2.1.11 Release Notes

[yiguolei]

Behavior Changes

• time_series_max_tablet_version_num controls the maximum number of versions for tables using the time - series compaction strategy. [enhancement](time_series) increase the version limit for the time series table #51371

• Fixed the issue where the HDFS root_path did not take effect during hot - cold tiering. [fix](cold hot separation) Fix the issue of root_path not working in HDFS resource #48441

• In the new optimizer (Nereids), when the depth or width of an expression in a query exceeds the threshold limit, the query will not fall back to the old optimizer regardless of whether the fallback has started. branch-2.1: [fix](nereids) exceed expr limit error not fallback to legacy planner #52431

• Unified the name checking rules for enabling or disabling unicode names. Now, the non - unicode name rules are a strict subset of the unicode name rules. branch-2.1: [opt](identifer) let unicode format as a superset of latin format #48078 #53264

New Features

Query Execution Engine

• Introduced the system table routine_load_job to view information about routine load jobs. [improve](information schema) introduce routine load job system table #48963

Query Optimizer

• Supported MySQL's GROUP BY roll - up syntax GROUP BY ... WITH ROLLUP. branch-2.1: [improvement](nereids)Support GROUP BY ... WITH ROLLUP syntax #51948 #51978

Improvements

Query Optimizer

• Optimized the performance of collecting statistical information on aggregate model tables and primary key model MOR tables. branch-2.1: [improvement](statistics)Agg table set preagg on when doing sample analyzing. (#49918) #51675

Asynchronous Materialized View

• Optimized the planning performance of transparent rewriting. [opt](mtmv) optimize mtmv rewrite performance (#49514) #51309

• Optimized the refresh performance. https://github.com/apache/doris/pull/51493

Bug Fixes

Import

• Fixed the problem that the display result of show did not meet expectations after altering the attributes of routineload. [fix](job) fix show routine load job result incorrect after alter job property #53038

Lakehouse Integration

• Fixed the issue of incorrect data reading for Iceberg equality delete in certain cases. [fix](iceberg) fix the iceberg eq-delete filter resize-fill bug. #51253

• Fixed the error of Iceberg Hadoop Catalog in the Kerberos environment. [fix](iceberg) fix iceberg hadoop type kerbros issue. #50623 [fix](iceberg)Ensure proper authentication context before accessing Iceberg Catalog  #52149

• Fixed the problem of failed transaction submission when writing to Iceberg tables in the Kerberos environment. [fix](iceberg)Fix the thread pool issue used for commit. #51508

• Fixed the error in transaction submission when writing to Iceberg tables. [fix](iceberg) Fix transaction issues #52716

• Fixed the error when accessing Hudi tables in the Kerberos environment under certain circumstances. [fix](hudi catalog) Fix the Kerberos authentication error when querying hudi table #51713

• SQL Server Catalog supports identifying IDENTITY column information. [improvement](jdbc catalog) Optimize the acquisition of indentity type in SQLServer #51285

• Fixed the issue that Jdbc Catalog tables could not obtain row count information in some cases. [fix](jdbc catalog) fix a jdbc catalog npe #50901

• Optimized the decompression performance of ORC zlib in the x86 environment and fixed potential problems. [fix & opt](orc) ORC-1525: Fix bad read in RleDecoderV2::readByte and Decompress zlib by libdeflate. #51775

• Added indicators related to Parquet/ORC condition filtering and delayed materialization in the Profile. [Enhancement](multi-catalog) Add PredicateFilterTime, DictFilterRewriteTime, LazyReadFilteredRows profile metrics in parquet orc profiles. #51248

• Optimized the reading performance of ORC Footer. [Optimize](orc-reader) Optimize stripe footer multiple reads in orc reader. #51117

• Fixed the problem that Table Valued Function could not read compressed JSON files. [fix](tvf) support compressed json file for tvf and refactor code #51983

• Fixed the issue of inconsistent metadata caused by concurrent catalog refreshing in some cases. [fix](catalog) synchronize reset methods in catalog classes and remove Lombok annotations #51787

Index

• Fixed the query error of the inverted index when processing IN predicates containing CAST operations to avoid returning incorrect query results. https://github.com/apache/doris/pull/50860

• Fixed the memory leak problem of the inverted index in abnormal execution situations. https://github.com/apache/doris/pull/52747

Semi - structured Data Type

• Fixed the problem that some JSON functions returned incorrect results when dealing with null values.

• Fixed some bugs related to JSON functions. [fix](json) Add . after in JSON path to support correct token parsing #52543 [fix](function) JSON_EXTRACT_STRING should return NULL instead of the string 'null' when encountering a NULL value #51516

Query Optimizer

• Fixed the issue that the query could not continue execution when parsing a string into a date failed. branch-2.1: [fix](nereids) fix parse date time exception #50810 #50900

• Fixed the problem of incorrect constant folding results in individual scenarios. branch-2.1: [fix](nereids) fix cast ipv4 to string #51546 #51738

• Fixed the issue that individual array functions could not be planned normally when encountering null literals as input. branch-2.1: [fix](Nereids) could not work well when check precision for null literal #50815 #50899

• Fixed the problem that enabling local shuffle might lead to incorrect results in extreme scenarios. [fix](nereids) fix colocate agg + join compute wrong result (#48934) #51313 branch-2.1: [fix](coordinator) fix cte with local shuffle throw exception #52870 #52871

• Fixed the issue that replace view might cause column information not to be visible when using desc view. branch-2.1: [fix](metadata) replace view will break view's metadata #51058 #52043

• Fixed the problem that the prepare command might not be executed correctly on non - master FE nodes. branch-2.1: [fix](Prepared Statment) Fix exec prepared insert stmt in non master error (#48689) #52265

Asynchronous Materialized View

• Fixed the problem that query failure might occur after transparent rewriting when the data type of the base table column changes. branch-2.1: [fix](mtmv)before mtmv refresh,check column type if change #50730

• Fixed the problem of incorrect partition compensation in transparent rewriting in individual scenarios. [fix](mtmv) Fix compensate union wrongly when direct query is empty relation (#51700) #51899 [fix](mtmv) Fix compensate union all wrongly when partition mv use partition external table #52218

Query Execution Engine

• Fixed the problem that TopN calculation might core dump when encountering variant column types. [Bug](topn) variant column read in topn may coredump #52573

• Fixed the problem that the function bitmap_from_base64 would core dump when inputting incorrect data. [Bug](function) fix bitmap_from_base64 function cause heap-buffer-overflow error #53018

• Fixed the problem of some incorrect results of the bitmap_union function when dealing with ultra - large amounts of data. [Bug](aggregate) fix bitmap_union return error result in query sql #52033

• Fixed the calculation error of multi_distinct_group_concat when used in window functions. [Bug](distinct) fix distinct function with over return error result #51875

• Fixed the problem that the array_map function might core dump at extreme values. [bug](function) fix array_map cause coredump as NULL #51618 [bug](function) fix array_map coredump when no rows #50913

• Fixed the problem of incorrect time zone handling. [Fix](timezone) Fix incorrect DST handling #51454

Others

• Fixed the inconsistent behavior of multi - statements between the master FE and non - master FE. branch-2.1: [fix](fe) multi statements different behavior in master and follower (#52144) #52632

• Fixed the error of prepared statements on non - master FE. [fix](Prepared Statment) Fix exec prepared insert stmt in non master error #48689

• Fixed the problem that the rollup operation might cause CCR interruption. [fix](cooldown) allow cooldown_ttl = 0 when altering storage policy #50830

[wuzg2000]

arm64的文件4.1G，比x86_64的2.7G大很多，解压后目录结构和3.0.x一样，是不是文件挂错了？

[Ruffianjiang]

arm版的包不太对，大小和3.0的一样了。
正常不到3G才对