fields API support #81266
- Make wildcard accessible from the scripting field API #82763
- Ordinal field data plumbing #80970 (issue: #79105)
- Support boolean fields in Fields API #80043 (issue: #79105)
- Time series compile and cache evict metrics #79078 (issue: #62899)
- Infra/Settings
-
-
Optimize duplicated code block in
MetadataUpdateSettingsService#82048
-
Optimize duplicated code block in
- Machine Learning
-
- Add ability to update the truncation option at inference #80267
- Add error counts to trained model stats #82705
- Add latest search interval to datafeed stats #82620 (issue: #82405)
- Adds new MPNet tokenization for NLP models #82234
- Force delete trained models #80595
- Improve error message on starting scrolling datafeed with no matching indices #81069 (issue: #81013)
- Report thread settings per node for trained model deployments #81723 (issue: #81149)
- Set default value of 30 days for model prune window #81377
- Track token positions and use source string to tag NER entities #81275
- Warn when creating job with an unusual bucket span #82145 (issue: #81645)
- Improve skip_model_update rule behavior #2096
- Prevent over-subscription of threads in pytorch_inference #2141
- Mapping
- Monitoring
- Network
- Recovery
-
-
Adjust
indices.recovery.max_bytes_per_secaccording to external settings #82819
-
Adjust
- SQL
- Search
-
-
Add
scripted_metricagg context tounsigned_long#64422 (issue: #64347) - Add field usage support for vectors #80608
- Allow doc-values only search on boolean fields #82925 (issues: #82409, #81210, #52728)
- Allow doc-values only search on date types #82602 (issues: #82409, #81210, #52728)
- Allow doc-values only search on ip fields #82929 (issues: #82409, #81210, #52728)
- Allow doc-values only search on keyword fields #82846 (issues: #82409, #81210, #52728)
- Allow doc-values only search on number types #82409 (issues: #81210, #52728)
-
Rewrite
matchandmatch_phrasequeries totermqueries onkeywordfields #82612 (issue: #82515) - Short cut if reader has point values #80268
-
Support combining
_shardspreference param with<custom-string>#80024 (issue: #80021)
-
Add
- Security
- Snapshot/Restore
- Stats
- TSDB
- Transform