Graph database NebulaGraph v3.6.0 is officially released, strengthening full-text indexing capabilities

This v3.6.0 version mainly strengthens the full-text indexing capability and optimizes the MATCH performance in some scenarios.

strengthen

Strengthen and enhance the full-text indexing function, specific pr see: #5567 , #5575 , #5577 , #5580 , #5584 , #5587

Support the use of variables when using clauses to retrieve VID or attribute indexes, for specific pr see: #5468 , #5553 MATCH
Support to start RocksDB instances in parallel to speed up the startup of the Storage service, for specific pr see: #5521
Optimize the prefix search performance of the RocksDB iterator after executing the DeleteRange operation, see for specific pr: #5525
Optimize appendLog sending logic to avoid affecting write performance after follower downtime, specific pr see: #5571
Optimize the performance of the MATCH statement when querying attributes that do not exist. For specific pr, see: #5634

Fix the problem of inconsistency of Meta data, see for specific pr: #5517
Fix the problem that the leader's lease is invalid due to the RocksDB import operation. For specific pr, see: #5534
Fix the problem of stored statistical logic error, specific pr see: #5547
Fix the problem that setting the flag of invalid request parameters causes the web service to crash. For specific pr, see: #5566
Fix the problem of printing too many logs when listing sessions, for specific pr see: #5618
Fix the problem that a single large query causes the Graph service to crash. For specific pr, see: #5619
Fix the problem that the execution of the Find All Path statement causes the Graph service to crash. For specific pr, see: #5621 , #5640
Fix the problem that some expired data will not be recycled at the bottom layer. For specific pr, see: #5447、#5622
Fix the problem that adding a path variable in the MATCH statement will cause the pushdown optimization of the all() function to fail. For details, see: #5631
Fix the problem that the wrong result is returned when querying the self-loop through the shortest path in the MATCH statement. For specific pr, see: #5636
Fix the problem that the Graph service crashes due to deleting edges through the pipe character. For specific pr, see: #5645
Fix the problem that the returned result lacks edge attributes when matching multiple hops in the MATCH statement. For specific pr, see: #5646

Enhanced full-text index features, involving the following changes: the original full-text index function is changed from calling Elasticsearch's Term-level queries to Full text queries.
After the change, in addition to supporting the original wildcards, regular expressions, fuzzy matching and other methods (but the grammar has changed), support for word segmentation is also added (depending on the word segmenter used by Elasticsearch itself), and the query results include scoring results. For more syntax, please refer to the official Elasticsearch documentation .

For specific steps, refer to the upgrade document: Upgrade Version - NebulaGraph Database Manual

This v3.6.0 version is contributed by the following users, in no particular order:

Finally, special thanks to @luyade , @songqing , @shixiangz for their contributions to this release.