Skip to main content

Spark Doris Connector Release Notes

25.2.0

Improvements

  • Add FE retry and load balancing strategy for Doris Flight SQL Reader #318
  • Add load balance strategy for frontends and backends #329
  • Add schema-less option #332

Bug Fixes

  • Fix filter push issue with OR in SQL #317
  • Fix the problem that retry may get stuck when FE hangs up #331
  • Improve accept_any_schema to options #326
  • Fix V2Expression build error #339
  • Fix the issue where configuring doris.benodes is not working #346

Credits

@JNSimba @wanggx @wary

25.1.0

Improvements

  • Support overwrite for DataSource V2 #281
  • Support read DECIMAL128 type #290
  • Compatible with read IP data type #289
  • Improve BE Thrift read log and stream load log #279 #313
  • Add Doris source/sink itcase and Spark 3.2/3.4/3.5 CI #296 #297 #302 #312
  • Enhance the FE nodes port check and improve write performance #303
  • Add retry interval when load fails #307
  • Add code check for FE request #309
  • Remove unused code after refactor #295

Bug Fixes

  • Fix filter IN clause compilation #277
  • Fix UT for row convertor and schema convertor #278
  • Fix NPE when empty partition commit txn #286
  • Fix prune column result isn't pushed down issue #287
  • Fix performance degradation caused by interval mistake #294
  • Fix the issue when Spark pushes down in 'case when' #300
  • Fix the issue where the loading task got stuck when stream load ended unexpectedly #305
  • Fix the problem that the query could not be performed when pushing UTF-8 encoding #311
  • Fix the conflict for rename ArrowWriter and ArrowUtil #315

Credits

@aoyuEra @gnehil @JNSimba @IsmailTosunTnyl @qg-lin @vinlee19 @wary