Elasticsearch version 8.0.0-rc1
  
Also see Breaking changes in 8.0.
Known issues
- 
Do not upgrade production clusters to Elasticsearch 8.0.0-rc1. Elasticsearch 8.0.0-rc1 is
a pre-release of Elasticsearch 8.0 and is intended for testing purposes only.
Upgrades from pre-release builds are not supported and could result in errors or
data loss. If you upgrade from a released version, such as 7.16, to a
pre-release version for testing, discard the contents of the cluster when you are
done. Do not attempt to upgrade to the final 8.0 release.
 
- 
For Elasticsearch 8.0.0-rc1, the Elasticsearch SQL JDBC driver requires
Java 17 or newer. In Elasticsearch 8.0.0-rc2, the JDBC driver will only require Java 8
or newer. https://github.com/elastic/elasticsearch/pull/82325
 
 
Breaking changes
- 
Infra/Core
 
- 
- 
All system indices are hidden indices #79512
 
 
 
- 
Snapshot/Restore
 
- 
- 
Adjust snapshot index resolution behavior to be more intuitive #79670 (issue: #78320)
 
 
 
 
Deprecations
- 
Engine
 
- 
- 
Deprecate setting 
max_merge_at_once_explicit #80574
 
 
 
- 
Machine Learning
 
- 
- 
Deprecate 
estimated_heap_memory_usage_bytes and replace with model_size_bytes #80554
 
 
 
- 
Search
 
- 
- 
Configure 
IndexSearcher.maxClauseCount() based on node characteristics #81525 (issue: #46433)
 
 
 
- 
Transform
 
- 
 
 
Enhancements
- 
Authorization
 
- 
- 
Granting 
kibana_system reserved role access to "all" privileges to .internal.preview.alerts* index #80889 (issues: #76624, #80746, #116374)
 
- 
Granting 
kibana_system reserved role access to "all" privileges to .preview.alerts* index #80746
 
- 
Granting editor and viewer roles access to alerts-as-data indices #81285
 
 
 
- 
ILM+SLM
 
- 
- 
Make unchanged ILM policy updates into noop #82240 (issue: #82065)
 
 
 
- 
Indices APIs
 
- 
 
- 
Infra/Settings
 
- 
 
- 
Ingest
 
- 
- 
Extract more standard metadata from binary files #78754 (issue: #22339)
 
 
 
- 
Machine Learning
 
- 
- 
Add 
deployment_stats to trained model stats #80531
 
- 
The setting 
use_auto_machine_memory_percent now defaults max_model_memory_limit #80532 (issue: #80415)
 
 
 
- 
Monitoring
 
- 
- 
Adding default templates for Metricbeat ECS data #81744
 
 
 
- 
Network
 
- 
- 
Improve slow inbound handling to include response type #80425
 
 
 
- 
Recovery
 
- 
- 
Fix 
PendingReplicationActions submitting lots of NOOP tasks to GENERIC #82092 (issues: #77466, #79837)
 
 
 
- 
Reindex
 
- 
- 
Do not scroll if max docs is less than scroll size (update/delete by query) #81654 (issue: #54270)
 
 
 
- 
Security
 
- 
- 
URL option for 
BaseRunAsSuperuserCommand #81025 (issue: #80481)
 
 
 
 
Bug fixes
- 
Autoscaling
 
- 
- 
Autoscaling use adjusted total memory #80528 (issue: #78750)
 
 
 
- 
Data streams
 
- 
- 
Prohibit restoring a data stream alias with a conflicting write data stream #81217 (issue: #80976)
 
 
 
- 
ILM+SLM
 
- 
- 
Less verbose serialization of snapshot failure in SLM metadata #80942 (issue: #77466)
 
 
 
- 
Indices APIs
 
- 
- 
Fix 
ComposableIndexTemplate equals when composed_of is null #80864
 
 
 
- 
Infra/REST API
 
- 
 
- 
Ingest
 
- 
- 
Adjust default geoip logging to be less verbose #81404 (issue: #81356)
 
 
 
- 
Machine Learning
 
- 
- 
Set model state compatibility version to 8.0.0 #2139