Studio

• Fixed accessibility of Certificates view

Server

• [Attachments] Fix the attachment key data extraction
• [Audit Log] Added logging on database settings/index configuration changes
• [Audit Log] Added logging of opening a TCP connection
• [Audit Log] Added logging of memory dumps endpoints usage
• [Audit Log] Fixed a log message of a backup operation
• [Backup] Fixed slow restore of a backup with revisions
• [Backup] Enabled support for multi-part uploads up to 5TB to S3
• [Backup/Export] Implemented rate limiting for backup/restore and import/export operations
• [Certificates] Removed Oracle Key Usage support for newly generated client certificates to fix compatibility with .NET 9
• [Corax] Added support for unbounded time queries in Corax between operations
• [Corax] Optimized performance of cold queries on large data sets
• [Corax] Made default value of Indexing.Corax.MaxMemoizationSizeInMb dependent on total physical installed memory
• [Corax] Fixed query results deduplication when a query has projection and specific terms
• [Corax] Adjusted the starts with behavior inside the Search method to match Lucene
• [Corax] Fixed the handling of document double deletion in internal structures
• [Corax] Fixed backward compatibility for indexes that do not have NonExisting posting list support
• [Data Archival] Archived flag will not be removed when modifying an archived document
• [Debug] Switched to Microsoft.SourceLink.GitHub and introduced .snupkg
• [Debug] Fixed /databases/*/admin/debug/txinfo debug endpoint
• [Debug] Fixed handling of cancellation gracefully in LogTrafficWatch and preventing immediate exit in Raven.Debug traffic command
• [Debug] Enhanced the replication debug endpoint functionality
• [Embeddings Generation] Fixed waiting for embeddings to be stored in a task
• [Embeddings Generation] Fixed embedding generation - skip whitespace / empty strings
• [ETL] Connection string changes will now restart the relevant ETL process
• [JSON] Fixed internal parsing of JSON due to invalid handling of escape positions
• [Logging] Fixed log messages that contain "Raven.Server.Documents.DocumentDatabase" instead of actual database name
• [Memory] Optimized unmanaged memory usage when doing deep paging and using encryption
• [Memory] Fixed internal calculations of memory utilization
• [Memory] Enhanced low memory detection on Windows
• [Patching] Added WaitForIndexes functionality for patch operations
• [Performance] Disabled performance impacting logging configuration
• [PowerBI] Fixed experimental feature error confusion
• [Querying] Return an empty query when an analyzer removes all terms in the search method instead of throwing OutOfRangeException
• [Subscriptions] Fixed OperationCanceledException on waiting subscriptions
• [TestDriver] Made RavenTestDriver testable and fixed testing of ThrowOnInvalidOrMissingLicense
• [Time Series] Preserved time series name order after conflict resolution for Client API compatibility
• [Vector Search] Fixed fetching of fields in auto index handling vectors
• [Vector Search] Fixed C# class generation for RavenVector fields
• [Vector Search] Improved performance of vectors indexing by changing HNSW insert to parallel mode
• [Vector Search] Added ability to query vector searching using a document ID
• [Vector Search] Fixed auto vector search indexes that use quantization and load embeddings from a task
• [.NET] Updated to 8.0.16

Client

• [Querying] Added support for MemoryExtensions.Contains and MemoryExtensions.ContainsAny in LINQ provider and LINQ to JS coverter
• [Qyerying] Fixed an issue that query builder didn't handle properly ContainsAll of an empty array
• [Querying] Fixed the problem with Query improperly serializing an enum when it had an EnumMember attribute

Smuggler

• [Certificates] Revamped certificates filtering/sorting view
• [Export] Added export to stream overload for Smuggler API

Studio

• [Edit] Added ability to save a document in Studio via cluster transaction
• [Querying] Vector Search - Autocomplete Enhancements
• [Revisions] Added a button to copy revision dates

Server

• [Audit Log] Logging enhancements
• [Certificates] Fixed an issue with loading a certificate with a password after removing the password
• [Corax] Fixed NRE when using wildcard search with multiple analyzers
• [Corax] Optimizations and enhancements for deletions
• [Corax] Performance optimizations for range queries on big datasets
• [Databases] Improved management of database idleness detection timers
• [Embeddings Generation] Fixing an error on creating Embeddings Generation task when some other is disabled
• [ETL] Fixed testing functionality of SQL connection strings
• [Indexing] Fixed an issue where indexes could start before custom analyzers and sorters had been loaded
• [Indexing] Fixed memory corruption when indexing Vectors and optimized vector writes
• [Indexing] Improved validation of an index definition when using Run Test functionality
• [Indexing] Optimized vector writes
• [Indexing] Throw relevant exception for embeddings length mismatch
• [Licensing] Introduced grace period for upgrade required modal - 14 days
• [Logging] Fixed logging levels in ETL tasks logs
• [Memory] Prevented possible memory fragmentation in internal allocations
• [Monitoring] Added SNMP endpoint for getting the time since the creation of oldest active transaction
• [Monitoring] Fixed incorrect CPU usage calculation on machines with two physical processors
• [Performance] Low level performance improvements
• [Querying] Throw when using the filter together with vector.search()
• [Querying] Merging scores of duplicated vectors when performing Vector Search
• [Replication] Fixed an issue in which the incoming replication recreated a revision that has been deleted
• [Setup] Fixed ports reservation for server
• [Subscriptions] Snapshot backup restore will restore an original subscription name
• [TestDriver / Embedded] Added more details on how to handle invalid or missing license
• [Voron] Fixed a potential issue with missing removal of an active transaction from in-memory state in some edge conditions
• [.NET] Updated to 8.0.15

Client

• [Querying] Added RavenQuery.Include method which extends existing include functionality to allow more complex operations

Studio

• [Cluster] Improvements to Cluster Debug view
• [Connection strings] Fixed an issue where custom model & endpoint were not shown when editing AI connection string
• [Embeddings Generation] Fixed layout of the creation of a new AI connection
• [Index Errors] Opening the errors by default
• [Indexes] Fixed filtering arrow icons
• [Revisions] Exposed size of a revision
• [Tombstones view] Exposed info about what is blocking tombstones deletion

Features

• [AI] Added ability to generate embeddings via external AI providers or embedded model with the usage of Embeddings Generation Task Docs | Guide

Server

• [Backup] Better selection of databases to wake up for to run a backup
• [Backup] Enhanced backup operation status error reporting
• [Backup] Fixed error when saving the periodic backup status
• [Backup] Improved logic for determining when a backup should run
• [Configuration] Removed obsolete "Storage.TransactionsModeDurationInMin" option
• [Configuration] Default value of Indexing.Corax.VectorSearch.DefaultMinimumSimilarity got changed from 0.75 to 0
• [Corax] Fixed and issue with duplicates when performing AndWith with TermReader
• [Corax] Fixed invalid order of query results when using order by
• [Corax] Fixed missing handling of non existing values in MultiUnaryMatch
• [Corax] Fixed NRE on the dictionary training phase for indexes defined on @all_docs
• [Corax] Fixed querying regression introduced in 7.0.0 build which resulted in the increased latency
• [Counters] Fixed the mechanism of counter tombstones cleanup
• [Indexing] Fixed WaitForIndexesAfterSaveChanges() implementation to handle indexes defined on @all_docs
• [Logging] Fixed logging levels of couple of too detailed messages
• [Memory] Improvement to memory cleanup mechanism of Lucene's managed memory resources
• [Memory] Improved the memory allocation mechanism in Lucene to reduce GC pressure
• [Memory] Fixed calculation of unmanaged memory the debug stats
• [Memory] Fixed an issue with unmanaged memory allocation for a specific pattern
• [Memory] Fixed an issue with managed memoru usage during backups
• [Notifications] Improved implementation of notifications storage
• [Revisions] Fixed possible infinite loop when enforcing revision configuration
• [Security] Added audit logs for export and import operations
• [Voron] Fixed edge cases in SeekBackwardFrom implementation that could affect Time-Series retention
• [Voron] Improved performance of the free space detection to allocate
• [.NET] Updated to 8.0.13

Client

• [Querying] Fixed an issue that "Id" field was not populated in some projections
• Fixed an issue that the Read balance behavior configuration changes was not triggering the speed test

Studio

• [AI] Added new view in AI Hub to manage AI connection strings and Embeddings Generation tasks
• [Index View] Fixed an issue with the default search analyzer not shown in UI
• [Ongoing Tasks View] Better observability of replication tasks
• [Revisions] Added revision bin cleaner to the settings

Features

• [Vector Search] added support for vector search. Docs | Guide
• [ETL] added Amazon SQS support. Docs | Guide
• [ETL] added Snowflake support. Docs | Guide
• [Logging] introduced new logging system. Docs | Guide | Article: Using Grafana Loki

Upgrading from previous versions

• please refer to our Server migration guide before proceeding with Server update and check our list of Server breaking changes available here, Client API breaking changes available here

Server

• [JavaScript] updated to Jint 4.1.0
• this version contains same set of changes as RavenDB 6.2.3

Client API

• [Bulk Insert] enabled compression by default
• [HTTP] switched default http compression algorithm to zstd
• this version contains same set of changes as RavenDB 6.2.3

Studio

• [Admin Logs] revamped view
• [Feedback] revamped form
• [Revisions] added new view to Studio for easier Revision management