What's new
3.0.30115 - 2016/04/10
Server
- [Major]
[Prefetching]
Fixed memory leak that could lead into slowness of the server under heavy load impacting indexing and replication mechanisms, - [Major]
[Indexing]
Fixed issue when new indexes could stuck at 32768 entries under certain conditions when precomputed batch flow was executed, [JavaScript]
Fixed float conversion- Added memory statistics thread state to stats (
IsMemoryStatisticThreadRuning
)
Client
AlwaysWaitForNonStaleResultsAsOfLastWrite
convention marked as obsolete- Fixed concurrency errors when retrieving identity values concurrently by
DatabaseCommands.NextIdentityFor
- Fixed JSON parsing of data using base64 format
Studio
- Fixed import with credentials usage
- Fixed replicate all indexes and transformers options
- Fixed Etag verification when changing an identifier of an existing document
3.0.30100 - 2016/03/17
Server
[Indexing]
Fixed handling of indexes which were disabled,[Indexing]
Added support for nullable types to facets,[Indexing]
Fixed updates of side-by-side indexes,[Indexing]
Fixed index staleness detection,[Indexing]
Fixed an issue where we wouldn't treat JsonPropertyAttribute correctly during the indexing,[Indexing]
Added support for custom type arrays in an index definition,[Prefetching]
Adjustments to document prefetching mechanism,[Esent]
Fixed Esent storage schema update (4.8 to 5.0),[Voron]
Fixed handling of overflow pages[JavaScript]
Fixed math random init in Jint,[Replication]
Fixed "Resolve to latest" conflict resolution strategy,[Configuration]
ExposedRaven/Voron/SkipConsistencyChecks
setting,[Configuration]
AddedRaven/MaxPrecomputedBatchSizeForNewIndex
setting,- Fixed document caching mechanism,
- Fixed validation of OEM/ISV licensed affecting the startup performance,
- Fixed UDP port leak in licensing component,
- Fixed debug info generation,
- Fixed UniqueConstraintsPutTrigger, which shallowed a snapshot of a document,
- Added
debug/sl0w-lists-breakd0wn
debug endpoint exposing internal lists breakdown, - Fixed timing of dynamic queries,
Client
- Fixed querying with transformer usage - respecting query type instead of transformer one,
- Fixed
session.Advanced.Lazily.Load()
andsession.Include().Load()
methods against passing duplicate identifiers, - Fixed includes in a sharded session,
- Fixed loads in a sharded session (when array of identifiers specified),
- Fixed insertion of multiple indexes with given priorities,
- Fixed internals of UniqueConstraint bundle,
- Fixed
AllowMultipleIndexEntriesForSameDocumentToResultTransformer
flag application, - Added extension methods to work with authorization bundle (
Raven.Client.Authorization
namespace), - Fixed filtering of ignored headers,
- Optimized memory usage when streaming documents with missing properties according to a given type,
- Optimized usage of HttpClient cache,
- Fixed reconnection issue in Changes API,
- Added
PutIndexAsyncWithOperation
command method
Studio
- Fixed traffic watch for file systems,
- Fixed documents visibility on the patching page,
- Fixed vertical scrolling bar on the collections and document view,
- Fixed import to CSV to preserve column ordering, skipping and document identifier if present,
- Fixed info view of currently running tasks,
- Fixed escaping of queries with contains
:
character
Smuggler
- Fixed issue that an export file didn't include documents created during the export operation,
- Fixed export of file systems,
- Added
--batch-size
option for smuggling file systems, - Fixed import operation for large attachments,
- Fixed smuggler "between" option to ensure documents added during the operation are transferred as well
3.0.30037 - 2016/01/09
Server
[DTC]
Fixed NonAuthorativeInformation detection,- Added
admin/generate-oauth-certificate
endpoint, - Exports will now include documents created during the export time,
- Fixed licensing timer leaks,
- Minor fixes and tweaks,
- Added a mechanism preventing from opening a resource with a different storage type than is defined in configuration,
- Added a mechanism preventing from creating a different resource type than is defined in data directory,
[Scripted Index Results]
will work correctly if there are multiple loads and puts of the same document in same batch,[JavaScript]
decreased max recursion from 1024 to 128,[Replication]
addedLastReplicatedAttachmentEtag
to the replication statistics,[Patching]
Inc operation will now assume that the value is Int64, not Int32,- Better mechanism preventing index corruption after server crash
Client API
- support for DNX Core 5.0 (only unstable builds greater than 30011 due to NuGet policies),
- Streams now contain full document metadata,
- Added
SetResultTransformer
overload toIDocumentQuery
andIAsyncDocumentQuery
, - Query
DurationMilliseconds
is set to -1 when response comes from cache
Studio
- Fixed FileSystem import,
- Studio should not allow to 'save' locked index or should indicate after pressing save that the index is locked and changes will not be saved,
- Minor fixes and tweaks
FileSystem
- Failure to start FileSystem should result in 503, not 500 status code,
- Fixed issues with recreating IndexSearcher during index reset, which could result in a failure to access index files,
- Fixed FileSystem exports,
- Fixed issue with streaming, now streams will exclude internal files (tombstones, deleted, synchronized),
- Stability improvements
StorageExporter
- Added
SkipConsistencyCheck
option
Smuggler
- Better export timeout handling,
- Enhanced server version discovery
3.0.30000 - 2015/11/19
- [Breaking Change] Changed the build number in the RavenDB version (3.0.30000)
Server
[Prefetching]
Fixed performance issues,[Indexing]
Fixed an issue with accessing index files (System.ObjectDisposedException: Cannot access '_[xxx].fdt' because the index input has been disposed or System.IO.FileNotFoundException: Can not load ICSharpCode.SharpZipLib.dll),[Indexing]
Fixed reduction phase which skipped docs when first time it was performed in a single step but later it was processed as a multi step operation,[Indexing]
Fixed race condition in Lucene.net spatial contrib,[DTC]
Fixed an issue with deletes performed in multiple concurrent threads (while in a distributed transaction) were not being replicated,[Scripted Index Results]
Fixed missing execution of index update triggers when removing from map-reduce index using RemoveFromIndexTask,[Scripted Index Results]
Fixed conversion of null string properties,[SQL Replication]
Fixed bug when comparing withnull
in a sql replication script,[Configuration]
AddedRaven/Tenants/MaxConcurrentResourceLoads
andRaven/Tenants/ConcurrentResourceLoadTimeout
settings which limit concurrent load of resources (databases, filesystems, etc)
Client API
- Added support for transactional sessions with DTC under async sessions,
- Fixed an issue with unbounded results API which returned up to 128 documents when doing spatial query using Customize(),
- Enabled compression for more requests to shrink the amount of data transferred through the network,
- Added
InMemoryDocumentSessionOperations.UnregisterMissing
and invoke it before loadingConstraintDocument
in bundle, - Added new overload of
SetResultTransformer
toIDocumentQuery
that allows strong-typing of the transformed result independently of the type of the index entries.
Data Subscriptions
- Changed the approach of retrieving and processing documents to avoid connection breaks caused by consuming incoming data too slowly,
- Fixed issues related to opening a subscription depending on a specified strategy,
- Fixed a timeout handling in subscriptions if all documents are filtered out
File systems
- Fixed OutOfMemoryException when uploading large files or synchronizing between servers,
- Fixed an issue that a file were accessible even though its upload has been aborted,
- Fixed file uploads with Windows Auth enabled
Smuggler
- Fixed the import of dump files containing
Raven/Subscription/...
identities used by Data Subscrptions, - Added support for transform / filter scripts on database export.
Studio
- Added transform script validation for import/export operations. Fixing help message and outdated links,
- Fixed Patch by index - query not filtering matching documents,
- Fixed acquisition of the debug info package if the server machine has .NET installed in version 4.5.2 or 4.6,
- Faster transitions,
- Fixed race condition between loading the studio version and generating help link,
- Fixed Patch page - when selecting a collection the documents are overlapping the Before Patch and After Patch,
- Fixed Indexes page - show if index is map reduce
version.3800 - 2015/09/21
Server
- Improved formatted index generation with better error handling,
- Fixed issue with single OAuth tokens caching,
- Larger batches are now handled better by Map-Reduce indexes,
- Added support for HEAD request for streams,
- Fixed
ArgumentOutOfRangeException
that could occur during reading from Lucene index
File systems
- Better handling of larger files with longer names
Client API
- Fixed issue with saving documents to proper database in ShardedBulkInsert operation,
- Fixed issues with index generation,
- IndexCreation now takes into account conventions,
- Added the option to specify timeout of a subscriptions pull request
Studio
- Fixed replication topology graph
- Fixed replication settings page