Server

• [Replication] fixed issue where under low memory we could stop sending any documents, now we will send at least one
• [.NET Core] updated to 3.1.25

Server

• [Backups] fixed issue with backups timers not being rescheduled when database fails to load
• [Voron] fixed potential deadlock when getting flushing and write transaction locks
• [Voron] fixed potential race condition between backup and journal applicator
• [.NET Core] updated to 3.1.24

Setup

• Signed setup-as-service.ps1

Server

• [Backup] fixed issue with concurrent access to headers files when doing snapshot backup
• [Backup] fixed issue with trying to backup already deleted journals when doing snapshot backup
• [Cluster] fixed unhandled cases for raft snapshot
• [Debug] fixed potential invalid memory access during creation of Debug Package
• [Indexing] fixed wrong decompression usage in map-reduce indexes that could result in low level Voron errors
• [Indexing] fixed issue with compacting an index with NGram analyzer
• [Indexing] fixed issue with putting ReplacementOf indexes in the database record
• [Indexing] fixed index error when trying to parse a short type
• [Indexing] fixed problem with undisposed indexes and UnauthorizedAccessException if an error was thrown during a database initialization
• [Licensing] fixed the downgrade to dev license from commercial one
• [SNMP] fixed the count of faulty databases when doing compaction
• [SNMP] fixes issue with time since oldest backup showing wrong values for never backed up dbs
• [SNMP] added number of faulty indexes (per database and total)
• [.NET Core] updated to .NET Core 3.1.23

Client

• [Bulk insert] forced data flush on initial write of data to the server to fix potential connection problem when using SSL

Server

• [Cluster] fixed issue with observer moving node from member to rehab back and forth
• [Certificates] when database is being deleted, access will be removed from the certificate as well, after recreation of the database permissions needs to be set again
• [Counters] properly removing counters and counters-snapshot from metadata upon counter deletion
• [Debug] added database record to the debug package
• [Encryption] fixed issue with encrypted stream flushes not being done in 4096 increments when using gzip
• [Indexes] properly handle SelectMany with DateTime
• [JSON] support for 'yyyy-MM-ddTHH:mm:ss.fff' format
• [Querying] fixed issue with using in projections fields from documents loaded using 'load'
• [Replication] fixed possible freed memory usage

Client API

• [Bulk Insert] fixed issue with sending very large documents
• [Subscriptions] do not fail subscription worker if MaxErroneousPeriod was spent during registration of subscription connection

Studio

• [Backup] region should not be mandatory when using custom S3 host

Server

• [Cluster] idle indexes should not be taken into account when calculating promotable state
• [Compare Exchange] fixed possible NRE when 'AddOrUpdateCompareExchangeCommand' is executed due to possibility of early context release
• [Configuration] added 'Indexing.TimeSinceLastQueryAfterWhichDeepCleanupCanBeExecutedInMin' configuration option
• [Configuration] added 'Indexing.IndexStartupBehavior' configuration option
• [Backup] fixed issue with testing AWS credentials when custom URL is used
• [Backup] fixed issue with rescheduling backup when during processing it is being moved to another node
• [Backup] added 'ForcePathStyle' option to S3 settings
• [ETL] fixed possibility of infinite loop in ETL (backported from 5.2)
• [Indexing] early stop reference processing if server is under memory pressure
• [Indexing] fixed issue with retrying 'OutputReduceToCollectionCommand' in transaction merger which could result in not deleting relevant documents
• [Querying] fixed issue with calculating total number of results when using collection query on a non-existent property
• [Replication] fixed issue with calculating properly HasRevisions flag when replicating document delete
• [Replication] revision configuration should create revision bin for documents deleted by replication
• [Transactions] fixed issue with re-running some of the commands when merging fails and they need to be executed once again one by one
• [Voron] using 'mmap' instead of 'posix_memalign' when allocating 4KB aligned memory for encryption buffers to avoid memory fragmentation
• [.NET Core] updated to .NET Core 3.1.22

Other

• [Docker] added the ability to override server URL via RAVEN_ServerUrl environment variable