OpenMLDB Weekly Update：提升性能稳定性

Summary

本周合并 Pull requests 18个，新增Pull requests 9个，关闭 Issues 10个，新增 Issues 26个。总计94个文件修改，新增1502行代码，删除7764行代码。发布Release版本v0.2.3。了解OpenMLDB

Merged Pull Requests

• docs: add the demo link in readme#305 
• docs: add a new logo#364 
• refactor: refact AppendEntries in log replicator#327 
• fix: sql and ns client desc result#341 
• docs: add compile doc#346 
• docs: new shilds#355 merged 
• docs: added new shields#348 
• fix: importer fix and doc#328 
• build(deps): bump hive-exec from 3.0.0 to 3.1.1 in /java/openmldb-import#307 
• update the blog link#299 
• refactor: remove unused scripts#343 
• fix: revert spark and hadoop version#344 
• build(deps-dev): bump junit from 4.11 to 4.13.1 in /java/openmldb-import#308 
• feat: update spark and hadoop version#332 
• feat: update openmldb version#310 
• feat: add offline update for changelog#326 
• build(hybridse): prepare version to 0.2.3#325 
• feat: build nearline tablet framework#246 

Open Pull Requests

• feat: add java common lib#347 
• feat: support column query with the same name in window skew optimization#349
• feat: enhance plan optimization for group and filter#350 
• feat: add batchjob module#359 
• feat: add task manager module#361 
• ci: openmldb java deploy workflow#366 
• feat: add openmldb jmh#367 
• test: modify test case#368 
• update readme_add ai pipeline#370 

Close Issues

• docs: Add a new OpenMLDB logo#363 
• bug: sql client `desc` result#331 
• Add compile doc#345 
• 在readme里添加知乎、oschina等媒体链接#298 
• Support reading and write tables in hive metastore catalog#202 
• Remove unused scripts#337 
• Update Spark version to 3.1.2#333 
• TestWindowUnion failed: 20 did not equal 10 (TestWindowUnion.scala:58)#122 
• Update the verison of openmldb-jdbc, openmldb-native and openmldb-import before auto release#324 
• Build NearLineTablet framework#218 

Open Issues

• feat: support and validate query table with multiple-levels path#365 
• feat: refactor of sql query router#362 
• Add TaskManager service to submit OpenMLDB Batch jobs#360 
• ut: so many unit test have memory leak, we should use sanitizer to check#358 
• bug: cicd build failed but report ok#357 
• Multiple columns with the same name can't execute when last join and over window#356 
• bug: cicd python test#354 
• bug: create index can't set ttl with no units#353 
• Add OpenMLDB docs#352 
• Add module for OpenMLDB Batch to run custom SQL and submit by TaskManager#351 
• Add java common lib#342 
• Replace `rtidb` to `openmldb` in some files#340 
• Enable the unit test of TestWindowUnion#339 
• Test and register the iceberg tables without using default Spark catalog#338 
• Enable UnsafeRowOpt by default and resolve the related issues#336 
• Enable WindowSkewOpt by default and resolve the running issues#335 
• Add the unit test for physical nodes like GroupByAgg for offline#334 
• Add deployment script#330 
• Reduce the size of docker image#329 
• Automatic packaging openmldb-import after tag creation#323 
• Add `exit` command for sql_client#322 
• Not able to find out SQL syntax docs from readme.#321 
• feat: support distributed query on BatchMode under some restrictions#318 
• feat: engine plan optimization for where and group with the same partition#317 
• feat: support integration test for java/python sdk and offline batch#316 
• enable python style check via pylint#315 

Contributors

• aceforeverd(teapot@aceforeverd.com)
• dependabot[bot](49699333+dependabot[bot]@http://users.noreply.github.com)
• dl239(dl239@126.com)
• HuangWei(huangwei@apache.org)
• imotai(codego.me@gmail.com)
• Lucifer(63491234+ashish-patwal@users.noreply.github.com)
• luyuxiao211(79981940+luyuxiao211@users.noreply.github.com)
• Rohan Sharma(rhnsharma5113@gmail.com)
• tobe(tobeg3oogle@gmail.com)
• Wang ZeKai(1290561498@qq.com)
• wuyou10206(zw1020688@163.com)
• xuman2019(52193163+xuman2019@users.noreply.github.com)

Highlights

本周发布Release v0.2.3版本，在功能特性、性能稳定性上都在稳步提升。在线存储引擎数据导入 工具 支持批量导出（Bulk load），在线计算引擎批处理模式（Batch mode）支持带参数的 SQL 查询，离线存储引擎支持数仓元数据服务（Hive metastore）以及数据湖存储格式（Apache Iceberg），数据分析能力支持Trino计算集群，离线计算引擎支持全局 排序 节点（SortByNode）。修复大量性能和功能相关的Bug，如端到端离线计算引擎测试时使用相同SQL导致的错误，在线存储引擎添加索引使用desc来展示TTL值有误等等。SQL语法也有拓展支持nvl、nvl2等函数，支持”&”、“|”、“^”、“~“等位操作函数，支持Between谓词逻辑等。详细Release日志请参考 Release OpenMLDB v0.2.3 · 4paradigm/OpenMLDB 。

OpenMLDB整体架构也在升级重构中，全新的近实时存储服务Nearline Tablet和批处理任务管理服务TaskManager也在初步设计和实现中，新增模块在OpenMLDB的易用性和可拓展性上有进一步增强，相信在不久的v0.3.0正式版中可以与开发者见面。

欢迎更多开发者关注和参与OpenMLDB开源项目。