Neo4j 2.3 Community changelog

2.3.12

Kernel

• Close statement in node relationships iterable's #9752

Cypher

• TransactionBoundQueryContext no longer creates new statements #9790

Docker

• Fixed setting config values via environment values #91

2.3.11

Kernel

• Don't cache unchanged records in DirectRecordAccess #9386

Cypher

• Fixed an unnecessary error when loading an empty file using LOAD CSV WITH HEADERS #9183
• Fixed error with accessing maps and lists with null, so it now evaluates to null instead #9272

Docker

• Configuration can be set from environment variables #83
• Fixed issue which prevented dbms.mode=ARBITER from working #89

2.3.10

Kernel

• Fix an issue in batch importer where id duplication detection would miss some 1 in 10.000 ids #8731

Cypher

• Fixed bug with relationship direction in create unique with path variable #8678

Misc

• Improved hostname resolution for Graphite connections #8787

2.3.9

Kernel

• Fix a bug that could prevent recovery from finding the latest check point record in the logs, preventing adequate recovery of the store. #8418
• Fix a bug that on rare occasions could lead to leaking paged memory. #8499

Cypher

• Better error messages for property assignment to invalid collections #8513
• Remove cypher dependency on kernel IndexDescriptor #8615

Security

• Generated SSL certificate files should only be readable by owner. #8510

2.3.8

Kernel

• Fixes #7204 an issue where log rotation sometimes would cut a transaction in half making it problematic for recovery to read such transactions. Transactions are now never split into multiple log files. Any db experiencing this issue can now also start that db with this fix included due to improved log reading abilities. #7895
• Lock node without relationships during detach delete #8001
• Retry count store lookup in case of concurrent state change. #8003
• Setting dbms.tracer=null will no longer prevent the database from starting when metrics are enabled #8402

Cypher

• Tweaks to query re-planning #7881
• Fix a bug where aliasing of aggregations caused queries to fail in semantic checking. #8278
• RETURN * with ORDER BY should not leak internal variable names in the final result due to AST rewriting. #8389

Browser

• Performance improvement - Limit default number of meta items shown in database sidebar drawer #8299

Docker

• Listen on localhost only when setting credentials #53

2.3.7

Kernel

• Open lucene indexes in read only mode when database started in read only mode (not applicable to ha setup) to prevent cases when usage of interrupts will result in deleted index. #7530
• Fixed exception thrown when iterating over reverseNodes() if Path was a single node long #7609 #7619
• Fixes an issue where failure during forcing of transaction log after multiple concurrent threads appending to it wouldn't have all threads notice the kernel panic before marking their transactions as committed. In rare cases the effect may be that there could be a skipped transaction id in the otherwise sequential gap-free transaction log. #7621
• Don't take locks for entities created in the same transaction. #7628
• Automatically rebuild indexes that fail to open on startup. #7745

Misc

• Made possible to allow node id reuse using 'dbms.ids.reuse.types.override' setting. #7600

2.3.6

Kernel

• Transaction termination aware locks #7343
• Safe freeing of ids for HA slaves #7477

Misc

• Fixes a problem where deleted property record ids may be reused twice after a role switch in HA, causing properties to be overwritten #7342
• Partitioned instances revert to PENDING state while allowing read only operations #7423

2.3.5

Kernel

• Safe freeing of aggressively reused ids #7072

• Prevents null from being added to legacy indexes #7097

• Reading properties scales better by no longer needing to acquire read locks. #7179

  This also removes experimental.use_read_locks_on_property_reads setting.

• Fixes issue with map size in IndexDefineCommand #7213

• Reintroduced read locks on property reads until discovered issues have been resolved #7348

• Fixes issue with relationship group type overflow #7363

Cypher

• Misc improvements for LOAD CSV #7149
• Fix bug in the IN operator #7190
• Handle deleted entities in property operations #7217

Packaging

• Fix: NEO4J_ULIMIT_NOFILE is now respected on Debian package service restart #7269
• Allow Debian package users to configure shutdown parameters via defaults #7285

Tools

• Errors related to input data will not print stack trace unless supplying --stacktrace #7263

Misc

• Improved logging in online backup #7287

2.3.4

Kernel

• Safe release of outdated entries in TimedRepository #6782
• Change counts store to write -1 as trailer when empty. #6849
• Fix deadlock caused by different lock ordering on nodes for relationships #6930
• Separate checkpoint triggers for backup and store copy #6981
• Fix NPE after Recovery #7049
• A little faster decoding of short string/array #7054
• Handling of unused records in property chain #7060

Cypher

• Modulo function cannot only use doubles #6756
• Fix NPEs in NiceHasher and NiceHasherValue #6814
• Fix computing effects in DeleteEntityAction #6896
• Fix the logic around isolating aggregations #6917
• Fix bug in rule planner pattern matcher #6922
• Deleting a node/relationship should never fail #6928
• Optimize Cypher aggregation memory usage #6948
• Fix sampling of lucene index size #6963
• Fix IDP solver configs #7047
• Add missing initialization of variables on each iteration #7061
• Remove compiled runtime from 2.3 #7107

Packaging

• Added ability to set max number of open files in /etc/default/neo4j (default behavior remains unchanged) #7095
• Neo4j can now be started on OS X if only Java 8 is installed #6895 #7113

High Availability

• Introduce new parameter max_survivable_failed_members. #7011

Misc

• Added print of Neo4j version #7057

2.3.3

Kernel

• Report error and starting line in import if quote is unbalanced #6217
• Fix raw ASCII support in import tool #6221
• Fix AStar result in case End equals Start #6223
• Adaptive work-ahead queue size in ParallelBatchImporter #6225
• Fix Integer overflow for large number of dense nodes #6253
• Ability to specify db config in import tool #6284
• Add a timer to arbiter's shutdown hook in case of deadlock #6293
• Handle leading/trailing whitespace in importer #6321
• Fixes CC issue where some relationship inconsistencies may be overlooked #6517
• Specify sender address for network communication #6562
• Bigger page size during import #6611
• Fixes issue with incremental backup failing with invalid high id #6627
• Fix record counting in legacy CC #6697
• Adversarial and Devious PageCacheRule #6712
• Fix for rare permanent self-deadlocks in the Enterprise lock manager #6654 #6754

Cypher

• Fix delete operation in cypher 2.0 compatibility mode #6206
• Fix missing predicates from the solved query graph when using greedy #6210
• Fix bug around disconnected patterns after WITH #6216
• Improve error message for invalid unicode literal #6236
• Fix issue with deleting paths #6237
• 2.3 idp long chains #6269
• Relationship by id should work with arguments #6333
• Make DISTINCT work with arrays inside nested structures #6340
• Fix queries having multiple pattern sharing same rel names but different node names #6368
• Do not create label ids when removing non-existant labels via Cypher #6369
• Don't overflow on aggregation #6430
• Improve unwind to not materialize collections and avoid OOM #6463
• Fixed eagerness on reads after deletes #6564
• Cypher should be able to run queries when query cache size is 0 #6569
• Fixed eagerness with deletes followed by reads #6571
• Fix plan caching #6602
• Make ExpandIntoPipe deterministic #6724
• Using Kahan's algorithm to compute AvgFunction and SumFunction #6745
• Fix label cardinality estimation on the rhs of apply when projecting #6751

Misc

• Fixed a bug in Windows with HA mode when connecting to self #6667
• Fixes issue where some properties could get dropped during large imports #6692

2.3.2

Kernel

• Log the first heartbeat after a series of missed #5629
• Introduce ascending-order fixed tx push strategy. #5712
• Fix cluster uri resolution #5869
• Add missing checks for tx changes before augmenting iterators #5888
• Deprecate metrics reporting into a single csv file #5892
• Fix Cluster Uri regression again #5932
• Fixed issue with last committed tx id in query planning #5935
• Add an echo " done" and newline to stdout on stop #5949
• Add check "hasTxStateWithChanges" when creating an unique constraint #5953
• ALL and NONE predicates evaluated during shortest path traversal #5974
• Deprecate untested code merged by mistake #5993
• Enable the use of enterprise edition from neo4j-shell -path ... #5995
• Fixed command aggregation for legacy indexes #6022
• More robust placeholder parser in batch endpoint, fixes #5664 #6032
• Accepts raw ascii code as delimiter in import tool #6034
• Release lock client in KTI #6075
• Make SlaveLockClient cope better with ComExceptions #6102
• Add the option to trace ComExceptions #6117
• Fix reading of props with different block sizes #6141
• 2.2 update pull and coms #6146
• Ability to import > 2.4 billion nodes in import tool #6180

Cypher

• Equality should treat chars as strings #5909
• Correct status code when using hint with no index #5925
• Fix plans generated with USING JOIN hints #5927
• Fix bug with cost planner NodeIndexSeekByRange #6063

Misc

• Don't show a file limit warning on OS X when running neo4j console #6080

2.3.1

Cypher

• Abort rewriting of predicates with too large DNF #5821