Server

• [Cluster Transaction] fixed regression introduced in '5.4.202' causing 'InvalidOperationException (XYZ is an atomic guard and you cannot load it via the session)' in Client API version 5.4.201 or older when session with ClusterWide mode is loading a non-existing document

Server

• [Audit] added audit log to the database configuration changes
• [Backup] fixed issue with backup not waking up immediately the database when 'UpdateResponsibleNodeForTasksCommand' command is executed
• [Backup] properly restoring Refresh, Integrations and RevisionsForConflicts configuration during the restore process
• [Backup] fixed issue with backup restore cancellation not working
• [Backup] import properly checks for license restrictions
• [Backup] fixed backup timer calculations
• [Cluster] fixed issue with not all databases waking up when server starts to calculate heal and backup status due to concurrent database load timeout
• [Cluster] redirecting to leader PutDatabaseCommand
• [Cluster] added more grace time before moving node into rehab for nodes that are lagging behind to avoid constant topology changes
• [Cluster] fixed potential issue of Heartbeat spam when CPU credits are low on CPU bounded instances like T3 family in AWS
• [Cluster] stabilized the behavior of cluster
• [Cluster Transaction] fixed possible race with creation of document with atomic guard and not getting concurrency exception
• [Debug] added the ability to log internal .NET events to the log
• [Indexes] more strict parsing of TimeSpan to avoid false positives when strings containing timespans are passed to the index
• [JSON] fixed possible JSON corruption due to reading native memory and HttpClient handling serialization tasks in some cases in the background without waiting for them to complete when HTTP/2 is used
• [Http] fixed possible ArgumentException when HTTP/2 is used and we are establishing a connection
• [Monitoring] fixed number of threads returned via SNMP for Windows instances
• [Notifications] introduced max queue size for the notifications that are sent to the Studio
• [Querying] fixed issue with parameters not being added to high latency query alerts
• [Replication] fixed leak of SINK tags in Hub/Sink Replication when Sink Task is defined
• [Replication] fixed issue with SINK tags not being restored as TRXN when getting the document back from the replication
• [Replication] fixed issue with Pull Replication not connecting properly when 2+ Sinks have the same database name
• [Revisions] fixed issue with making revisions orphaned after executing revert revision without configuration
• [Subscriptions] fixed issue with not taking into account if subscription is disabled or not when updating the subscription
• [Subscriptions] fixed memory leak in Subscription when revisions are being processed and we do a lot of skip operations
• [Threads] implemented cleanup mechanism for built-in PoolOfThreads to avoid having a lot of threads in the pool that hold the memory
• [.NET] updated to .NET 8.0.8

Client API

• [Indexes] fixed issue with IndexCreation.CreateIndexes ignoring SearchEngineType property
• [JSON] fixed possible JSON corruption due to reading native memory and HttpClient handling serialization tasks in some cases in the background without waiting for them to complete when HTTP/2 is used
• [Subscriptions] properly registering missing Ids in subscription session
• [NuGet] added ContinuousIntegrationBuild=true to the CI when package is being built

TestDriver

• [Licensing] added configuration options in TestServerOptions to pass license and license related properties in a more easy manner with the ability to throw if no license is provided or license is invalid

Server

• [Backup] Fixed an issue with the backup running on UTC instead of server's local time
• [Backup/Import/Export] Handling the export and import of time series deleted ranges
• [Backup/Import/Export] Fixed an potential race condition issue with applying commands during import / export
• [Compression] Fixed a potential issue with stats of a collection when compression was turned on/off during large import
• [Containers] Set HTTPs scheme in Docker automatically if cert options were passed
• [Containers] Publishing Docker multi-arch images
• [ETL] Replaced deprecated System.Data.SqlClient dependency with Microsoft.Data.SqlClient (existing connection strings will default to Encrypt=Optional option which is mandatory since version 5.0 of Microsoft.Data.SqlClient to ensure backward compatibility)
• [Indexing] Improved memory usage of loaded documents that are compressed
• [Indexing] Fixed the hash of query results after resetting an index
• [JS Admin] Added ability to use internal operation contexts to execute some emergency operations
• [Logging] Audit log improvements
• [Monitoring] Added ability to monitor number of ongoing tasks via SNMP
• [Monitoring] Exposed Monitor.LockContentionCount via SNMP
• [Setup] Fixed regression issue with RavenDB deb package dependencies
• [Setup] Added Ubuntu Deb Packages distribution for Ubuntu 24.04 and dropped distribution for Ubuntu 18.04
• [Patching] Fixed and issue with patching a document metadata
• [Replication] Fixed issue that WaitForReplicationAfterSaveChanges() didn't account itself to calculate majority
• [Server] Added -i | --info switch to Raven.Server
• [Subscriptions] Throwing less SubscriptionDoesNotBelongToNodeException on unstable cluster
• [.NET] Updated to .NET 8.0.7

Studio

• Added navigation buttons to Revisions Compare view

Articles

• Getting started with GraphQL and RavenDB

Server

• [Cluster] Fixed possible race that upon cluster tx failover it could have waited for the task after it was already completed
• [Cluster] Fixed an issue that WaitForIndexNotification wasn't waiting until the command was done
• [Databases] Fixed an issue with a database getting occasionally stuck in due to The database 'xxx' has been unloaded and locked by IdleOperations
• [Debug] Increased timeout for getting Debug Package
• [ETL] Fixed an issue of respecting cancellation during the load phase of Elasticsearch, Kafka and RabbitMQ ETLs
• [ETL] Upgraded Npgsql driver from v5 to v8
• [Expiration/Refresh] Fixed an issue with unreleased buffers on compressed databases
• [Expiration/Refresh] Added a transaction size limit in the Expiration and Refresh commands
• [Import/Export] Decrypting and Encrypting streams should not use synchronous IO in async methods
• [Indexing] Include the index names that are stale after getting a timeout on WaitForIndexesAfterSaveChanges() in the error message
• [Licensing] Fixed an issue with license signature verification on some Linux distributions
• [Logging] Fixed a bug when downloading logs from the Studio returns empty zip
• [Logging] Fixed a potential deadlock in the logging
• [Logging] Unified audit log lines
• [Replication] Fixed NRE in /databases/*/debug/replication/all-items endpoint
• [Subscriptions] On update subscription populate PinToMentorNode from existing state if needed
• [Time Series] Optimized Time Series retention
• [Voron] Fixed a bug in clustering code which could result in runnnig and committing two write transactions concurrently. This could result in InvalidJournalException after the restart.
• [.NET] Updated to .NET 8.0.5

Client

• [Certificates] Throwing meaningful exception when HttpClientHandler does not support certificates
• [Querying] Fixed an issue with missing parentheses when combining Where and Search clauses
• [Changes API] Don't allow automatic fail over in Changes API when tracking node-specific operations
• [Changes API] Fixed an issue with cache wasn't invalidated after Changes API reconnection
• [HTTP] Flushing request stream to send the headers in all customized HttpContent implementations

Studio

• [Cluster Dashboard] Added the Lucene Unmanaged Allocations to the memory widget
• [Indexes] Showing referenced collections
• [Traffic Watch] Fixed an issue that it couldn't view a MultiGet request

Server

• [Certificates] Adjusted client certificate trust logic to ensure compatibility with shorter intermediate certificates