-
Notifications
You must be signed in to change notification settings - Fork 870
OrientDB 3.2 Release Notes
tglman edited this page Aug 28, 2024
·
39 revisions
- 3.2.32 - 29 August 2024
- 3.2.32 - 18 July 2024
- 3.2.31 - 19 June 2024
- 3.2.30 - 24 May 2024
- 3.2.29 - 5 March 2024
- 3.2.28 - 21 February 2024
- 3.2.27 - 26 January 2024
- 3.2.26 - 9 January 2024
- 3.2.25 - 30 November 2023
- 3.2.24 - 26 October 2023
- 3.2.23 - 21 September 2023
- 3.2.22 - 17 August 2023
- 3.2.21 - 25 July 2023
- 3.2.20 - 15 June 2023
- 3.2.19 - 11 May 2023
- 3.2.18 - 13 April 2023
- 3.2.17 - 9 March 2023
- 3.2.16 - 8 February 2023
- 3.2.15 - 17 January 2023
- 3.2.14 - 23 December 2022
- 3.2.13 - 1 December 2022
- 3.2.12 - 10 November 2022
- 3.2.11 - 19 October 2022
- 3.2.10 - 15 September 2022
- 3.2.9 - 11 August 2022
- 3.2.8 - 26 June 2022
- 3.2.7 - 8 June 2022
- 3.2.6 - 28 April 2022
- 3.2.5 - 14 February 2022
- 3.2.4 - 28 December 2021
- 3.2.3 - 16 November 2021
- 3.2.2 - 21 September 2021
- 3.2.1 - 8 September 2021
- 3.2.0 - 29 April 2021
- Marked methods deprecated for removal in the next major
- Make sure to handle record with missing id (lightweight edges) in delete query, issue #10239
- Re-add support of indexes for between queries in new query engine
- Return every time the correct document instance in APis and queries, issue #10280
- Re-add fetch from variable for traverse in the new query engine implementation, issue #10281
- Avoid to use index when the condition is empty list
- Make sure to respect sorting on IN conditions with indexed fields
- Handle correctly lightweight edges in save
- Support of indexes in 'is null' operator
- Corrected ORestricted case for count operations in new query engine
- Make sure to bind g variable for groovy-gremlin scripts
- Update the root password if env variable is set, issue #10273
- Fix bug in SQL engine variable name resolution
- Keep order for asSet() function
- Minor optimization in query engine class/cluster filtering
- Make sure to load record from database all the times by query engine
- Make sure that toJson of list of ids get expanded, issue #10264
- Fix IndexReader ref leak when Lucene index modified and read in transaction
- Removed copied code in favor of dependency
- Reduced logging for block size detecting
- Add support of empty array as from target of query
- Make sure to run let query every record when there is a context variable
- Make sure to be able to start and load databases on startup also when not run in distributed mode
- Add skipping of failed record on import, improved import logging add content of failed record
- Corrected parsing of rid to allow white space around it in import
- Corrected imported to avoid to terminate in case of failure to read a record
- Make sure to retry to install the database later if it fail to install the first time
- Fix disk compatibility on some specific CPU architectures
- Add Support of preserving record ids on database import
- Fix issue on delegated records on raw access, issue #10171
- Minor fix in transaction not document record handling
- Make sure to check the security policy record exists before creating instance
- Correct id swap revert in distributed transaction during retry
- Make sure to execute full sync in case of delta sync failure
- Check if the index exists before checking it's properties
- Make sure to initialize maps on update of value in query engine
- Add name to the OrientDB internal timer
- Removed optimization in query engine that cause failing to update linked records, issue #10077
- Corrected query with content of nested JSON documents to use embedded class
- Add count on the create link result set
- Make sure to call on drop on database drop
- Double Write Log is fsynced on writes. Looks like usage of SYNC flag is not enough
- Double Write Log is not truncated on file flush but only if it reaches file limit
- Double Write Log is truncated every time there are tail segments in queue
- We do not need to make fsync during Write Ahead Log truncation it will be forced by Double Write Log itself
- sync flag is removed for Double Write Log channel because we perform sync on data write
- Dirty counter is updated on write sync
- Fixed database sync with enterprise agent
- Fix concurrent issues in not x86 architectures
- removed duplicate cancel from client code
- Minor correction on value conversion for comparison in queries
- Corrected implementation of tracing executor to trace source properly
- Minor fix on storage open failure
- Correct set of target class when query target a cluster
- Minor fix on retrying next host from client
- make sure to ignore empty response from distributed requests
- Refactor distributed configuration management to be handled by distributed context
- Minor fix on wait for shutdown
- Corrected order by with database locale
- Used the correct property name in alter property statement propagation
- Make sure that old query engine use public schema API for create property
- Re-enabled tests for bonsai tree with fixes
- Simplification of index implementation in storage and lucene
- Introduced executor for tracking submitter in case of exception
- Simplification and optimization in query engine results
- Corrected insert content query to respect default values
- Report DDL query errors in distributed correctly
- Make sure the distributed index drop works in the same way of embedded
- Minor change in configuration reading on context initialization
- Add support of @fieldTypes in the json from new query engine
- Add support for multiple json or parameter in insert content in the query engine
- Correct lucene search on class function
- Fix query engine integration with geospatial lucene
- Optimized OResult to load records lazily
- Optimized OResult to avoid early load of link in properties
- Minor optimizations in the query engine runtime
- Fix insert into index step to use the right value
- Minor dependencies updates
- Fix regression in rid comparison in case of embedded query
- Fix regression in direct insert into index query
- Fix regression in update index step
- Avoid storage to go in read only mode if index key type validation fails
- Optimization to avoid load record from indexes that do not match the target class
- Correct delete query in new query engine to consider also binary record
- Avoid to create temporary records in query engine equal operator
- Correct the delete edge with single from or to
- Correct optimization of count queries, issue #10021
- Moved logic for truncate cluster to the database layer
- Avoid to scan all the database files for identify a database, issue #10024
- Add check to avoid to try lock not existing ridbags
- Avoid to cast to the underlying specific type while reading a ridbag from the network, issue #9499
- Minor fix for lookup of remote session from thread local
- Make sure that lucene functions do no fail when no class is defined
- Correct distributed metadata query
- Optimizations of escaping in serialization of json strings
- Avoid to store in result metadata not network serializable values, issue #10004
- Re-enable transaction methods in scripts issue #9690
- Optimization index rebuild with transactions
- Minor optimization in free space map logic
- Fix space reuse, only use spaces in pages that have enough of it, issue #10005
- Make sure to reload correctly view index informations on context restart
- Changed logic for tracking the cluster and index views in the queries
- Optimized view refresh using batch transactions
- use everytime the most specific index when multiple are found
- Update build dependencies to most recente versions
- Refactor in query engines for execution simplification
- Corrected failure to wait for migration when storage is reused, issue #9974
- Made last refresh time of views persistent
- Minor fix in views to skip refresh of deleted views
- Reduced logging logic that is not used anymore
- Inverted order of finish check for chunks to avoid leaving data in the buffer, issue #9985
- Make sure that client executor pool size never have size 0
- Correct the class for scanning vertex on graph repair, issue #9970
- Reduced some dependencies from external libraries
- Removed not needed implementations in native logic
- Removed the need to load a native library to check some system settings
- Moved sources in main modules to avoid split packages, issue #8001
- Minor fix for new internal experimental configuration store
- Revert upgrade of minor dependency that cause issues on older OS
- Minor fixes from static analysis
- Make sure to unblock other storage open operations only when migration is complete, issue #9974
- Deduplication and simplification of distributed logic
- Removed operation waiting flow not needed any more
- Avoid startup failure in case of rebalancing failure
- Avoid that failure on load of database block the server starting
- Add command to fetch distributed configuration from SQL
- Add a specific executor, for all the io operations
- Make sure to use the context executor for all background ops
- Fix automatically remove null links on write of embedded ridbags, issue #9767
- Make sure to use the plugin class loader to load the plugin class
- Avoid repeated distributed database shutdowns
- Fix race in registering installing database to guard against concurrent installs
- Upgraded dependencies
- Minor fix on embedded storage reload
- Increased separation of schema logic for remote and embedded cases
- Corrected edge iteration in remote after commit, with relative test case issue #9805
- Upgraded to the last release of gremlin libraries
- Removed not needed dependencies, to reduce the agent bundle size
- update dependencies
- make sure to make view query wait for drop and create operations
- changed security checks for the import logic to avoid spurious failures
- optimizations in disk cache implementation
- pre-check cluster existence to avoid storage to go in read only mode in case of duplicate cluster
- add check to avoid failure when engines are unloaded, issue #9629
- removed not used configurations
- moved clean-up of old structures out of the view refresh window
- replaced global executor with context executor
- use correct index in immutable view metadata
- add configuration for define the max number of threads for the context executor
- make sure to print the right protocol for studio address when ssl is enabled
- minor fix on view metadata serialization and demineralization over network
- Removed not used single cluster management in distributed sync
- Fix spacial function filtering when used trough the new query engine
- removed the need to buffer changes before adding the to the tx
- make sure to stop scheduling of view refresh before closing the OrientDB instance
- improved refreshing and cleaning of views
- minor fix to handle misstyped records in delta serialization
- minor correction on metadata snapshot logic on release of index manager lock
- make sure to use transaction aware stream also with legacy index implementations, issue #9914
- reduced the number of times the storage goes in read only mode, and allow to delete not empty datastructures
- make sure that on index load the schema snapshot are available, issue #9915
- changes in index manager and view implementation to avoid cross locking during updates
- avoid to re-index new cluster when swapping the cluster in views.
- make sure to initialize the timeout checker before the system db
- made remote index changes go through the transaction instead of the query
- make sure to have the correct transaction in the context for distributed phases execution
- managed correctly the check of all value of a unique index during first phase checks
- give more clear error when view metadata are not correct on view creation
- improved validation of embedded types to avoid use of vertices and edges in embedded
- handle additional case for shape type conversion in query engine
- removed not needed synchronization in views structures
- reduced logging of the security module
- minor fix for no-tx changes on lucene indexes
- removed duplicate logic in index implementation
- reduced the thread local lookups and database isClosed checks
- removed some locale translate done only on specific use cases
- avoid to update the index manager when there are no changes
- replaced storage error state lock with atomic reference
- removed server and client dependency from lucene module
- minor fix on view creation and refresh in case of distributed environment
- Optimization on graph delete operations (issue #9821)
- Fix on remote deserialization of specific exception
- Fix on index logic for multi-page split
- Fix limit of index key size, now it fail with a user message when the key is over the maximum size (10Kb)
- Optimization in index write
- Optimization of security checks in writes
- Fix wrongly oversize WAL records
- Fix views locking logic to update the view atomically
- Fix failing schedule of view if previous view fail to refresh
- Add support of indexing of embedded types in views
- Fixed script poll release issue
- Make sure to use high level transactions in all index change cases
- Made index rebuild batch changes in transaction
- Fix view unlocking in case of refresh error
- Optimized update of vertices that use ridbag trees
- Fix regression on OResult hash code, issue #9900
- Ignore index corrupted errors on drop
- Batched view refreshing in transaction
- Optimized of usage of read/write lock in storage
- Fix check of index type also on chained indexes
- Reduced usage of modifiable schema to reduce contention on read/write lock
- Fix waiting for last task in ViewManager close
- Fix running of ddl while incremental backup is running do not block anymore other write operations.
- Fix invalid cluster id do not make storage readonly
- Fix in handling of node missing on refresh
- Add timeout check in next query page fetching
- Updated dependencies
- Optimized Security Policies checks
- Optimized query engine property fetching and filtering
- Removed not used anymore dependencies
- Fix in truncate clusters logic
- Improved string caching logic
- Made the direct memory pointer soft reference to retain it and reduce the allocation overhead
- Add a correct error message when opening a not existing database in a in memory context
- Fix regression in sql function execution after refactor, issue #9888
- Optimization in storage record reading
- Fix OClass.setCustom() in ditributed setup
- Fix regression database creation, issue #9889
- Limited the maximum number of metrics kept in memory
- Made query statements metrics ognostic of values
- Use correct configuration to run javascript functions on truffle script engine of graal
- Minimize duplicate logic in index searching for query engine
- Add caching of graal Context with relative database functions
- Reduced allocations in query engine
- Add possibility to unload cached information for specific script executor
- Optimized the cluster computation in query engine
- Disabled the sharding check logic for distributed
- Removed dependencies not used anymore
- Correct query interrupt based on command timeout configuration
- Minor updates of dependencies
- Make put of index everytime transactional
- Converted all left put operations to use transaction directly
- Fix logging of log messages with exceptions
- Refactor for simplify index creation logic
- Optimization in index write logic
- Optimization in ridbag write logic
- Optimization in page write logic
- Optimization in script pool
- Disabled write operations without atomic operation
- Fixes in Json parsing
- Re-enabled computation of record sizes for cluster issue #9870
- Replaced custom lock implementation with jdk implementation
- Avoid to read null values from json for ridbags
- Automatically start a transaction in the client side if a script started it server side
- Correct lucene remove after transactional index operations refactor
- Make sure rebuild of lucene works after refactor
- Fix copyStream to not terminate on 0 byte read
- Mark HA STATUS idempotent so it can be used from SQL API.
- Include individual DB status across cluster in HA STATUS result
- Add exception details to some logging that lost stack traces.
- Make sure to drop the database also when drop listeners fail
- Improve thread pool executors to scale up and down properly
- Make sure to release lock on close of failed storage
- Removed background thread to mark storage broken not needed anymore
- Add missing file extension from list of database files extensions to allow proper cleanup on delete
- Remove no needed refresh of immutable metadata during index metadata load
- Avoid that wrong validation value make a database read only
- Corrected regression on btree v3 value replace
- Refactor to simplify index v3 logic
- Prevent Live Query thread hot looping on idle queue.
- Improved distributed hosts change notification after client optimization
- Release session lock in any case of error
- Make sure to close context on process shutdown
- avoid to make the storage read only in case of error in record serialization
- correct iterable value set during update operations issue #9781
- avoid to put storage in read only in case of errors during read operations
- fixed multivalue composite index queries to not return duplicates
- correct indexes rebalancing causing error on delete operations
- local creation of a database in plocal with console old command will create a admin user
- unblock distributed request handling in case of install error, issue #9815
- Cap string length in binary channels
- Support parameterised metadata in Lucene queries
- make sure that lucene parsing exception do not set the storage in read only mode, issue #9498
- Fix the locking in the Lucene indexer
- fixed delete of lucene indexes
- make sure to do a clean metadata re-init after incremental backup database restore
- handle thread interruption while data read and write with correct exception
- Speed up of record updates.
- remove edges with in or out null during bonsai repair,issue #9806
- minor correction in composite index creation query
- cloned distributed lock keys that may contain a record id for avoid mutation
- properly handle database instance used by lucene indexes while distributed sync
- Install ShutdownLogManager in distributed mode
- LSN of latest applied record returned during restore
- Binary compatibility issue was fixed
- Update jackson dependency to the latest release
- Disabled memory pre-allocation by default
- Ported fix for shortest path, issue #9774
- Minor update of dependency version
- Storage state lock was replaced by standard JDK lock
- Minor fix for distributed on deploy database backup in case of source folder and dest folder in different volumes
- Fixes on network layer between nodes
- Restore flag is triggered once physical restore is started.
- Tracking of consistency of atomic operation on restore was added.
- If Unsafe is present it is used for memory allocation/de-allocation.
- Fix toJson() with spaces in map key names.
- Prevention of recursive flow of exception is case of internal error.
- DWL files do not trigger errors during the database backup.
- Fix ORDER BY with LET
- Minor fix in distributed stat collecting logic
- Check to avoid distributed backup path to be in the same location of databases
- Minor refactor to avoid to do version promises for not changed records
- Fix of bug when removal of entry from the BTree causing damage of the tree structure.
- add some methods in internal database API for avoid to use storage interface
- reduce use of direct storage API where is not needed
- Remove Log4J testing dependencies
- Fix npe on traversal of graph
- refactor: remove of last few usages of storage for client use cases
- separated the logic for remote ridbag management from the embeded one
- cleanup of not needed methods in remote index manager
- completely split the index remote and embedded logic
- minor refactor in remote storage to remove dependency to core components
- removed additional code in distributed module not used anymore
- add reset of previous sync backup on ddl execution
- minor fix ddls
- enabled two phase ddl with retry
- add tests for new ddl messages serialization and execution
- first definition of coordination of ddls
- first definition of new ddl messages for distributed ddls
- made distributed unit tests more reliable
- re-enabled distributed in github actions
- add method for only log distributed sequential
- removed old delta sync implementation not used anymore
- minor fix in distributed transaction data test
- minor fix in transaction data test
- minor fix in the distributed sequence manager
- Fix NPE on network reconnect
- start implementing support for Java 17
- made sure that the client network deserializer is used in most of the places
- merged Basic Transaction interface in Transation interface
- minor refactor to remove all storage usages from abstract database
- Fix of IT tests, to make them work under Java 17
- Import - Fix RID conversion in list of embedded objects
- Fix illegal memory access in Unsafe byte array comparator
- All threads given a name and default exception handler
- add method in local cache to clear referred record instances
- add property for generate and hold database instance id
- Fix DB backup UUID checks
- removed redundant cache clear from rollback operations
- minor refactor to avoid cache clear on rollback
- Reduce log level on JSON serializer
- updated icu dependency to keep only one dependency jar in the final package
- removed cleanup of locks on shutdown from HZ lock manager, try finally should already cover all cases
- minor fix to avoid a storage to go to read only if the backup fail
- update fix to avoid deadlock, to avoid the start to block
- minor fix in the loop resolution of temporary ids
- minor fix in storage validation logic
- minor fix on server startup logic
- Fix corner cases of LET expression evaluation with $current
- Add new interfacte for live query batch. Optimise live query execution in remote (Send results in batches)
- refactor of distributed drop to move distributed drop logic outside the listener and in the distributed context
- fix drop in case of distriubed environment with distributed disabled
- fixed reconnect of distributed to reduce session and delay on error
- add quorum in error message for not enough nodes online
- removed quorum all for index changes
- additional cleanups in case of full sync failure
- refactor distributed startup to avoid to deadlock on remote databases install
- minor fix on record id reset in case of distributed concurrent creation
- minor fix on sync backup reset
- Fix ETL execution with LINK transformers
- Fix management of predicate security on DELETE
- Minor performance fix in listing on databases during subscribe on push notifications
- Minor fix in server users password reading
- Automatic renew token in distributed channel when close to expire
- Import: removed unused v3 approach and started using parser for brokenrids
- Fix Studio login
- Removed distributed logging not needed anymore
- Fix Nashorn js execution
- Force Nashorn usage when GraalVM is disabled by config
- Many fixes to database import/export
- Refactor transaction index changes
- Reload schema after incremental restore
- Optimise predicate security checks on property validation
- Allow concurrent access to active queries stats
- Optimise OTxIndexChangesList and enable it for index tx management
- Allow binary connections without session token (disabled by default)
- Add security check on cached sequences
- Optimise record save
- Replaced micro tx with transaction in embedded noTx operations
- Removed micro transaction implementation not needed anymore
- Optimise views management
- Fix index space consumption
- Fix security checks on Class in remote
- Check that prevents restore of database in another version after crash was introduced
- Allow choice of storage cluster version via global config
- Fix StackOverflow error when adding the owner document to OTracked
- Fix OClassRemote.hashCode()
- Maximum allowed size of WAL segment has been increased.
- Fix issues #9658 #9562
- Fix ORDER BY with big LIMIT
- Implement DROP CLASS with parameter
- Enhance live queries to allow projections
- Implement exclude projections in SELECT
- Fix management of embedded objects on Live Query DELETE
- Fix UPDATE UPSERT with cluster as target
- Implement support for index chains in SQL with CONTAINSANY
- Add Secure header to session cookie
See What's New
Copyright (C) Orient Technologies LTD - All rights reserved.