v2.4.0-release-note.txt

The main revisions of V2.4.0-release focus on:
1. Distributed computing performance improvement:
● The hasAggs/having/sort/limit/Stable function/ optimization is pushed down to the DN for execution, and the performance is improved nearly a hundred times.
● Distributed process ProcLock segmentation lock optimization reduces lock conflicts and improves execution efficiency by about 5 times.
● Execution plan optimization: When the number of redistributed nodes is 1, the remote operator is skipped and the execution plan is simplified.

2. 2PC transaction optimization:
● Deadlock checking performance optimization: obtain gxids in batches; change gxids from traversal search to binary search; increase try round limit.
● 2PC adds a protection mode to prevent partial submissions.
● The prepare process is added to drop database to ensure that there will be no residual node failures.

3. Enhanced usability:
● Added pooler multi-thread log function, which is convenient for users to analyze problems.
● GTM log optimization, logs are aggregated into one file, and active logs are automatically split, which is convenient for users to view at any time.
● Support gtm_ctl -l to specify the log file path to start, which is convenient for users to manage logs.

4. New feature support:
● Self-developed partition table pruning, adding IN syntax pruning, to accelerate pruning in advance.
● Data shuffle supports user-defined functions.
● Allow dropping columns of primary tables with distributed foreign key constraints.