Elasticsearch version 7.13.0
editElasticsearch version 7.13.0
editAlso see Breaking changes in 7.13.
Known issues
edit-
Frozen tier: (Windows only) The frozen data tier relies on multiple caching mechanisms to speed up access to searchable snapshot files. One of these caches uses sparse files to avoid creating large files on disk when it is not strictly required. A bug prevented files from being created with the right options to enable sparse support on Windows, leading Elasticsearch to create potentially large files that can end up consuming all the disk space.
This issue is fixed in Elasticsearch versions 7.15.2 and later. There is no known workaround for earlier versions. Filesystems that enable sparse files by default are not affected. For more details, see #79371.
-
If autoscaling is enabled for machine learning, the administrator of the
cluster should increase the cluster setting
xpack.ml.max_open_jobs
to the maximum value of512
. This allows autoscaling to run reliably as it relies on assigning jobs only via memory. Havingxpack.ml.max_open_jobs
as a small number may cause autoscaling to behave unexpectedly. -
Snapshot and restore: If an index is deleted while the cluster is concurrently taking more than one snapshot then there is a risk that one of the snapshots may never complete and also that some shard data may be lost from the repository, causing future restore operations to fail. To mitigate this problem, set
snapshot.max_concurrent_operations: 1
to prevent concurrent snapshot operations:PUT _cluster/settings { "persistent" : { "snapshot.max_concurrent_operations" : 1 } }
This issue is fixed in Elasticsearch versions 7.13.1 and later. It is not possible to repair a repository once it is affected by this issue, so you must restore the repository from a backup, or clear the repository by executing
DELETE _snapshot/<repository name>/*
, or move to a fresh repository. For more details, see #73456. -
If local and remote clusters are on different patch releases, response serialization fails for requests to a remote cluster that use the field capabilities API with CCS. Because Kibana uses this API internally, requests to CCS index patterns will also fail.
This issue is fixed in Elasticsearch version 7.13.3 and later. For more details, see #74504.
Deprecations
edit- Aggregations
- Authentication
- Features/Indices APIs
- Infra/Core
- Infra/Settings
New features
editEnhancements
edit- Aggregations
-
- Improve handling of incompatible after key types in composite aggregations #70839 (issue: #70480)
-
Increase
search.max_bucket
default value by one #70645 (issue: #57042) -
Modest memory savings in a
date_histogram
followed by a terms aggregation #68592 -
Speed up
top_metrics
on hot shards #70579 (issue: #70453) - Speed up aggregations with sub-aggregations #69806 (issues: #63643, #68871)
- Speed up terms aggregation when alone #69377 (issue: #68871)
- Speed up terms aggregation when not force merged #71241 (issue: #71086)
-
Use
#updateTop
to speed upInternalComposite#reduce
#71278
- Allocation
- Analysis
- Audit
- Authentication
-
- Add a deprecation message if a REST wrapper implementing plugin presents #66827
- Service Accounts - Authentication with file tokens #70543
- Service Accounts - Fleet integration #70724
- Service Accounts - Get service account API #71315
- Service Accounts - Initial bootstrap plumbing to add essential classes #70391
- Service Accounts - New CLI tool for managing file tokens #70454
- Service Accounts - delete index backed service account token #71165
- Service Accounts - token name in response to Authenticate API #71382
- Support metadata on API keys #70292 (issue: #48182)
- Authorization
- Autoscaling
-
- Frozen tier autoscaling decider based on shards #71042
- Cluster Coordination
- Distributed
- EQL
- Engine
-
-
Support
include_unloaded_segments
in node stats #69682
-
Support
- Features/ILM+SLM
-
-
Add
max_single_primary_size
as a condition for the ILM rollover action #68917 (issues: #63026, #67842) - Reject creating ILM policies with phase timings that are not greater than or equal to the previous phase #70089 (issue: #70032)
-
Switch built-in policies to
max_primary_shard_size
#69995 (issue: #63026)
-
Add
- Features/Indices APIs
-
- Date math support for aliases #67226 (issue: #20367)
- Introduce separate shard limit for frozen shards #71392 (issues: #34021, #71042)
- Support specifying multiple templates names in delete component template api #70314 (issue: #69973)
- Support specifying multiple templates names in delete composable index template api #70094 (issue: #69973)
- Features/Ingest
-
- Accept more ingest simulate params as integers or strings #66197 (issues: #23823, #65992)
- Extract device type from user agent info #69322
- Network direction processor additions #68712
- Summary option for listing ingest pipelines without their definitions #69756 (issue: #31954)
-
MurmurHash3
support for fingerprint processor #70632 (issue: #69182)
- Features/Java Low Level REST Client
-
- Support new data roles #66947
- Features/Stats
- Features/Watcher
- Infra/Core
- Infra/Logging
- Infra/Scripting
-
- Add Runtime Fields Contexts to Painless Execute API #71374 (issue: #70467)
- Add a new ANTLR lexer for Painless suggestions #70517
- Improve null def access error message #69226 (issue: #53129)
-
Make the available
ScriptContexts
accessible fromScriptService
#70465 - Whitelist the CIDR convenience API #71258 (issue: #60668)
- Query Languages
- Machine Learning
-
- Add new delete trained model aliases API #69195
- Add runtime mappings to data frame analytics source config #69183 (issue: #65056)
-
Adding new
_preview
endpoint for data frame analytics #69453 - Adding support for composite aggregations in anomaly detection #69970
- Allow datafeed and job configs for datafeed preview API #70836 (issue: #70264)
- Improve messages related to assigning machine learning jobs #69752
- Add put and delete trained model alias APIs to high-level REST client #69214
- Speed up training of regression and classification models for data sets with many features #1746
- Adjust the syscall filter to allow mremap and avoid spurious audit logging #1819
- Avoid overfitting in final training by scaling regularizers to account for the difference in the number of training examples. This results in a better match between train and test errors for classification and regression and often slightly improved test errors #1755
- Mapping
-
- Add calculated numeric fields #69531
- Allow specify dynamic templates in bulk request #69948 (issue: #61939)
-
Change default format for
date_nanos
field #70463 (issues: #67063, #69192) - Expose if a field is a metadata field in the field capabilities response #69977
- Field capabilities index action should not fork its execution #69865
- Improve error message for invalid field name #70972 (issue: #70960)
-
New queryable
_tier
metadata field #69288 (issue: #68135) - Output script stats for indexed fields #71219
-
Preserve
half_float
precision in fields API #70653 (issue: #70260) - Support fetching flattened subfields #70916 (issue: #70605)
- Network
- Ranking
- Recovery
-
- Fix retention lease expiry to not mark stale #68577
- SQL
- Search
-
-
Add
_size
and_doc_count
to fields output #70575 (issue: #63569) -
Add
positive_score_impact
torank_features
type #69994 (issue: #68619) -
Add earlier validation for some
SearchSourceBuilder
settings #69548 (issue: #54958) - Allow format sort values of date fields #70357 (issue: #69192)
- Cancel searches earlier #69795
- Close search contexts on reassigned shard #68539
-
Improve lookup for
include_unmapped
field pattern #69984 (issue: #69983) -
Support fetching
_tier
field value #71379 (issues: #63569, #68135)
-
Add
- Security
- Snapshot/Restore
-
- Add searchable snapshot stats for reads from Lucene #70464
- Add support for range reads and retries to URL repositories #69521
-
Change tier preference for
shared_cache
searchable snapshots to frozen only #70786 (issue: #70341) -
Enforce
data_frozen
for partial searchable snapshot_tier_preference
#71155 (issues: #70786, #71014) - Forbid dedicated frozen nodes w/ unfrozen indices #71395
- Include min/max/average file size in Searchable Snapshots Stats API #70294
- Make searchable snapshot cache size effectively zero on non-frozen nodes #71134 (issues: #70341, #70846, #71013)
-
Skip
TRANSLOG
stage for searchable snapshots recovery stage #70311 (issue: #65531) - Use default application credentials for GCS repositories #71239
- Transform
Bug fixes
edit- Aggregations
- Analysis
-
- Ukrainian language plugin can fill up heap #71998
- Authentication
-
- Fix inconsistency of internal user checking #70123
- Engine
- Features/Features
- Features/Java High Level REST Client
- Features/Java Low Level REST Client
- Geo
-
-
Fix overflow in
GeoTileGridTiler
#70222
-
Fix overflow in
- Infra/Logging
- Infra/Scripting
- Machine Learning
-
-
Consider
xpack.ml.max_ml_node_size
ineffective_model_memory_limit
#70473 (issue: #70069) - Do not create machine learning annotations index in upgrade mode #71175
- Do not track machine learning usage when collecting monitoring #71314
-
Ensure
auc_roc
curve is monotonic #70628 - Exclude nested fields in data frame analytics #71400
- Improve readability of messages written when assigning machine learning jobs to nodes #69629 (issue: #59602)
- Ensure the same hyperparameters are chosen if classification or regression training is stopped and restarted, for example, if the node fails #1848
- Fail gracefully if insufficient data is supplied for classification or regression training #1855
- Fail gracefully on encountering unexpected state in restore from snapshot for anomaly detection #1872
- Use appropriate memory ordering flags for aarch64 with string store to avoid excessive string duplication #1888
- Fix autoscaling bug where many jobs take a long time to open #72423
- Use appropriate master timeouts for master actions #72492
- Fix empty overall_buckets response #72542
- Check the out stream exists before consuming it #72455
- Prevent data frame analytics freeze after loading data #72412
-
Consider
- Mapping
- Search
- Transforms