Release 1.1.4
In this release, Doris Team has fixed about 60 issues or performance improvement since 1.1.3. This release is a bugfix release on 1.1 and all users are encouraged to upgrade to this release.
Features
Improvements
- Do not acquire mutex in metric hook since it will affect query performance during heavy load.#10941
BugFix
The where condition does not take effect when spark load loads the file. #13804
If function return error result when there is nullable column in vectorized mode. #13779
Fix incorrect result when using anti join with other join predicates. #13743
BE crash when call function concat(ifnull). #13693
Fix planner bug when there is a function in group by clause. #13613
Table name and column name is not recognized correctly in lateral view clause. #13600
Unknown column when use MV and table alias. #13605
JSONReader release memory of both value and parse allocator. #13513
Fix allow create mv using to_bitmap() on negative value columns when enable_vectorized_alter_table is true. #13448
Microsecond in function from_date_format_str is lost. #13446
Sort exprs nullability property may not be right after subsitute using child's smap info. #13328
Fix core dump on case when have 1000 condition. #13315
Fix bug that last line of data lost for stream load. #13066
Restore table or partition with the same replication num as before the backup. #11942