What's new

Server

• [Voron] Fixed free space tree handling that returned already allocated pages as free what could result in overwriting actively used pages
• [SQL Replication] Removed recent touches optimization to ensure we don't skip documents in SQL replication
• [Indexing] Fixed map reduce indexing which could got stuck in when the batch size is 256 and the values for a reduce key is more than 1024

Client

• [HiLo] Fixed HiLo generator which could get stuck with the aggressive cache enabled

This is an aggregated build containing fixes from builds 3.0.30156 and up

Server

• [Voron] stability improvements,
• [Voron] journal flushing stability improvements,
• [Voron] Fixed possible release of scratch files being currently in use,
• [Indexing] fixed possible infinite loop that could occur in indexing when replication of delete occured,
• [Indexing] Removed support for using 'minimum ETag replacement strategy' when doing Side-By-Side on Map-Reduce indexes,
• [Licensing] ability to use 3.5 standard license in 3.0,
• [Embedded] Prevented from conflicting RavenDB.Database and Metrics.Net assemblies loaded in the same project,
• [Configuration] Added Raven/SqlReplication/CommandTimeoutInSec setting,
• [SQL Replication] Added CommandTimeout property to SqlReplicationConfig that allows to override command timeout per destination,
• [SQL Replication] Fixed issue with setting LastReplicatedEtag when SQL batch times-out,
• [SQL Replication] Fixed calculation of deletes duration,
• [Querying] Lucene will now escape text phrases AND and OR properly when they are not the operators,
• [Querying] fixed query parsing when using @in operator,
• [Prefetching] fixed prefetching behavior that could cause inifinite indexing loop,
• [Prefetching] Fixed creation of too many future batches,
• [Versioning] Fixed revisions download of a deleted file,
• BulkInsert now supports auto-generated ids for '/' id pattern e.g. 'orders/',
• BulkInsert now works properly with Ids that contain whitespace characters at the end,
• Fixed disposal of the documents cache when disposing of Voron storage
• Fixed an issue where replication will be triggered unnecessarily when a lot of indexing is going on
• Added support for storing Database and FileSystem transactional logs in the same directory

Client

• Fixed session.Advanced.LoadStartingWith with a transformer having LoadDocument which returned inconsistent results,
• [Side-By-Side] Client will throw when 'minimum ETag replacement strategy' is used on Map-Reduce indexes,
• TransformerParameters will have it's data escaped properly now,
• Better detection if server is down for various edge cases so FailoverBehavior can work properly,
• Fixed conflicts resolving by Query and LoadStartingWith method,
• Fixed sharding operations when custom database parameter was passed

FileSystem

• [Client] Fixed possible NullReferenceException that can occur when file that does not exist is being loaded,
• Fixed problem with declining reading performance in file systems with large number of files

Smuggler

• support for 4.0 format

Studio

• Fixed an option to overwrite a name of an exported database file,
• Better memory utilization when viewing Running Tasks view,
• Fixed displaying of more that 1024 databases

Tools

• [Backup/Restore] Taking into account custom index storage path
• [Storage Exporter] Allowing to specify custom path of journals / logs

Who should upgrade

Users who see high CPU usage with replication

Server

• Fixed an issue where replication will be triggered unnecessarily when a lot of indexing is going on

Server

• [Prefetching] Fixed creation of too many future batches
• [SQL Replication] Fixed calculation of deletes duration
• [Voron] Fixed possible release of scratch files being currently in use
• [Versioning] Fixed revisions download of a deleted file
• Fixed disposal of the documents cache when disposing of Voron storage

Client

• Fixed conflicts resolving by Query and LoadStartingWith method
• Fixed sharding operations when custom database parameter was passed

Studio

• Fixed displaying of more that 1024 databases

Tools

• [Backup/Restore] Taking into account custom index storage path
• [Storage Exporter] Allowing to specify custom path of journals / logs

Server

• BulkInsert now supports auto-generated ids for '/' id pattern e.g. 'orders/',
• BulkInsert now works properly with Ids that contain whitespace characters at the end,
• [Querying] fixed query parsing when using @in operator,
• [Prefetching] fixed prefetching behavior that could cause inifinite indexing loop,
• [Voron] journal flushing stability improvements,
• Added support for storing Database and FileSystem transactional logs in the same directory

Client

• TransformerParameters will have it's data escaped properly now,
• Better detection if server is down for various edge cases so FailoverBehavior can work properly

Smuggler

• added support for 4.0 export files