Elasticsearch version 7.17.0
Elasticsearch version 7.17.0
Also see Breaking changes in 7.17.
Known issues
- Elasticsearch fails to parse system index migration data from the cluster state (issue: #84115)
- Feature migration refuses to run if all features requiring upgrade have encountered errors (issue: #83917)
- Migrating system indices fails for tasks with
mapper_parsing_exception
(issue: #83779)
These issues will impact any subsequent upgrade to 8.x, and are fixed in 7.17.1. We strongly recommend that all users upgrade to a later patch release of 7.17 before upgrading to 8.x.
- 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) - The deprecated
index.mapper.dynamic
setting can break your cluster. It can only be set using the Update index settings API. Symptoms include nodes failing to start or shards failing to allocate. Do not use this setting in versions prior to 7.17.22. The bug is fixed in 7.17.22. (issue: #109160)
Deprecations
Mapping
Transform
Enhancements
Audit
Authorization
- Allow read template with cluster monitor privilege #82046 (issue: #78832)
- Make authorization performance log message configurable and disabled by default #82648 (issue: #75439)
- Add Kibana system permissions for Endpoint action indices #81953
ILM+SLM
- Add an index→step cache to the
PolicyStepsRegistry
#82316 (issue: #77466) - Migrate legacy/v2/component templates away from custom attributes routing #82472 (issue: #82170)
- Migrate to data tiers API dry run on any ILM status #82226 (issue: #82169)
Infra/Core
- Allow scaling executors to reject tasks after shutdown #81856 (issues: #77017, #77178)
- Prevent direct upgrade of indices from 6.8 to 8.0 #82689 (issue: #81326)
Infra/Logging
- Adjust ILM policy for deprecation logs #82833
- Do no use
x-opaque-id
for deduplicating Elastic originating requests #82855 (issue: #82810)
Infra/Settings
Java Low Level REST Client
- Expose HTTP client and allow overriding meta header #81955
License
- Add
DEBUG
log when document level security and field level security usage is detected #82182 (issue: #79152)
Machine Learning
- Move datafeed stats action off of master node #82271
Watcher
- Prevent watcher from starting if its templates are missing #82395
Bug fixes
Aggregations
- Fix bucket keys format for range aggregations on float field #81801 (issue: #81749)
- Fix cardinality aggregation in asynchronous search #82108
- Fix missing fields in range aggregation response for date fields #82732 (issue: #82688)
Allocation
- Correct context for batched reroute notifications #83019
Authorization
CRUD
- Fix potential listener leak in
TransportBulkAction
#81894
Geo
- Handle degenerated rectangles in vector tiles #82404 (issue: #81891)
GeoPolygonDecomposer
might fail due to numerical errors when calculating intersection with the dateline #82953 (issue: #82840)
ILM+SLM
- Migrate the coldest node attribute for data tier routing #81940 (issue: #81633)
- Retry ILM step to safely refresh the cached phase #82613 (issue: #81921)
- Fix ILM allocate action to allow only
total_shards_per_node
#81944 (issue: #81943)
Indices APIs
Infra/Core
Infra/Logging
- Add
doPrivileged
section in deprecation logger #81819 (issue: #81708) - Always emit product origin to deprecation log if present #83115
Ingest
- Filter enrich policy index deletes to just the policy’s associated indices #82568
- Fix enrich cache corruption bug #82441 (issue: #82340)
Java Low Level REST Client
- Fix version resolution and encoding in LLRC #81989
Machine Learning
- Fix annotations index maintenance after reindexing #82304 (issue: #82250)
- Improve cleanup for model snapshot upgrades #81831 (issue: #81578)
- Make delete intervening results more selective #82437
- Skip time to next interval with data for datafeeds with aggs #82488 (issue: #82406)
- Update running process when global calendar changes #83044
- Avoid transient poor time series modeling after detecting new seasonal components. This can affect cases where there are fast and slow repeats in the data, for example 30 minutes and 1 day, and the job uses a short bucket length. #2167 (issue: #2166)
Monitoring
Network
- Correct context for
ClusterConnManager
listener #83035
Search
- Fix bug where field is not returned if it has the same prefix as a nested field #82922 (issue: #82905)
Settings
- Change
deprecation.skip_deprecated_settings
to work with dynamic settings #81836 - Check both node and cluster settings in
NodeDeprecationChecks
#82487 (issue: #82484) - Ignore dynamic settings specified by
deprecation.skip_deprecated_settings
in node deprecation checks #82883 (issue: #82889)
Snapshot/Restore
- Always fail snapshot deletion listeners on master failover #82361 (issue: #81596)
- Fix potential repository corruption during master failover #82912 (issue: #82911)
- Remove requirement for key setting on Azure client settings #82030
- Support GKE workload identity for searchable snapshots #82974 (issue: #82702)
Stats
- Correct context for
CancellableSOCache
listener #83021
Upgrades
Search
Upgrade to Lucene 8.11.1 #81900
- Elasticsearch version 7.16.3 »