What's new

Server

• [Encryption] Fixing the recovery process on storage startup which could cause overwrite of valid data and result in "Unable to decrypt page" error
• [Encryption] Fixing potential segmentation fault error on Linux on attempt to encrypt empty page that effectively was not in use
• [Indexing] Fixing internals in Lucene to prevent from killing the server by an unhandled exception

Server

• [Alerts] Using standard deviation for Low swap size warning
• [Backups] Fixing ocasionally happening problem that backups were not triggered
• [Backups] Fixing error handling for backups to S3
• [Backups] Added Backup.TempPath configuration option
• [Cloud]Retry CPU credits sync on startup
• [Cluster] Prevent concurrent bootstrap
• [Cluster] Cluster observer will move node to rehab upon large gap from the other nodes
• [Cluster] Fixing missing revisions from cluster transactions
• [Cluster] Limiting cluster log truncation by time
• [Cluster] Promote from rehab with indexes that were explicitly disabled
• [Cluster] Stay in rehab with index error
• [Configuration] Adding Indexing.MaximumSizePerSegmentInMb, Indexing.MergeFactor, Indexing.LargeSegmentSizeToMergeInMb, Indexing.NumberOfLargeSegmentsToMergeInSingleBatch, Indexing.MaxTimeForMergesToKeepRunningInSec
• [Debug] Adding /admin/stats end point to Debug Package
• [Docker] Updating docker base image
• [Encryption] Introducing global stack to EncryptionBuffersPool for better memory management
• [ETL] Fixing IndexOutOfRangeException while iterating the ETL processes what could result in process crash
• [Indexing] Fixing partitioning in Bloom Filters
• [Indexing] Fixing JS map-reduce indexes when grouping by nested object
• [Indexing] Adding support for complex reduce keys in JS map-reduce indexes
• [Indexing] Fixing the index replacement behavior if any error happened meanwhile
• [Indexing] Improving memory management in Lucene to reduce GC cost
• [Indexing] De-referencing index resources under low memory
• [Indexing] Better management of Lucene's merges to reduce the size of transaction
• [Indexing] Fixing CreateField returning a single result when passing the same field name
• [Indexing] Fixing progress calculation of indexes processing counter or time series
• [Indexing] Improve the performance of LoadDocument when referencing very large number of documents
• [Indexing] Improve the performance of initial index run
• [Indexing] Adding support for return block statements in group by + allow arrow functions which return objects (w/o explicit return) in JS indexes map
• [Indexing] Restricting number of items in fields cache of LuceneDocumentConverter
• [Indexing] Increasing default value of Indexing.Encrypted.TransactionSizeLimitInMb configuration option to 96 MB
• [Linux] Fixing SWAP detection on HDD over SSD in Posix systems when NVMe SSD drives are available
• [Patching] Fixing regression issue with pathing large number of documents
• [Querying] Fixing metadata projection returning LastModified in incorrect format
• [Querying] Allow to include on the output of facets
• [Querying] Converting >= and <= queries to between queries automatically
• [Querying] Fixing potential issues with float comparison
• [Replication] Avoiding the update of sibling etag on new connections
• [Replication] Fixing the issue with change vector is not updated after replication
• [Replication] Fixing issue that resolving document conflict in Studio removed the attachment
• [Replication] Fixing issue with missing attachment breaking the replication
• [Replication] Fixing duplicate attachments problem in conflicted documents
• [Replication] Checking conflict status for attachment and deleted attachment which caused infinite loop
• [Replication/Smuggler]Adding unused database Ids to import/export for preventing "conflict" in import destination cluster replication
• [Revisions] Fixing issue that deleting a document with Resolved flag created a revision
• [Revisions] Fixing unhandled cases of conflicts in revisions
• [Subscriptions] Fixing the subscription behavior on SubscriptionDoesNotBelongToNodeException
• [Subscriptions] Fixing calculation of redirect node in case of failover
• [Time Series] Fixing optimistic concurrency violation error when patching multiple types
• [Time Series] Fixing parsing issue when querying with the usage of time series names containing spaces
• [Voron] Fixing the validation of overflow pages during the storage recovery process on startup
• [Voron] Fixing possible AccessViolationException in when deleting entries from fixed size trees
• [Voron] Fixing NRE on attempt to get temp page from DecompressionBuffersPool
• [Voron] Deleting large journals instead of adding them to recycle pool to speed up the startup time
• [Voron] Adding cleanup of Temp directory on storage startup
• [Voron] Fixing seeking to a record when using startAfter parameter
• [.NET Core] Updated to .NET Core 3.1.11
• Improved error handling in timer threads

Client API

• [Conventions]added DisableTopologyCache convention
• [Changes API] Fixing changes API interal operation so it should use the same node as the operation
• [Changes API] Marking Changes API connection state as faulted if failed to initialize
• [Operations] Fixing WaitForCompletion throwing InvalidOperationException for RevertRevisionsOperation
• [Querying] Adding option to project the last modified date from queries
• [Querying] Fixing calculation the hash of complex objects when querying that caused query hangs
• [Querying] Support TimeSpan in projections
• [Querying] Making HashCalculator use DateTime.GetDefaultRavenFormat
• [Querying] Fixing time series queries using the string interpolation of call expression in GroupBy
• [Serialization] Optimize client side JSON serialization
• Fixing backward compatibility of the health check endpoint
• Fixing issue when using lazy query + aggressive caching returned Index does not exists exception
• Better exception for actions that require database but none was specified via parameter or there is no default one
• Properly detecting changes in HasChanged when we have deletes
• Suggest X509KeyStorageFlags.MachineKeySet when passing invalid certificate to document store

Studio

• Adding navigation when clicking the backup alert
• Fixing hight CPU usage on the dashboard
• Auto-reconnect to web socket in Admin Logs & Traffic Watch
• When node is passive and license is not activated, stating so explicitly on the License View
• Giving better error message if trying to run update in query window
• Adding more details from Lucene do Idexing Performance view
• Fixing issue that setting value to zero in Database Record saved it as null
• Fixing broken failed Enforce Revision Configuration message

Server

• [Encryption] fixed possible AVE when encryption is being used

Server

• [Compression] fixed possible data corruption that could manifest when compression is used
• [Encryption] fixed memory leak that occurred when encryption was used
• [Querying] increased performance and decreased memory usage of collection queries with where clauses for operations (e.g. patching by collection or deleting by collection)

Client API

• [Querying] fixed NRE that could occur during streaming of query results

Server

• [Querying] fixed endless query issue when optimized IN is done against more than 128 items inside the IN operator
• [Threading] fix race condition in background workers that could occur under high memory conditions that could result in background task not ending
• [Storage] fixed possible NRE in decompression buffers that could occur under low disk space conditions
• [.NET Core] updated to 3.1.10

Client API

• [Querying] fixed possible infinite recursion that could occur during query hash calculation
• [Querying] fixed issue with hashing parameter of type 'Type'

Studio

• [About] disable renew license button when user does not have sufficient privileges
• [Backups] allow to export revisions even when revision configuration is disabled
• [Ongoing Tasks] fixed dirty flag behavior